Latest / The Pragmatic Engineer / Amazon, Google and Vibe Coding with Steve Yegge

Transcript

Amazon, Google and Vibe Coding with Steve Yegge · The Pragmatic Engineer

0:00
stevie's platform ran because it was a really good criticism of
0:03
google it was a
0:04
really realistic picturing of amazon including jeff bezos not
0:08
giving a shit about
0:09
your day he still doesn't did you write this kind of things all
0:12
the time i was fed
0:13
up i've been there six years and i still couldn't get a
0:15
platform out of anybody i
0:16
went nuts and then a bottle of wine later i told him how it was
0:19
but you were
0:20
actually right in hindsight i was right about all of it but
0:24
they never said sorry
0:24
SPEAKER_03: stevie's
0:25
Steve Yaghi is widely known for his writing and rants in
0:27
software engineering.
0:29
SPEAKER_03: His blog posts,
0:29
SPEAKER_03: Get That Job at Google,
0:30
SPEAKER_03: was circulated by Google HR for hiring purposes for 15 plus years.
0:34
And his Google Platforms rant,
0:35
written a decade ago,
0:36
is still heavily cited across the industry.
0:39
SPEAKER_03: Steve worked for seven years at Amazon,
0:40
SPEAKER_03: 13 at Google,
0:42
SPEAKER_03: and is now building AI tools at Sourcegraph.
0:44
SPEAKER_03: In this direct conversation with Steve,
0:46
we cover the infamous Google platform rant and why Steve thinks
0:50
Google is still
0:51
SPEAKER_03: terrible at building platforms.
0:53
SPEAKER_03: Why Steve unretired from tech and coding thanks to AI tools.
0:57
Why Steve thinks more dev should vibe go together with AI and
1:00
many more interesting topics.
1:02
If you're interested in how AI tools will change how tech
1:04
companies operate,
1:05
SPEAKER_03: how us developers can keep up with them,
1:07
or why the core DNA of tech giants like Google and Amazon seem
1:10
to change very
1:11
SPEAKER_03: little over 20 years,
1:12
SPEAKER_03: then this episode is for you.
1:14
If you enjoy the podcast,
1:15
please subscribe to it on any podcast platform and on YouTube.
1:19
SPEAKER_03: So Steve, just welcome to the podcast.
1:20
SPEAKER_03: It's so nice to also meet you in person.
1:23
SPEAKER_03: KAREK
1:24
WILKIEWICZKIEWICZKIEWICZKIEWICZKIEWICZKIEWICZKIEWICZKIEWICZKIEWICZKIEWICZKIEWICZKIEWICZKIEWICZKIEWICZKIEWICZKIEWICZKIEWICZKIEWICZKIEWICZKIEWICZKIEWICZKIEWICZKIEWICZKIEWICZKIEWICZKIEWICZKIEWICZKIEWICZKIEWICZKIEWICZKIEWICZKIEWICZKIEWICZKIEWICZKIEWICZKIEWICZKIEWICZKIEWICZKIEWICZKIEWICZ
1:47
And the two things that helped me most was a course at Stanford
1:50
about cracking the
1:51
SPEAKER_03: Google interview and your article,
1:53
SPEAKER_03: Get That Job at Google.
1:54
SPEAKER_03: And what really stuck with me, this article is still up there.
1:57
And I just tweeted recently that I think after like almost 15
2:00
years,
2:01
it's still very relevant.
2:03
One of the things I really liked is you put this important
2:06
takeaway is if you don't
2:07
SPEAKER_03: get an offer,
2:08
SPEAKER_03: you may still be qualified to work there.
2:10
SPEAKER_03: So don't blow your ego at all.
2:14
SPEAKER_03: What motivated you to write this article?
2:15
SPEAKER_03: Yeah.
2:16
SPEAKER_01: Getting turned down by a bunch of places.
2:20
SPEAKER_01: No, you know, it's true.
2:21
SPEAKER_01: Actually, a lot of my friends got turned down.
2:23
SPEAKER_01: I knew they were good, right?
2:24
SPEAKER_01: So I saw the false positives,
2:26
SPEAKER_01: or sorry,
2:26
SPEAKER_01: false negatives,
2:28
SPEAKER_01: because they were so scared of a false positive.
2:30
SPEAKER_01: And they just, they were Google, and they could just turn people away.
2:33
SPEAKER_01: Turn great talent away.
2:34
SPEAKER_03: This was Google in 2008.
2:36
SPEAKER_03: So, like, this was, they barely went public.
2:38
SPEAKER_03: They were the hottest thing.
2:39
SPEAKER_03: They joined in 2005, actually.
2:40
SPEAKER_03: You joined in 2005, right?
2:44
SPEAKER_01: Yeah,
2:44
SPEAKER_01: so by the time I wrote that,
2:45
SPEAKER_01: I had seen three years of interviewing there and I knew what it took,
2:48
SPEAKER_01: right?
2:49
And I don't think it's changed that much in the last 15 years
2:52
or whatever.
2:53
SPEAKER_03: This episode was brought to you by WorkOS.
2:55
SPEAKER_03: If you're building a SaaS app,
2:57
SPEAKER_03: at some point,
2:57
your customers will start asking for enterprise features like
2:59
SAML authentication,
3:01
SPEAKER_03: SKIM provisioning,
3:02
SPEAKER_03: and fine-grained authorization.
3:04
SPEAKER_03: That's where WorkOS comes in,
3:05
SPEAKER_03: making it fast and painless to add enterprise features to your app.
3:09
Their APIs are easy to understand and you can ship quickly and
3:11
get back to building
3:12
SPEAKER_03: other features.
3:14
WorkOS also provides a free user management solution called
3:17
AuthKit for up to 1
3:18
SPEAKER_03: million monthly active users.
3:20
It's a drop-in replacement for Auth0 and comes standard with
3:22
useful features like
3:23
SPEAKER_03: domain verification,
3:24
SPEAKER_03: role-based access control,
3:26
SPEAKER_03: bot protection,
3:27
SPEAKER_03: and MFA.
3:28
SPEAKER_03: It's powered by Radix components, which means zero compromises in design.
3:31
You get limited customizations as well as modular templates
3:34
designed for quick integrations.
3:36
SPEAKER_03: Today,
3:37
SPEAKER_03: hundreds of fast-growing startups are powered by WorkOS,
3:39
SPEAKER_03: including ones you probably know,
3:41
SPEAKER_03: like Cursor,
3:41
SPEAKER_03: Vercel,
3:42
SPEAKER_03: and Perplexity.
3:43
SPEAKER_03: Check it out at workos.com to learn more.
3:46
SPEAKER_03: That is workos.com.
3:49
SPEAKER_03: If you want to build a great product, you have to ship quickly.
3:52
SPEAKER_03: But how do you know what works?
3:54
SPEAKER_03: More importantly, how do you avoid shipping things that don't work?
3:59
SPEAKER_03: The answer, Statsik.
4:01
SPEAKER_03: Statsik is a unified platform for flags,
4:03
SPEAKER_03: analytics,
4:04
SPEAKER_03: experiments,
4:04
SPEAKER_03: and more,
4:05
combining five plus products into a single platform with a
4:08
unified set of data.
4:10
SPEAKER_03: Here's how it works.
4:12
SPEAKER_03: First, Statsik helps you ship a feature via feature flag or config.
4:16
SPEAKER_03: Then,
4:17
SPEAKER_03: it measures how it's working,
4:18
SPEAKER_03: from alerts and errors,
4:19
SPEAKER_03: to replays of people using that feature,
4:22
SPEAKER_03: to measurement of top-line impact.
4:24
SPEAKER_03: Then you get your analytics,
4:25
SPEAKER_03: user account metrics,
4:26
SPEAKER_03: and dashboards to track your progress over time,
4:29
SPEAKER_03: all linked to the stuff you ship.
4:30
SPEAKER_03: Even better,
4:31
SPEAKER_03: Statsik is incredibly affordable with a super generous free tier,
4:35
SPEAKER_03: a starter program with $50,000 of free credits,
4:37
SPEAKER_03: and custom plans to help you consolidate your existing spend on flags,
4:41
SPEAKER_03: analytics,
4:41
SPEAKER_03: or A-B testing tools.
4:43
SPEAKER_03: To get started, go to statsik.com slash pragmatic.
4:47
SPEAKER_03: That is S-T-A-T-S-I-G dot com slash pragmatic.
4:50
SPEAKER_03: Happy building!
4:52
SPEAKER_03: Yeah,
4:52
and one thing that you wrote about is this thing called the
4:55
interview anti-loop,
4:56
SPEAKER_03: which I never heard about until then.
4:59
SPEAKER_03: What is it?
5:00
SPEAKER_03: Does it still exist?
5:01
SPEAKER_01: I mean,
5:01
SPEAKER_01: I made it up,
5:02
SPEAKER_01: but I mean,
5:03
SPEAKER_01: it's a phenomenon that I observed that everybody knows about,
5:05
that it was the one thing in that post that recruiting and HR
5:09
were a little,
5:11
SPEAKER_01: you know,
5:11
SPEAKER_01: I mean,
5:12
SPEAKER_01: worried about me publishing it.
5:13
And I was like,
5:14
well,
5:14
there's no point in doing the post if we don't talk about it,
5:16
right?
5:16
SPEAKER_01: Let's just be, it'll give us some credibility.
5:18
SPEAKER_01: Yeah.
5:19
SPEAKER_01: And I think it did ultimately, right?
5:20
SPEAKER_01: Which is,
5:21
SPEAKER_01: look,
5:22
you could just get unlucky and accidentally get the six people
5:24
at the company who
5:25
SPEAKER_01: just disagree with you the most on everything technical.
5:28
SPEAKER_01: Yeah.
5:29
SPEAKER_01: Right?
5:29
SPEAKER_01: And it's just like, just bad luck.
5:31
SPEAKER_03: In fact,
5:31
I think a lot of tech companies have this policy or at least
5:35
used to have it until
5:36
SPEAKER_03: recently.
5:37
SPEAKER_03: Maybe they'll do that.
5:38
SPEAKER_03: You can reapply after six months exactly for this reason.
5:41
SPEAKER_01: Yep.
5:42
SPEAKER_01: And I knew a bunch of people who reapplied at Google multiple times.
5:46
One guy I knew got in on his fifth attempt and then went on to
5:50
get promoted really
5:51
SPEAKER_01: fast and rise up the ranks and everything.
5:52
He was very obviously a false negative,
5:54
but it just took a bunch of tries to get in.
5:56
So a super critical criticism that a lot of people who read
5:59
that article have is
6:01
SPEAKER_03: like,
6:02
SPEAKER_03: well,
6:02
SPEAKER_03: oh,
6:02
SPEAKER_03: if this is what it takes in to get Google or Meta or whatever,
6:06
SPEAKER_03: which is,
6:06
SPEAKER_03: you know,
6:06
SPEAKER_03: it's my skill,
6:08
SPEAKER_03: not my matter as much as the interviewers,
6:11
SPEAKER_03: I don't want to do that.
6:12
Like,
6:12
I really appreciate that you just like you didn't hold back and
6:15
you just kept it real.
6:15
SPEAKER_03: But what is your take on people who are like, well, that's not fair.
6:19
SPEAKER_03: It's not meritocracy, that kind of stuff.
6:22
SPEAKER_01: You know, interviewing is not really a very good signal.
6:25
SPEAKER_01: I empathize with their viewpoint.
6:27
SPEAKER_01: In fact,
6:28
SPEAKER_01: at several points in my career,
6:29
SPEAKER_01: I've sort of kind of given up on interviewing and just said,
6:33
SPEAKER_01: like,
6:33
SPEAKER_01: you guys do it.
6:34
There's a lot of people who think they're really good at it and
6:36
they think that
6:37
SPEAKER_01: they know how to do it well and so on.
6:38
Even though the statistics at Google,
6:40
they ran many,
6:40
many statistical analyses and found that –
6:43
SPEAKER_01: There isn't really a lot of correlation between,
6:46
SPEAKER_01: you know,
6:46
how you score and whether you get an offer and whether you get
6:48
an offer and whether
6:49
SPEAKER_01: you do well and so on.
6:50
SPEAKER_01: And so I kind of lost faith in the process a little bit.
6:54
SPEAKER_01: I noticed that I was a referral.
6:56
I was a reference,
6:56
I should say,
6:57
for a buddy of mine who was applying at Anthropic.
7:02
SPEAKER_01: Recently, right?
7:03
SPEAKER_01: And I got a call, right?
7:04
SPEAKER_01: Just a regular reference call.
7:06
SPEAKER_01: Reference call, yeah.
7:07
SPEAKER_01: And the person was the hiring manager, not a recruiter.
7:11
SPEAKER_01: And the hiring manager talked to me for probably at least 40 minutes,
7:14
SPEAKER_01: digging into all the things that you don't pick up in an interview,
7:18
SPEAKER_01: right?
7:18
SPEAKER_01: Because he recognized,
7:19
SPEAKER_01: just like we do,
7:20
SPEAKER_01: that interviewing is a really flawed process and it's a trade-off.
7:25
that the company has to make between sort of like effort that
7:28
they expend trying to
7:29
SPEAKER_01: find good candidates and being really accurate in their assessments.
7:33
SPEAKER_01: That's a tradeoff.
7:35
SPEAKER_03: Yeah,
7:35
SPEAKER_03: and then,
7:36
SPEAKER_03: interesting enough,
7:38
SPEAKER_03: some people are saying,
7:39
SPEAKER_03: I guess a lot of people are saying this is unfair.
7:42
SPEAKER_03: There's also a criticism of coding interviews, lead code, etc.
7:45
SPEAKER_03: And they're like, why can't these companies just ask me to do the work?
7:50
SPEAKER_03: And then,
7:51
SPEAKER_03: plot twist,
7:51
SPEAKER_03: some companies are doing that these days,
7:53
like Linear and some of the formal companies who have a strong
7:57
enough brand.
7:57
SPEAKER_03: They're like,
7:57
SPEAKER_03: we will pay you your day rate,
8:00
SPEAKER_03: week rate,
8:01
SPEAKER_03: and for a week,
8:02
SPEAKER_03: you will work with us remotely.
8:04
SPEAKER_03: Now, of course...
8:05
And I'm actually talking with the engineer manager on my team,
8:10
their first engineer manager.
8:10
SPEAKER_03: He's like,
8:11
you can use AI tools like they're immune to everything because
8:13
you're actually
8:14
SPEAKER_03: doing the work.
8:14
SPEAKER_03: Now, the downside is it's a week of your life, right?
8:17
SPEAKER_03: And people are like, well, I can't interview at five different places.
8:19
SPEAKER_03: And I feel, you know, there's all these trades.
8:21
SPEAKER_03: I thought, well, yeah, but now it is real world, right?
8:23
SPEAKER_03: So there's this spectrum of interviews.
8:25
SPEAKER_03: And as you said, like in the end, just, I guess, pick your poison, right?
8:28
SPEAKER_01: That's right.
8:29
SPEAKER_01: That's right.
8:29
SPEAKER_01: And I know, look, man, I've been in the industry for 30, 35 years.
8:32
I've seen people try all sorts of different variations on
8:35
trying to improve this.
8:36
SPEAKER_01: Like,
8:37
the first company I worked for required you to do a six-month
8:39
co-op before you
8:40
SPEAKER_01: could get a full-time offer there.
8:41
SPEAKER_01: What's that?
8:42
SPEAKER_01: GeoWorks.
8:42
SPEAKER_01: GeoWorks.
8:43
SPEAKER_01: And they had probably the highest hiring bar I've ever seen.
8:46
And they got acquired by Amazon,
8:48
and Amazon was just blown away by their hiring bar.
8:50
Yeah.
8:50
SPEAKER_03: In fact,
8:50
SPEAKER_03: we should probably mention,
8:51
SPEAKER_03: I mean,
8:52
SPEAKER_03: I think you and me have both seen this,
8:53
but there's this like open secret in the industry where if you
8:56
go to the website
8:57
SPEAKER_03: for like Google,
8:58
SPEAKER_03: Meta,
8:58
SPEAKER_03: a bunch of big tech,
9:00
SPEAKER_03: even Microsoft,
9:01
SPEAKER_03: you're not going to see software development engineer one.
9:05
SPEAKER_03: Because they fill all those up with interns.
9:07
SPEAKER_03: So the internship is actually a recruitment operation.
9:10
SPEAKER_01: It is.
9:10
SPEAKER_01: It is.
9:11
SPEAKER_01: It's a really cutthroat.
9:13
SPEAKER_01: College hiring is super cutthroat in the industry.
9:15
And the big companies like Microsoft,
9:17
you know,
9:17
and Google,
9:17
they sort of dominate it.
9:18
SPEAKER_01: They have the resources to build all the relationships with the schools.
9:22
SPEAKER_01: And it's, yeah, so they get the cream of the crop, you know.
9:26
SPEAKER_03: Yeah.
9:26
SPEAKER_03: And then they fill up an entry-level internship.
9:28
SPEAKER_01: I'm really proud of any intern that goes off to a startup, really.
9:32
SPEAKER_03: I actually just talked with someone.
9:34
SPEAKER_03: She'll be on the podcast.
9:36
SPEAKER_03: She had returned offers from Microsoft and Google.
9:39
SPEAKER_03: And she talked with her mentor at Microsoft.
9:41
SPEAKER_03: It's a good mentor.
9:42
SPEAKER_03: And the mentor was saying,
9:43
SPEAKER_03: look,
9:44
SPEAKER_03: you can do big tech,
9:45
SPEAKER_03: but with startups,
9:46
SPEAKER_03: you have a very different skill set.
9:47
SPEAKER_03: And she thought about it for a long time.
9:49
SPEAKER_03: And in the end, she took a risk and she went to CODA.
9:53
SPEAKER_03: She's now at OpenAI, actually, but I think that experience helped her.
9:55
SPEAKER_03: And she talked through her mentality,
9:58
SPEAKER_03: and I was like,
9:58
SPEAKER_03: wow,
9:59
SPEAKER_03: she sounded like a wise,
10:01
SPEAKER_03: experienced person.
10:02
SPEAKER_03: And, yeah, I did not expect it because it was paved.
10:07
SPEAKER_01: I see a lot of this, too.
10:08
SPEAKER_01: I mean, college kids are savvy these days.
10:11
SPEAKER_01: They know that stuff really inflects.
10:14
SPEAKER_01: And in fact,
10:14
SPEAKER_01: all the stuff we talked about,
10:16
even many of the things that we talked about in the blog post
10:18
that seemed timeless
10:19
SPEAKER_01: about getting a job at Google,
10:21
SPEAKER_01: getting a job is just hard as a software engineer right now.
10:23
SPEAKER_03: The other thing that really resonated with this article is,
10:26
SPEAKER_03: as you wrote,
10:27
SPEAKER_03: I'm going to quote it,
10:28
SPEAKER_03: when you get an offer from a tech company,
10:30
SPEAKER_03: you just happen to squeak by.
10:32
SPEAKER_03: And at the time when I read it, I didn't really believe it from outside.
10:36
SPEAKER_03: But now that I've also,
10:38
SPEAKER_03: you know,
10:38
SPEAKER_03: I've gotten jobs,
10:39
SPEAKER_03: I've been a hiring manager and made so many offers,
10:42
SPEAKER_03: you know,
10:42
SPEAKER_03: people who are coming in and they're like,
10:44
SPEAKER_03: oh,
10:44
SPEAKER_03: I smashed the interview.
10:45
SPEAKER_03: Actually,
10:46
SPEAKER_03: like out of maybe a hundred interviews,
10:48
SPEAKER_03: roughly that I've been the hiring manager at Uber.
10:51
There was,
10:52
like,
10:52
two that was,
10:53
like,
10:54
we had more than one person do a double thumbs up.
10:56
SPEAKER_03: We had thumbs up, double thumbs up.
10:57
SPEAKER_03: The rest were a mix of, like, thumbs up, thumbs down.
11:01
SPEAKER_03: And then we came to a decision.
11:02
SPEAKER_03: And it was, like, it could have gone either way.
11:04
SPEAKER_03: Like, when we went to the debrief.
11:07
SPEAKER_03: So, like...
11:08
SPEAKER_03: I now really appreciate it.
11:10
I feel this is one of the things which is hard to believe from
11:12
the outside.
11:13
SPEAKER_01: The best story is when I was at Google,
11:15
SPEAKER_01: I was on their hiring committee,
11:17
SPEAKER_01: which is a blind,
11:20
SPEAKER_01: double blind.
11:20
SPEAKER_01: They don't see the candidates.
11:22
SPEAKER_01: They don't know the interviewers who's doing it.
11:24
SPEAKER_01: They're just reading feedback packets.
11:26
SPEAKER_01: And the interviewers don't bias each other.
11:28
SPEAKER_01: And one day they didn't experiment with us, okay?
11:30
Because we were the ones that ultimately made that decision
11:33
that you just talked about,
11:34
right?
11:34
SPEAKER_01: The thumbs up, thumbs down type thing.
11:36
SPEAKER_01: Not the interviewers.
11:37
Google has a separate committee that actually looks at all the
11:39
feedback,
11:40
right?
11:41
And the recruiters did an exercise with us where they presented
11:44
a bunch of packets,
11:45
SPEAKER_01: hypothetical packets,
11:46
SPEAKER_01: say,
11:47
SPEAKER_01: of candidates who had been rejected or accepted.
11:51
SPEAKER_01: Actually, they didn't even tell us.
11:52
SPEAKER_01: They just said, these were just a bunch of candidates.
11:54
SPEAKER_01: We're going to go and do the process on them.
11:56
SPEAKER_01: We had feedback on them, though, okay?
11:58
SPEAKER_00: Yeah.
11:58
We went through and we evaluated them all and decided we were
12:00
going to not hire 60%
12:02
SPEAKER_01: of them,
12:03
SPEAKER_01: right?
12:03
SPEAKER_01: Mm-hmm.
12:03
SPEAKER_01: Have you figured this one out yet?
12:05
SPEAKER_01: No.
12:05
SPEAKER_02: No, not yet.
12:06
SPEAKER_01: We were reviewing our own packets, right?
12:09
SPEAKER_01: Yeah.
12:10
SPEAKER_01: So we voted not to hire 60 percent of ourselves.
12:13
SPEAKER_02: Yeah.
12:14
SPEAKER_01: OK.
12:14
SPEAKER_01: And it was a very sobering realization.
12:16
SPEAKER_01: And the next week or two was like the best time to apply to Google.
12:19
SPEAKER_01: So we were just like, come on through.
12:20
SPEAKER_01: Right.
12:21
SPEAKER_01: I mean, it was nuts.
12:22
SPEAKER_03: Well, because 60 percent is almost a coin toss.
12:24
SPEAKER_03: A coin toss is 50 percent.
12:25
SPEAKER_03: You're a little bit better.
12:27
SPEAKER_01: Right.
12:28
SPEAKER_01: And so,
12:29
SPEAKER_01: I mean,
12:29
SPEAKER_01: the whole I don't know,
12:30
the whole process is also so heavily biased towards whether you
12:32
like the person or
12:33
SPEAKER_01: not.
12:33
SPEAKER_01: You know, a lot of the decisions made in the first 10 seconds, they say.
12:37
SPEAKER_01: Yeah.
12:37
SPEAKER_03: Yeah,
12:37
SPEAKER_03: but my takeaway,
12:39
SPEAKER_03: and I think different people take different things,
12:41
SPEAKER_03: but the reason that really helped me,
12:42
SPEAKER_03: not just at that time when I got this first job in the UK,
12:45
SPEAKER_03: but actually I read Reddit later when,
12:48
SPEAKER_03: for example,
12:48
SPEAKER_03: later applied to Facebook.
12:50
SPEAKER_03: I narrowly didn't, but I didn't get it.
12:52
SPEAKER_03: And actually that rejection helped me get that position at Uber.
12:56
SPEAKER_03: which all of these are just cut to it.
12:58
SPEAKER_03: And what I took away from it is, this is how the process is.
13:02
SPEAKER_03: You might not like it,
13:03
SPEAKER_03: but you can either just complain or think it's unjust,
13:07
SPEAKER_03: or you can know it's unjust and you know that you just need to try hard.
13:10
SPEAKER_03: And when you do get it, don't take it for granted.
13:13
SPEAKER_01: So how did getting rejected by Facebook help you get a job at Uber?
13:16
SPEAKER_01: Because if it's helpful, I'll go get rejected at Facebook.
13:19
SPEAKER_03: What was helpful is I did a bunch of time preparing for Facebook.
13:24
SPEAKER_03: It was very clear at the time that they actually sent me materials.
13:28
SPEAKER_03: And the preparation did not go to waste.
13:31
SPEAKER_03: So,
13:31
SPEAKER_03: you know,
13:31
SPEAKER_03: I learned how to do the algorithmic coding,
13:33
SPEAKER_03: big O,
13:34
SPEAKER_03: like I knew some of that before,
13:36
SPEAKER_03: but I really refreshed it on the spot on Facebook for the system design.
13:41
I thought I nailed it because I heard the question before and I
13:44
just like drew up
13:45
SPEAKER_03: like it was like design Instagram,
13:46
SPEAKER_03: like I got this.
13:48
SPEAKER_03: you know, no conversation with the person.
13:50
And later I kind of got some feedback on like,
13:52
you know,
13:53
what I didn't do.
13:53
SPEAKER_03: And so by the time I got to Uber,
13:54
SPEAKER_03: I actually heard that,
13:56
SPEAKER_03: like,
13:56
SPEAKER_03: again,
13:56
SPEAKER_03: not many people got double thumbs up,
13:58
SPEAKER_03: but in hindsight,
13:59
SPEAKER_03: I kind of got the,
14:01
SPEAKER_03: I did get like two or three double thumbs up because I have practiced and
14:05
Also,
14:05
I think the other thing is that Uber at the time,
14:06
this was Amsterdam.
14:07
SPEAKER_03: So in London, a lot of people knew how to interview.
14:10
SPEAKER_03: Amsterdam, Uber struggled to have people who understood these interviews.
14:14
SPEAKER_03: So I guess I stood out because I prepared a year earlier.
14:17
SPEAKER_03: So the preparation does not go to waste.
14:19
SPEAKER_03: So...
14:20
SPEAKER_01: Yeah, preparation is so important.
14:22
SPEAKER_01: So important.
14:24
SPEAKER_01: But boy, what do you prepare for now?
14:26
SPEAKER_01: Like I've got a buddy who's out interviewing right now.
14:28
SPEAKER_01: He's just a very senior engineer.
14:30
And he says that the teams are all asking,
14:34
they want somebody to come teach them AI.
14:37
SPEAKER_01: That's what everyone's doing.
14:38
SPEAKER_01: So they want someone who knows AI because they don't.
14:40
SPEAKER_01: That's the theme right now.
14:42
SPEAKER_01: So what do you prepare for?
14:44
SPEAKER_03: Well,
14:44
SPEAKER_03: I just talked with someone,
14:45
SPEAKER_03: again,
14:46
SPEAKER_03: she'll be on the podcast,
14:47
SPEAKER_03: Janvi,
14:48
SPEAKER_03: who interviewed 46 AI companies.
14:50
SPEAKER_03: She's the engineer who went to Coda, became an AI engineer there.
14:55
SPEAKER_03: So she interviewed 46, and she said it's a mess.
14:58
SPEAKER_03: And this is for mid-level, so we're not talking staff level, but
15:02
SPEAKER_03: A lot of them are still doing the usual lead code style interviews.
15:06
SPEAKER_03: And then they might ask you a few things about AI.
15:09
And she said that there is one new type of project that she
15:11
actually really likes
15:12
SPEAKER_03: is a project,
15:13
SPEAKER_03: especially for AI,
15:13
SPEAKER_03: you know,
15:14
SPEAKER_03: build something based on AI.
15:15
And she said she loves it because she can actually show off
15:18
what she's capable of doing.
15:19
SPEAKER_03: It seems it's a mess.
15:20
SPEAKER_03: I don't think people know what to do.
15:22
SPEAKER_03: Yeah.
15:23
SPEAKER_03: I don't think even a lot of companies know what AI engineer is.
15:25
SPEAKER_03: We'll get into this.
15:26
SPEAKER_03: But before we go,
15:28
SPEAKER_03: so you wrote the Get That Job at Google in 2008,
15:31
SPEAKER_03: and 10 years later,
15:32
SPEAKER_03: you wrote another one called Get That Job at Grab.
15:35
SPEAKER_03: You were at Grab.
15:36
Now,
15:36
you would think that these two are kind of connected,
15:39
but Get That Job at Grab
15:41
SPEAKER_03: was more of an article about the job market at the time in 2018.
15:46
You wrote,
15:46
I'll quote,
15:47
because something very strange is going on in industry.
15:50
It started maybe a couple of years ago and it escalated a lot
15:52
around a year ago.
15:53
SPEAKER_03: And then what was completely crazy about six months ago,
15:57
What happened is this global demand for software engineers
15:59
completely outsized supply.
16:00
And I think it might be happening because we missed the market
16:03
correction.
16:03
SPEAKER_03: Sometimes the past five years,
16:05
it was the article was basically a bit of a heads of saying the
16:08
market is really
16:08
SPEAKER_03: hot.
16:09
SPEAKER_03: And now that I read it back, I was a bit of amazed because.
16:13
SPEAKER_03: You wrote this one or two years before anyone mentioned it.
16:17
SPEAKER_03: It was happening.
16:18
SPEAKER_03: It was heating up to be the hottest job market.
16:20
SPEAKER_03: And, you know, we saw it in 2021.
16:23
SPEAKER_03: It was the peak.
16:25
You saw this and you were pretty much advertising it to anyone
16:27
who was actually
16:29
SPEAKER_03: listening to whatever you were preaching.
16:31
SPEAKER_01: Yeah,
16:32
SPEAKER_01: well,
16:32
SPEAKER_01: I mean,
16:32
they're the early warning system that recruiters are that will
16:36
tell you what's
16:36
SPEAKER_01: going on with the market,
16:37
SPEAKER_01: right?
16:37
Because they're directly in touch with the hiring managers who
16:39
are the ones who
16:40
SPEAKER_01: are,
16:41
SPEAKER_01: you know,
16:42
in touch with the people with the budgets who are deciding what
16:44
the company is
16:44
SPEAKER_01: going to focus on.
16:45
SPEAKER_01: And so the recruiters,
16:46
SPEAKER_01: if you're in touch with your recruiter network,
16:48
SPEAKER_01: right,
16:48
SPEAKER_01: you know kind of what the trends are and all that stuff.
16:51
SPEAKER_01: And so I started noticing that the world was running out of engineers.
16:54
SPEAKER_03: Yeah.
16:54
SPEAKER_01: That's fundamentally what was happening back then.
16:57
SPEAKER_03: Yeah.
16:57
SPEAKER_03: And I mean,
16:58
SPEAKER_03: you know,
16:58
SPEAKER_03: like you also,
16:59
SPEAKER_03: I think some people were externally,
17:01
SPEAKER_03: it looked a bit surprising because you were,
17:03
SPEAKER_03: you were doing great at Google and you went to this scale up grab.
17:06
SPEAKER_03: I mean,
17:06
SPEAKER_03: they're growing fast,
17:07
SPEAKER_03: but I think some people are thinking,
17:08
SPEAKER_03: well,
17:09
SPEAKER_03: why is TV going after Google to grab?
17:12
SPEAKER_03: Why were you going, by the way?
17:14
Well,
17:14
well,
17:14
you know,
17:15
I mean,
17:15
GeoWorks,
17:16
Amazon,
17:16
Google,
17:18
all really similar in a lot of ways.
17:21
SPEAKER_01: You know, GeoWorks was more like device software, but still, right?
17:26
SPEAKER_01: Yeah.
17:27
SPEAKER_01: You know, Grab, I had a buddy from Google who was CTO there, right?
17:30
SPEAKER_01: Theo and Vasilakis.
17:32
SPEAKER_01: And he was like, man, this is an adventure.
17:34
SPEAKER_01: You got to come.
17:34
SPEAKER_01: So I started chatting with them and realized they were on just this,
17:37
SPEAKER_01: I mean,
17:38
that Southeast Asia in general is just this incredible
17:40
productivity explosion.
17:41
SPEAKER_01: And it just, it seemed fun, right?
17:43
SPEAKER_01: Yeah.
17:44
it turned out to be actually really fun it was and then code
17:47
killed it so you know
17:48
back back then like at the get this get that job job at grab
17:51
you did describe how
17:53
the market was was really heating up and you know some things
17:55
happened at covet but
17:57
what you wrote here is so so now there's a gut of investor
18:00
money as creating a lot
18:01
of startups a lot of startups including some very big ones and
18:04
they're gobbling up
18:05
SPEAKER_03: all the energy left on the planet and now it's a fight
18:08
SPEAKER_01: Yeah.
18:08
SPEAKER_01: It got worse after that.
18:10
SPEAKER_03: Yeah.
18:10
I was asked,
18:11
like,
18:11
how did you see it play out and how does it continue all this
18:14
today?
18:15
Because I feel today we might see something similar in a
18:17
different area,
18:18
right?
18:18
SPEAKER_01: Yeah.
18:19
SPEAKER_01: I mean, there's a lot of investment coming in for sure.
18:22
SPEAKER_01: It's coming in hot.
18:24
SPEAKER_01: Right.
18:24
We went through a huge spike right after I posted that because
18:29
shortly afterwards
18:30
SPEAKER_01: was COVID,
18:30
SPEAKER_01: right?
18:31
Two years later,
18:31
we had the stimulus package and that gave everybody a lot of
18:35
money.
18:35
SPEAKER_01: And that was like tons of startups appeared because of that.
18:37
SPEAKER_01: Yeah.
18:39
SPEAKER_01: So many founders.
18:41
SPEAKER_01: And so great time to be a remote engineer, basically.
18:43
SPEAKER_01: Right.
18:45
Then the stimulus package and the stimulus money went away and
18:49
things started to kind of crash.
18:50
SPEAKER_01: And then AI came out and everybody got really uncertain.
18:53
SPEAKER_01: And so it kind of dipped a little.
18:54
SPEAKER_01: It has dipped, I think.
18:56
SPEAKER_01: If you just look at Indeed's report,
18:58
SPEAKER_01: you can see jobs have dipped pretty heavily since their peak in 2021.
19:03
SPEAKER_01: 2021 or 2022.
19:05
But we also see a productivity explosion on its way,
19:08
like a boom of jobs coming.
19:10
SPEAKER_01: So it goes up and down.
19:12
SPEAKER_01: But yeah,
19:12
SPEAKER_01: I think at the time,
19:14
SPEAKER_01: at that time in 2018,
19:16
SPEAKER_01: the market was showing signs that it was going to.
19:18
SPEAKER_01: And that's what, look, that's what everybody wants.
19:20
SPEAKER_01: They want to predict what's going to happen.
19:22
SPEAKER_01: Not just that they know what stocks to buy, right?
19:23
But also,
19:24
you know,
19:24
how to make the right decisions for their companies or their
19:26
careers.
19:27
And right now,
19:28
I think you and I both agree that things are kind of headed
19:30
back up right now.
19:32
SPEAKER_03: Yeah, and we'll get into that.
19:35
SPEAKER_03: But I want to go back to the second time that.
19:38
SPEAKER_03: So the first time I came across your blog,
19:40
SPEAKER_03: I didn't really even connect the name with the face back then,
19:43
SPEAKER_03: was get a job at Google.
19:45
The second time was a few years later,
19:47
which was this Google platform rant,
19:50
which was...
19:51
SPEAKER_03: published on Google, Google Plus, right?
19:55
SPEAKER_03: Yeah.
19:57
SPEAKER_03: So it was an internal facing document.
20:00
Apparently you wrote a lot of these or just like rants or like
20:03
meant for Google internal only.
20:05
And somehow it was set to anyone could read it on the internet
20:08
and Hacker News jumped on it.
20:10
SPEAKER_03: And as soon as it went out, you know, people archived it as well.
20:15
SPEAKER_03: First of all,
20:16
SPEAKER_03: How did this rant came along?
20:17
SPEAKER_03: Because this rant has been so referenced.
20:19
SPEAKER_03: It's now,
20:21
SPEAKER_03: I think,
20:21
on GitHub as well as TV's platform rant because it was a really
20:24
good criticism of
20:25
SPEAKER_03: Google.
20:26
SPEAKER_03: And not just that,
20:27
SPEAKER_03: but it was a kind of a really,
20:29
SPEAKER_03: really realistic,
20:31
SPEAKER_03: like,
20:31
SPEAKER_03: picturing of Amazon,
20:33
SPEAKER_03: including Jeff Bezos,
20:35
SPEAKER_03: not giving a shit about your day,
20:37
SPEAKER_01: which I think,
20:38
SPEAKER_01: you know,
20:38
SPEAKER_01: people were like.
20:41
SPEAKER_01: He still doesn't, you know.
20:43
SPEAKER_03: Yeah, but it just felt very real and raw.
20:47
SPEAKER_03: And clearly, I understand it wasn't meant for public consumption.
20:50
But,
20:51
you know,
20:52
like,
20:52
hey,
20:53
did you write these kind of things all the time?
20:55
SPEAKER_03: Like, because we only saw this one thing.
20:57
And I've heard that you had a history of just internally just
21:00
keeping it really real.
21:03
SPEAKER_01: I had other ones internally, sure.
21:05
SPEAKER_01: None of them were quite that, I guess, accusatory or whatever.
21:09
SPEAKER_01: I mean, like I was really taking Google to task because I was fed up.
21:14
I'd been there six years and I still couldn't get a platform
21:16
out of anybody,
21:16
right?
21:17
SPEAKER_03: Yeah.
21:18
SPEAKER_03: Like Google to ship a proper platform.
21:20
SPEAKER_01: Even internally.
21:22
SPEAKER_01: Like the code search team didn't want to give me an API.
21:25
It's inconceivable today that you'd give somebody a REST API to
21:28
your stuff,
21:28
right?
21:29
SPEAKER_01: That's the way we think today.
21:30
SPEAKER_01: Well, outside of Google, inside of Google, who knows?
21:33
SPEAKER_01: They're just not really big on internal services.
21:35
SPEAKER_01: They're just like, he's our product.
21:36
SPEAKER_01: Yeah.
21:37
SPEAKER_01: It drove me nuts.
21:39
SPEAKER_01: Completely nuts.
21:39
SPEAKER_01: I went nuts.
21:40
SPEAKER_01: And then a bottle of wine later, I told him how it was.
21:46
SPEAKER_03: Yeah,
21:46
SPEAKER_03: so let's recall some of that part because I'm going to link it,
21:50
SPEAKER_03: obviously,
21:50
SPEAKER_03: so people can read it.
21:51
SPEAKER_03: But first,
21:52
you started summarizing on what Amazon did right and then what
21:56
you observed
21:57
SPEAKER_03: throughout your time,
21:57
SPEAKER_03: right?
21:57
SPEAKER_03: You were early Amazon, right?
21:59
SPEAKER_01: Yeah, early-ish, yeah.
22:01
SPEAKER_01: I got there in late 1998.
22:03
SPEAKER_01: It was pretty small back then.
22:05
SPEAKER_01: We were in one building in downtown Seattle, just a three-story building.
22:09
SPEAKER_01: Wow, that's it?
22:10
A four-story building of which we occupied three floors,
22:12
I guess is like it.
22:13
SPEAKER_01: And,
22:14
SPEAKER_01: yeah,
22:15
SPEAKER_01: there was just one data center at the time,
22:17
and it was just a very small – it already had a cult-like sort
22:20
of feel to it,
22:21
SPEAKER_01: right?
22:22
SPEAKER_01: An electric feel.
22:24
SPEAKER_01: I mean, a sense that there was something really magical going on.
22:28
So was this still the bookstore part,
22:32
or was it already expanding beyond books?
22:34
SPEAKER_03: Yeah.
22:34
We when I joined,
22:36
we already had music and I think we were just launching video.
22:42
SPEAKER_01: Yeah.
22:42
SPEAKER_01: So I think we had just just brought our tab.
22:44
SPEAKER_01: It was really early on.
22:45
SPEAKER_01: I have to go back.
22:46
SPEAKER_01: Yeah.
22:47
SPEAKER_01: Yeah.
22:48
And then,
22:49
like,
22:49
you know,
22:51
you said that that basically Jeff Bezos mandated platforms,
22:55
APIs.
22:56
SPEAKER_03: What did you do there?
22:57
You know,
22:57
it's interesting because everybody thinks that there's a real
23:00
memo.
23:00
SPEAKER_01: The memo was – Jeff wouldn't write an actual memo, right?
23:05
SPEAKER_01: Why the fuck would he do that?
23:07
SPEAKER_01: He just tells people stuff and it happens.
23:10
But the customer service organization in particular was – I was
23:14
in customer service
23:15
SPEAKER_01: tools at the time.
23:16
SPEAKER_01: I may have been running customer service tools at the time.
23:19
Bezos would sit with us every week in a meeting and we would
23:22
look at the top ten
23:24
SPEAKER_01: reasons that customers were contacting us,
23:26
SPEAKER_01: right?
23:26
SPEAKER_01: Right.
23:26
SPEAKER_01: And he'd want to know,
23:28
SPEAKER_01: why are these customers still contacting us,
23:29
SPEAKER_01: saying they're getting triple charge for their books as a translator,
23:32
SPEAKER_01: that kind of thing,
23:32
SPEAKER_01: right?
23:33
SPEAKER_01: Number one was always, where's my stuff, right?
23:36
SPEAKER_01: Customer service had a really interesting need.
23:39
SPEAKER_01: I've never thought about this before,
23:40
SPEAKER_01: but it may have been Jeff's sort of affinity for customer service,
23:43
SPEAKER_01: wanting to be the Earth's most customer-centric company,
23:46
that led him down this path of forcing people to open up their
23:49
APIs because the
23:50
SPEAKER_01: customer service team kept saying...
23:52
We can't make any changes to OBDOS,
23:54
you know,
23:54
our web server,
23:55
because that's their code.
23:56
SPEAKER_01: We can't get into the supply chain code.
23:58
SPEAKER_01: We can't get into the fulfillment center code.
24:00
SPEAKER_01: The customer, we can't help the customer.
24:02
SPEAKER_01: And Bezos was like, all right, tell you what.
24:05
SPEAKER_01: Right.
24:05
SPEAKER_01: I'm going to blast anybody standing in the way of that.
24:08
SPEAKER_01: And what that turned into was,
24:09
SPEAKER_01: well,
24:10
you need to provide something to the customer service technical
24:12
team that's not
24:14
them going and linking against your code and trying to get it
24:16
to run locally in
24:17
SPEAKER_01: some different environment.
24:18
SPEAKER_01: Right.
24:19
SPEAKER_01: Yeah.
24:19
SPEAKER_01: Which is what they were doing with this awful C++ code.
24:22
SPEAKER_01: So, yeah.
24:22
SPEAKER_01: So that's that's kind of the origin story.
24:24
SPEAKER_03: Yeah, and then this was, like, around early 2000s, right?
24:27
Like,
24:27
before,
24:28
you know,
24:28
we even had things like services or microservices.
24:31
SPEAKER_01: Yeah,
24:31
SPEAKER_01: well,
24:31
SPEAKER_01: back then,
24:32
SPEAKER_01: the services were things like,
24:33
SPEAKER_01: they were proprietary protocols like Corba,
24:37
SPEAKER_01: like Kipco and Tolerian,
24:39
SPEAKER_01: the PubSub things.
24:40
SPEAKER_01: And they were all really nasty binary formats.
24:44
SPEAKER_01: And there was this possibility to do REST, right?
24:47
It had been invented at the time,
24:48
but everybody was kind of poo-pooing it,
24:49
saying,
24:50
no,
24:50
no.
24:51
SPEAKER_01: Nobody really kind of understood it for that.
24:52
SPEAKER_03: No type safety, no protocol.
24:55
SPEAKER_01: Yeah, it took years.
24:57
SPEAKER_01: But it turned out, yeah, that's what you need.
24:58
SPEAKER_01: You need an API.
24:59
SPEAKER_01: And that was the origin of my rant, too, right?
25:02
SPEAKER_01: Which was, I talked about Amazon does stuff mostly wrong.
25:08
SPEAKER_03: This is how he started your memo.
25:09
SPEAKER_03: So that was actually fascinating to read.
25:11
SPEAKER_03: I think it was clear that you were on Google's side, right?
25:15
SPEAKER_03: Like,
25:15
SPEAKER_03: even though you're trashing Google,
25:17
SPEAKER_03: it very clearly came through that you actually,
25:19
SPEAKER_03: like,
25:19
SPEAKER_03: wanted to,
25:20
SPEAKER_03: like,
25:20
SPEAKER_03: shake...
25:21
things up like hello like like the memo when i read it felt
25:24
like hey like we should
25:26
be better than amazon yeah here's all the reasons and here's
25:29
the things that
25:29
they're doing better and it's not that hard we we just need to
25:32
be do that well and
25:33
then we will be better right i mean it made sense right yeah
25:37
and i i just i just
25:39
felt like we were good at everything else we were good at a lot
25:41
of stuff google was
25:42
SPEAKER_01: extraordinarily good at a lot of things that amazon had no clue how to do
25:47
SPEAKER_01: Really?
25:48
SPEAKER_01: And it took Amazon years to catch up to Google in a lot of things.
25:51
SPEAKER_03: So let's talk about that.
25:52
SPEAKER_03: What were the things that Google was just really good at?
25:55
SPEAKER_01: Like Google had one service called Stubby.
25:57
SPEAKER_01: I think I even mentioned it in the post called Stubby.
26:01
SPEAKER_01: Or sorry, Chubby.
26:02
SPEAKER_01: Chubby was the locking service.
26:04
SPEAKER_01: Chubby and stubby, they went together.
26:06
SPEAKER_01: Yeah, the distributed locking service.
26:08
SPEAKER_01: Those are not easy to implement, okay?
26:10
We're talking,
26:11
you know,
26:11
Paxos times 10,
26:12
you know,
26:12
make sure the thing stands up all the time.
26:14
SPEAKER_01: It had seven nines of availability, which is like – Yeah, yeah.
26:18
SPEAKER_01: It was like basically 30 seconds of downtime every 10 years, okay?
26:22
SPEAKER_01: It was a very reliable service.
26:24
SPEAKER_01: Wow, okay.
26:25
SPEAKER_01: Five nines is hard to get to.
26:28
SPEAKER_01: Amazon, seven, five is almost insane.
26:31
SPEAKER_01: Seven is just like, what?
26:32
SPEAKER_01: So that was just one example.
26:33
SPEAKER_01: Bigtable early on,
26:35
SPEAKER_01: they had like free,
26:36
basically like unlimited NoSQL storage with some pretty good
26:39
query facilities for
26:40
SPEAKER_01: everybody in the MapReduce infrastructure.
26:42
SPEAKER_01: Google invented it, you know, and on and on and on, right?
26:45
SPEAKER_03: So like really,
26:46
really good hardcore engineering problems solved in a like way
26:50
that is like just
26:52
SPEAKER_03: tough,
26:52
SPEAKER_03: tough to do.
26:53
SPEAKER_01: I was very impressed.
26:54
I slapped myself like my forehead sometimes when I was like,
26:57
I see some of the stuff they did.
26:59
SPEAKER_01: I got there and I'm like, why didn't I think of this?
27:01
SPEAKER_01: Like I had this game that I had a custom RPC protocol.
27:04
SPEAKER_01: When I looked at Google's,
27:05
SPEAKER_01: which is now GRPC,
27:07
SPEAKER_01: it was called protocol buffers and stubby back then.
27:09
You look at it and you're like,
27:10
oh,
27:10
wow,
27:10
it's a forward compatible protocol.
27:12
I can add stuff to it without breaking it,
27:14
but it's binary and high performance.
27:15
SPEAKER_01: And it was beautiful.
27:16
SPEAKER_01: It is beautiful.
27:17
SPEAKER_01: Surprised more people don't use it, to be honest.
27:19
So,
27:19
yeah,
27:20
they did a lot of things really well,
27:21
but they didn't do platforms well at all.
27:23
SPEAKER_01: It wasn't part of their DNA.
27:24
SPEAKER_01: They just didn't get it.
27:25
SPEAKER_03: And it was they didn't do internal or external or neither.
27:28
SPEAKER_01: Neither.
27:29
SPEAKER_03: Neither.
27:29
SPEAKER_03: Neither.
27:30
SPEAKER_03: And then so you wrote this rant,
27:32
SPEAKER_03: which,
27:32
SPEAKER_03: again,
27:32
SPEAKER_03: like I think if you're listening to this,
27:34
SPEAKER_03: you need to read that rant.
27:35
SPEAKER_03: It is like one of the best things I've read.
27:37
SPEAKER_03: It's also very entertaining, by the way.
27:41
SPEAKER_03: What was the impact?
27:43
SPEAKER_03: Because obviously you sent it internally.
27:45
SPEAKER_03: It now leaked externally.
27:46
SPEAKER_03: So clearly, you know, people were making fun of Google.
27:49
SPEAKER_03: Did it achieve that shakeup effect?
27:51
SPEAKER_03: And how high did this thing get?
27:53
SPEAKER_03: I'm pretty sure it must have gotten pretty high.
27:56
Well,
27:56
I mean,
27:56
Google had a very open culture,
27:58
so it got brought up at the next TGIF,
28:00
right?
28:01
SPEAKER_03: Thank God it's Friday, right?
28:02
SPEAKER_03: It's Google's iconic Friday meeting.
28:05
SPEAKER_03: It's like all hands-ish.
28:06
SPEAKER_01: I remember Ben,
28:07
SPEAKER_01: the guy who was in charge of our data centers at the time,
28:14
SPEAKER_01: he stood up there and said,
28:15
SPEAKER_01: well,
28:15
SPEAKER_01: you know,
28:15
SPEAKER_01: we all read the rant.
28:18
SPEAKER_01: So, you know, they got a kick out of it, right?
28:20
SPEAKER_01: But, you know, Vic Condotra was pissed.
28:22
SPEAKER_01: I mean, he was really, really mad, right?
28:25
Because I had,
28:26
like,
28:26
told him he had an ugly baby,
28:28
and very,
28:28
very loudly and publicly.
28:30
SPEAKER_03: Yeah.
28:31
SPEAKER_01: You know, and I'd used his ugly baby to do it.
28:34
SPEAKER_03: This was the developer saw Google.com, maybe, or something else.
28:39
SPEAKER_01: Google Plus.
28:40
SPEAKER_03: Oh, Google Plus.
28:40
SPEAKER_01: I called Google Plus ugly, right?
28:43
SPEAKER_01: Yeah.
28:44
SPEAKER_01: And he was really gunning for the head spot at the time.
28:48
SPEAKER_01: And he had planted the seed of fear in Larry Page.
28:50
SPEAKER_01: He was like, Facebook's going to kill us.
28:52
SPEAKER_01: Facebook's going to kill us.
28:52
SPEAKER_01: They're going to kill us, right?
28:53
SPEAKER_01: We had to have a Facebook.
28:55
SPEAKER_01: Which is stupid for many reasons.
28:58
SPEAKER_01: Oh, so I'll tell this again.
29:00
SPEAKER_01: I'll say this again.
29:02
SPEAKER_01: That blog rant,
29:03
SPEAKER_01: that famous rant,
29:04
was actually part two of an 11-part series that I had
29:06
meticulously planned out.
29:08
And I never finished because I accidentally published the
29:10
second one externally.
29:12
And the implications were actually so big that I was kind of
29:15
like in hiding for a while.
29:18
But yeah,
29:18
no,
29:18
I was actually picking apart Google Plus dimension by
29:21
dimension.
29:22
SPEAKER_01: And platforms was just one of the dimensions where it was failing.
29:25
SPEAKER_03: But you were actually right in hindsight.
29:28
SPEAKER_01: I was right about all of it.
29:30
SPEAKER_01: But they never said sorry.
29:31
SPEAKER_01: I was also right about not getting into publication ads.
29:33
SPEAKER_01: I was right about a lot of things at Google,
29:34
SPEAKER_01: but I'm not very good at convincing people that I'm not.
29:37
SPEAKER_03: So tell me that story,
29:38
SPEAKER_03: because you've said you've told me the story once in the newsletter,
29:41
SPEAKER_03: and we mentioned it super briefly.
29:43
SPEAKER_03: You killed publication ads.
29:45
And this was like,
29:47
as I remember,
29:48
like what happened is you joined Google.
29:49
SPEAKER_03: And then what did you do the first time?
29:51
SPEAKER_01: I went around to all the projects.
29:53
I was allowed to pick whatever I wanted and I picked print ads
29:55
because I thought it
29:56
SPEAKER_01: sounded like a cool challenge.
29:57
SPEAKER_01: I became a domain expert over the next six months,
29:59
learned everything there was to know about magazines and
30:02
newspaper publication ads
30:03
in the United States and concluded that we were never going to
30:06
make a dime,
30:07
that all of them hated us and blamed us for their declining
30:10
revenues and they
30:10
SPEAKER_01: wouldn't want to talk to us and we were evil.
30:14
SPEAKER_01: And I wrote it up as a big decision tree.
30:15
SPEAKER_01: I said, we could try this.
30:16
SPEAKER_01: We tried this.
30:17
SPEAKER_01: It didn't work.
30:17
SPEAKER_01: Tried this.
30:18
SPEAKER_01: The whole thing.
30:18
SPEAKER_01: I mapped out the entire decision tree of everything you could do.
30:22
SPEAKER_01: And they said, well, what about illegal stuff?
30:23
And I was like,
30:24
I'm not going to entertain any of that stupid stuff,
30:26
all right?
30:26
SPEAKER_01: It was like, they didn't put that in writing.
30:28
But,
30:29
you know,
30:29
it was like,
30:30
what if we just sucked up the phone book type stuff?
30:32
SPEAKER_01: Yeah.
30:33
SPEAKER_01: So, like, you know, I declined.
30:35
SPEAKER_01: And then they got mad and they sent it to other teams.
30:37
SPEAKER_01: And the other teams failed and came back to me for my postmortem.
30:40
SPEAKER_03: So they tried to make it work?
30:42
SPEAKER_01: Yeah.
30:42
SPEAKER_01: They tried again in Mountain View, and then they tried in England.
30:44
SPEAKER_01: And they couldn't do it because I was right.
30:47
SPEAKER_01: I never got so much as a I'm sorry or a thank you or anything like that.
30:51
SPEAKER_01: No.
30:51
SPEAKER_01: Yeah.
30:52
SPEAKER_03: But, like, you concluded this is not worth it.
30:54
So,
30:55
well,
30:55
first of all,
30:56
you said,
30:56
like,
30:57
if you were you,
30:57
you wouldn't do it.
30:58
SPEAKER_03: And then you moved on to the next thing.
30:59
SPEAKER_03: And then they failed to retweet what, like, sounds like twice.
31:03
SPEAKER_01: I did make a proposal in the postmortem,
31:05
SPEAKER_01: which was very similar to what ultimately turned into Groupon.
31:08
SPEAKER_03: Yeah.
31:09
SPEAKER_01: Yeah.
31:10
SPEAKER_01: So, you know, I mean, it wasn't like complete shooting.
31:13
SPEAKER_01: You could do one thing,
31:13
SPEAKER_01: but I said,
31:14
SPEAKER_01: you will need a sneaker network of like 8,000 people somehow.
31:17
SPEAKER_01: Yeah.
31:17
SPEAKER_01: Which is what Groupon ultimately did.
31:19
SPEAKER_03: It's fun to be right.
31:21
It sounds like,
31:22
you know,
31:22
you just like,
31:23
you know,
31:24
you did the best that you could.
31:26
SPEAKER_03: You gave the best.
31:27
And then you also like,
31:27
it sounds like you were like,
31:28
look,
31:29
if you want to try it,
31:29
like do it.
31:30
SPEAKER_03: But like, I don't believe, I believe this, this will not work.
31:34
SPEAKER_03: I believe we could try this and then just leave it at that.
31:36
SPEAKER_03: Right.
31:36
SPEAKER_03: Like, you know, you did what you believe then.
31:38
SPEAKER_01: Yeah.
31:39
SPEAKER_03: What do you think happened to Google Plus?
31:41
Like,
31:41
I remember,
31:42
you know,
31:43
Google launched Wave,
31:44
which kind of,
31:45
like,
31:45
died down pretty quickly.
31:47
SPEAKER_03: It was supposed to be the next email.
31:48
SPEAKER_03: That was the first time I was,
31:50
SPEAKER_03: like,
31:51
SPEAKER_03: I remember,
31:51
SPEAKER_03: like,
31:52
SPEAKER_03: this early 2010s,
31:53
SPEAKER_03: Google could not do anything wrong.
31:55
And every time they launched something,
31:56
I was like,
31:57
wow,
31:57
this is the next big thing.
31:57
SPEAKER_03: They launched Google App Engine.
31:59
SPEAKER_03: I was like, it's the coolest thing.
32:01
SPEAKER_03: And I onboarded it, and it was so cheap.
32:02
SPEAKER_03: It was ridiculously cheap.
32:04
SPEAKER_03: Later, I figured out why, because they were subsidizing it.
32:06
SPEAKER_03: Right.
32:06
But Google Wave was the first one where I remember all the
32:10
online portals,
32:12
SPEAKER_03: TechCrunch,
32:12
SPEAKER_03: et cetera,
32:12
SPEAKER_03: I was like,
32:13
SPEAKER_03: Google has replaced email.
32:14
SPEAKER_03: And we're like, oh, wow, Google has replaced email.
32:16
SPEAKER_03: And you tried it out and it didn't work.
32:18
SPEAKER_03: And then Google Plus came along.
32:19
And I think we understood from the outside,
32:22
not as Googlers,
32:23
that Google was...
32:25
SPEAKER_03: trying to really take on Facebook.
32:27
SPEAKER_03: And if they didn't succeed, you know, Facebook would win.
32:29
And I don't think we,
32:30
from the outside,
32:31
it seemed like it was kind of going,
32:33
going.
32:33
SPEAKER_03: Yeah, it was pretty ugly.
32:35
SPEAKER_03: But then it just kind of stopped.
32:37
SPEAKER_03: You were on the inside.
32:38
SPEAKER_03: Like, how did this play out?
32:39
Because I think we've heard there's,
32:40
like,
32:41
books about,
32:41
like,
32:42
Facebook's went all in wartime.
32:45
SPEAKER_03: They were working, you know, hard.
32:47
SPEAKER_03: And they actually saw this as a major threat.
32:49
SPEAKER_03: And it energized them.
32:51
SPEAKER_03: What do you think might have been drawn there?
32:52
SPEAKER_03: Or like, how much advantage did you then have on this?
32:55
SPEAKER_01: I mean,
32:56
SPEAKER_01: I was there,
32:57
SPEAKER_01: you know,
32:58
SPEAKER_01: I talked to people who were in the heat of it,
32:59
SPEAKER_01: you know,
33:00
SPEAKER_01: and I'm like,
33:01
SPEAKER_01: Wave was targeting a space that ultimately got solved by Slack.
33:06
SPEAKER_01: Slack was the right form factor and Wave wasn't.
33:08
SPEAKER_01: And when I saw Wave,
33:09
SPEAKER_01: I was totally unimpressed,
33:10
SPEAKER_01: but it was like they had cast a spell over everybody.
33:12
SPEAKER_01: And I didn't get it, right?
33:15
SPEAKER_01: But I got Slack instantly, right?
33:17
SPEAKER_01: And we all did.
33:19
SPEAKER_01: So it was very similar, I think.
33:21
SPEAKER_01: It was Google had trouble, struggled to find the right form factor.
33:25
SPEAKER_01: This was why I wrote that 11-part series.
33:27
It's because I knew that if they basically acted right then and
33:31
got Reddit,
33:32
SPEAKER_01: just took them,
33:32
SPEAKER_01: just bought Reddit,
33:33
SPEAKER_01: okay,
33:35
SPEAKER_01: and took over that sort of that social network,
33:38
SPEAKER_01: they would have had something.
33:39
SPEAKER_01: They would have had something.
33:41
SPEAKER_01: This was long before Reddit was in the top 10 in the U.S., right?
33:44
SPEAKER_01: Reddit was hot, but only like tech geeks, right?
33:47
SPEAKER_03: Yeah.
33:48
SPEAKER_03: Dick was also big back then.
33:50
SPEAKER_01: Yeah, as before, pre-digs, you know, blow up or whatever.
33:55
SPEAKER_01: Oh, yeah.
33:56
So I wanted Google to either build a Reddit that was done kind
34:00
of like slightly
34:01
SPEAKER_01: better because,
34:02
SPEAKER_01: you know,
34:02
SPEAKER_01: Reddit evolved and even they want to change it or something,
34:06
SPEAKER_01: fix a lot of things.
34:07
Because it had to be different from Facebook or people wouldn't
34:10
be able to migrate
34:11
SPEAKER_01: because of the network effect fundamentally.
34:13
SPEAKER_01: Yeah.
34:13
SPEAKER_01: Right.
34:14
SPEAKER_01: And Google just I mean, it's so weird, man.
34:16
SPEAKER_01: Companies are like people.
34:17
SPEAKER_01: They're like human beings.
34:18
They like they like they just they make decisions and the
34:22
decisions can be just
34:23
SPEAKER_01: absolutely terrible.
34:24
SPEAKER_01: And everyone around them knows it.
34:25
SPEAKER_01: And they're all embarrassed.
34:26
And they try to tell the company and the company's like,
34:28
don't tell me what to do.
34:30
SPEAKER_03: Yeah, it feels like it.
34:32
SPEAKER_03: So now looking back so many years later,
34:35
SPEAKER_03: you know,
34:36
SPEAKER_03: you've left Amazon,
34:37
SPEAKER_03: like,
34:37
SPEAKER_03: I don't know,
34:38
SPEAKER_03: like 10 plus years,
34:39
SPEAKER_03: even more.
34:40
SPEAKER_03: It's the same with Google.
34:41
SPEAKER_03: How do you think Amazon... 20 years.
34:44
SPEAKER_03: 20 years.
34:45
SPEAKER_03: Yeah.
34:45
SPEAKER_03: How do you think both Amazon and Google have changed,
34:48
SPEAKER_03: but also in what sense have they not changed?
34:52
SPEAKER_01: I think Amazon's changed way more than Google.
34:54
SPEAKER_01: You're the first person ever to ask me this, so thank you.
34:56
Amazon has improved dramatically in almost every possible way
35:00
that you could improve.
35:01
SPEAKER_01: Really?
35:01
SPEAKER_01: Yeah.
35:02
SPEAKER_01: Amazon has always executed better than anybody on Earth,
35:05
SPEAKER_01: but they found a way to do it without,
35:07
SPEAKER_01: you know,
35:08
SPEAKER_01: having all of the flaws that I mentioned at the beginning of my post.
35:11
SPEAKER_00: Yeah.
35:11
SPEAKER_01: Yeah.
35:13
SPEAKER_01: They've it's really it's quite nice now.
35:15
SPEAKER_01: And people that I know who work there are pretty,
35:18
SPEAKER_01: pretty satisfied and they're doing well and they still execute well.
35:23
They're a company that makes good decisions by and large,
35:26
just like Apple.
35:27
SPEAKER_01: Of course, they fall on their face once in a while.
35:28
SPEAKER_01: What company doesn't?
35:29
SPEAKER_01: Yeah.
35:31
SPEAKER_01: Google has not changed since the fucking day I joined.
35:36
SPEAKER_03: So recently,
35:39
SPEAKER_03: someone at Google was asking me about,
35:41
SPEAKER_03: like,
35:42
SPEAKER_03: what do I think about Google's developer story?
35:44
SPEAKER_03: And I said, like, do you want me to be honest?
35:46
SPEAKER_03: I said, developer what?
35:48
SPEAKER_03: And my example that I showed this person is Flutter versus React Native.
35:54
SPEAKER_03: Now,
35:55
React Native is about 10 full-time people at Facebook and a few
36:00
other in the core
36:01
SPEAKER_03: team and maybe a few other people from some other companies,
36:04
SPEAKER_03: maybe like 15 person.
36:05
SPEAKER_03: But Facebook invests like 10 full-time people.
36:08
SPEAKER_03: And if you go to the showcase page of React Native,
36:12
SPEAKER_03: which is,
36:12
SPEAKER_03: you know,
36:13
SPEAKER_03: where you show,
36:14
SPEAKER_03: you immediately see logos,
36:15
SPEAKER_03: Meta,
36:16
SPEAKER_03: Microsoft,
36:17
SPEAKER_03: Amazon.
36:17
SPEAKER_03: I think they have someone big, just like flagship apps.
36:22
SPEAKER_03: And then you have, oh, and then you have Shopify.
36:24
SPEAKER_03: You have like all these big companies.
36:26
SPEAKER_03: And, you know, you will find the blog post.
36:27
SPEAKER_03: Shopify says why we went all in on React Native,
36:30
SPEAKER_03: why we have thousands of developers working on React Native.
36:33
SPEAKER_03: And you have all these case studies.
36:35
SPEAKER_03: React Native is inside of Meta's Facebook app.
36:38
SPEAKER_03: It's inside Instagram.
36:39
It doesn't run the whole thing,
36:40
but it's in there,
36:41
obviously,
36:41
their ads app.
36:42
SPEAKER_03: And then you go to Flutter page.
36:44
SPEAKER_03: Now, Flutter has at least 50 full-time people, so five times as many.
36:48
SPEAKER_03: And you see some small Google apps on the top.
36:50
It looks nice,
36:50
but then you scroll down and it looks like an intern made that
36:53
page.
36:54
SPEAKER_03: Like you have some random Chinese app that you never heard about.
36:57
And then BMW,
36:59
which is a brand that you know,
37:00
it's somewhere in the very bottom.
37:02
SPEAKER_03: And I'm like...
37:03
SPEAKER_03: There's no apps,
37:04
SPEAKER_03: there's no big apps,
37:06
SPEAKER_03: there's no big logos outside of,
37:07
SPEAKER_03: and even for the Google logos,
37:10
SPEAKER_03: Flutter is not using any of their flagship apps.
37:13
SPEAKER_03: So I'm like,
37:13
SPEAKER_03: startups who are deciding which ones to use,
37:15
SPEAKER_03: just based on this,
37:16
SPEAKER_03: they will go for React Native.
37:17
SPEAKER_03: It actually has the street cred.
37:19
SPEAKER_03: And I asked someone at Meta, like, how did you guys pull this off?
37:22
SPEAKER_03: Like with a smaller team,
37:23
SPEAKER_03: you executed clearly what I think is better in terms of like,
37:27
SPEAKER_03: you got the big customers,
37:28
SPEAKER_03: you're building for building.
37:29
SPEAKER_03: He said, like, at Meta, everything is about impact.
37:31
SPEAKER_03: And the React Native team, the first thing they did is drive impact.
37:35
They got React Native inside the biggest apps into Instagram,
37:39
Facebook,
37:39
et cetera.
37:40
SPEAKER_03: And then the rest came because Shopify is like,
37:42
SPEAKER_03: well,
37:42
if it's used inside of Facebook with I don't know how many
37:45
thousands of developers,
37:46
SPEAKER_03: we can use it as well.
37:47
SPEAKER_01: Yeah, I mean...
37:49
SPEAKER_01: Look, Google can't afford to be disintermediated in the mobile space.
37:54
SPEAKER_01: They can't afford to just become the plumbing that people can swap out.
37:57
SPEAKER_01: And that's always been an existential threat for them.
38:02
The Facebook application is a platform itself,
38:04
and you can write applications inside of it.
38:06
SPEAKER_01: And so,
38:07
SPEAKER_01: like,
38:07
if you're writing for the Facebook platform and you're the New
38:09
York Times or
38:10
SPEAKER_01: whatever,
38:10
SPEAKER_01: like,
38:10
SPEAKER_01: who cares if you're running on Android or iOS?
38:12
SPEAKER_01: And that's Google's worst nightmare, right?
38:14
SPEAKER_00: Yeah.
38:14
SPEAKER_01: And that's why Facebook,
38:16
SPEAKER_01: in the age of AI,
38:16
SPEAKER_01: hasn't laid off the React Native team,
38:18
SPEAKER_01: because that's their basically,
38:20
SPEAKER_01: hey,
38:21
SPEAKER_01: you don't own Android.
38:22
SPEAKER_01: We do, right?
38:23
SPEAKER_01: That's their play.
38:24
SPEAKER_01: And so Google, they'll never give up on it.
38:27
SPEAKER_01: What happened was,
38:28
SPEAKER_01: unfortunately,
38:29
SPEAKER_01: Flutter's not from the Android team,
38:31
SPEAKER_01: and that pissed the Android team off,
38:32
SPEAKER_01: because Android is...
38:33
SPEAKER_03: Politics.
38:34
SPEAKER_01: Android was an acquisition.
38:37
SPEAKER_01: The guy that ran it was very particular about them being sort of
38:42
SPEAKER_01: In charge of their own destinies and not beholden to anyone else.
38:45
SPEAKER_01: And he kept Android sort of running the way that they ran it inside.
38:48
SPEAKER_01: And they made all the decisions and the buck stopped there.
38:51
Flutter came along and sort of threatened their dominance as
38:54
the platform.
38:56
SPEAKER_01: And it pissed him off.
38:56
And Google has been sort of unable to reconcile those things
39:00
even since 2018 when I
39:01
SPEAKER_01: was looking at this problem,
39:02
SPEAKER_01: 2017.
39:03
SPEAKER_03: Yeah,
39:03
and one of my biggest question marks about Google and why they
39:05
have not changed
39:06
SPEAKER_03: this is around their cloud platform.
39:09
So when I worked at Microsoft,
39:10
well,
39:10
I like to say Microsoft,
39:12
it was Skype.
39:12
SPEAKER_03: They just bought Skype and they left us alone.
39:14
SPEAKER_03: So it was Skype.
39:15
SPEAKER_03: And then when I turned Microsoft,
39:16
SPEAKER_03: I kind of,
39:17
SPEAKER_03: I was like,
39:17
SPEAKER_03: all right,
39:18
SPEAKER_03: this is,
39:19
SPEAKER_03: I don't like that much.
39:20
SPEAKER_03: But they gave us a mandate.
39:22
SPEAKER_03: They said, you need to use Azure.
39:25
SPEAKER_03: And we were one of the first, like we were the new purchase.
39:28
SPEAKER_03: So they just dumped it on us.
39:29
SPEAKER_03: Azure was not ready.
39:31
SPEAKER_03: And I was sitting next to the data team,
39:32
the Skype data team who had all our data centers and they're
39:35
moving over and
39:36
SPEAKER_03: they're saying,
39:36
SPEAKER_03: it's just a huge pain.
39:38
It's like,
39:38
we don't want to do this,
39:39
but it was actually Balmer was forcing it on them.
39:43
SPEAKER_03: And it was this blood, sweat, and tears, and eventually they moved.
39:46
SPEAKER_03: But what I've seen is,
39:47
SPEAKER_03: like,
39:47
SPEAKER_03: over time,
39:47
SPEAKER_03: you know,
39:48
SPEAKER_03: now when I talk with teams at Microsoft,
39:49
SPEAKER_03: like,
39:49
SPEAKER_03: what are you guys using?
39:50
SPEAKER_03: Obviously, they're using Azure.
39:52
SPEAKER_03: Or Bing might not be using it, but it's fun.
39:54
SPEAKER_03: AWS is using AWS.
39:56
SPEAKER_03: And I talk with teams at Google.
39:57
SPEAKER_03: What are you guys using?
39:58
SPEAKER_03: Borg.
39:59
SPEAKER_03: Like, hold on.
40:00
SPEAKER_03: Why are you not using GCP?
40:01
SPEAKER_03: Well, it doesn't scale.
40:03
SPEAKER_03: It doesn't have the things we need.
40:04
SPEAKER_03: And,
40:04
SPEAKER_03: like,
40:05
how can you be gunning to be number two or one day maybe number
40:09
one cloud platform
40:10
SPEAKER_03: if your own company...
40:12
SPEAKER_03: comes up with excuses.
40:13
SPEAKER_03: And I never understood,
40:15
SPEAKER_03: I tried to ask this,
40:16
SPEAKER_03: like,
40:16
SPEAKER_03: on back channels from people working at GCPN,
40:18
SPEAKER_03: they always come up with excuses,
40:19
SPEAKER_03: but I don't understand,
40:20
how is it that it's the only cloud company that does not use
40:23
its own cloud service
40:24
SPEAKER_03: for their flagship service,
40:26
SPEAKER_03: for the flagship products?
40:28
I think it's all just who's been the most successful at
40:31
marketing and convincing
40:32
SPEAKER_01: people that they're using their own clouds,
40:34
SPEAKER_01: but...
40:34
SPEAKER_01: They are all currently huffing their own farts.
40:38
SPEAKER_01: Amazon doesn't use AWS.
40:39
SPEAKER_01: No, I heard so.
40:41
SPEAKER_01: Sable ain't AWS.
40:43
SPEAKER_01: Okay.
40:44
SPEAKER_01: I mean, like, right, for the retail side, for the ad side.
40:47
SPEAKER_01: I mean,
40:47
SPEAKER_01: like,
40:47
SPEAKER_01: of course,
40:47
SPEAKER_01: they want you to use AWS,
40:49
SPEAKER_01: but all the core,
40:50
SPEAKER_01: the core,
40:50
SPEAKER_01: core,
40:50
SPEAKER_01: core stuff,
40:51
SPEAKER_01: and they haven't migrated,
40:52
SPEAKER_01: man.
40:52
SPEAKER_01: So, like, it's all frou-frou as far as I'm concerned, right?
40:55
SPEAKER_03: I think it might have changed because it is less.
40:59
They had a name for the old stuff,
41:01
and I think more and more things are moving over.
41:02
SPEAKER_01: That's fair.
41:03
SPEAKER_01: Never bet against Amazon.
41:05
AWS may have actually graduated to the point where they can
41:07
actually use it internally.
41:09
SPEAKER_01: The hurdles for Google were insurmountable.
41:13
SPEAKER_03: So maybe this is fair, by the way.
41:15
SPEAKER_03: So maybe this criticism is not entirely fair,
41:17
because what I understand is their infrastructure is way bigger
41:20
and more complex
41:21
SPEAKER_03: than anything else.
41:22
SPEAKER_01: It's sort of fair to say that Google's cloud runs on top of Google's
41:26
SPEAKER_01: infrastructure,
41:27
SPEAKER_01: which actually does scale the biggest in the world,
41:30
SPEAKER_03: much bigger than Amazon.
41:31
SPEAKER_03: Well,
41:32
SPEAKER_03: one thing that I am wondering,
41:34
because I'm still waiting for what will the tool or platform be
41:38
that Google
41:38
SPEAKER_03: releases that their internal tool teams use it?
41:41
SPEAKER_03: And they're like,
41:42
SPEAKER_03: oh,
41:42
SPEAKER_03: we have,
41:42
SPEAKER_03: you know,
41:42
100,000 or like 50,000 or 100,000 software developers inside
41:46
Google using it.
41:47
SPEAKER_03: You should use it.
41:48
SPEAKER_03: You know, Microsoft did this with like Visual Studio.
41:50
SPEAKER_01: You need some non-Googler?
41:52
SPEAKER_03: No, no, no.
41:52
SPEAKER_03: It's not some Google tool.
41:54
SPEAKER_03: And I'm thinking,
41:54
SPEAKER_03: could we see this maybe with some AI tools,
41:57
SPEAKER_03: you know,
41:57
SPEAKER_03: like AI coding tools,
41:58
SPEAKER_03: et cetera.
41:59
SPEAKER_03: Could they finally do this?
42:00
SPEAKER_03: Or maybe this is not a Google way to do it.
42:02
SPEAKER_03: They'll be like,
42:02
SPEAKER_03: all right,
42:03
SPEAKER_03: we have our superior internal tools and we will build an external thing.
42:06
SPEAKER_03: You know, we have Borg.
42:07
SPEAKER_03: We'll build Kubernetes for everyone else.
42:10
SPEAKER_01: I don't think Google understands developers.
42:13
SPEAKER_01: I don't think they ever did.
42:15
SPEAKER_03: Ironic.
42:16
SPEAKER_01: It's really closely related to their blind spot around platforms, right?
42:20
SPEAKER_01: If you don't get platforms, it's because you don't understand developers.
42:23
SPEAKER_03: It's just ironic because Google,
42:24
SPEAKER_03: like no company or few companies treat developers as good as Google does,
42:29
SPEAKER_03: right?
42:29
SPEAKER_03: In terms of hampering them.
42:30
SPEAKER_03: Yeah,
42:31
and few companies have built a platform as incredible
42:34
internally as Google's is,
42:37
SPEAKER_01: you know,
42:37
SPEAKER_01: at the sort of foundation level.
42:39
SPEAKER_03: Yeah.
42:40
SPEAKER_03: You unretired.
42:42
SPEAKER_03: You retired for some time and then you unretired because of,
42:46
SPEAKER_03: well,
42:46
SPEAKER_03: initially Sourcegraph,
42:47
SPEAKER_03: but then also AI.
42:49
SPEAKER_03: What made you kind of come back into the game?
42:53
SPEAKER_01: It's not a binary thing.
42:54
SPEAKER_01: I've been gradually unretiring, if it makes any sense.
42:58
And it's because at first I was like,
43:00
you know what,
43:01
I'm really climbing the walls.
43:02
SPEAKER_01: I really want to just go work with some people.
43:04
SPEAKER_01: And so that's, you know, that's where I went up at Sourcegraph.
43:06
SPEAKER_01: Like that was familiar ground, right?
43:08
SPEAKER_01: That was Google Code search for everyone else.
43:10
SPEAKER_01: Yeah, yeah.
43:11
SPEAKER_01: And then shortly afterwards, the AI showed up.
43:14
SPEAKER_01: And I was like,
43:15
SPEAKER_01: that was like the next step up is,
43:16
SPEAKER_01: oh,
43:16
SPEAKER_01: man,
43:17
maybe I better get back into coding again for a while because
43:19
this looks really
43:20
SPEAKER_01: different.
43:20
SPEAKER_03: So fun fact is last time we talked about three years ago,
43:24
SPEAKER_03: you were head of engineering at Sourcegraph.
43:26
SPEAKER_03: And actually people told me at Sourcegraph you came in,
43:28
SPEAKER_03: you made some changes,
43:29
SPEAKER_03: which were actually like pretty well received,
43:31
SPEAKER_03: but like you shook up,
43:32
SPEAKER_03: you introduced where people could drop there,
43:35
SPEAKER_03: that kind of stuff.
43:37
SPEAKER_03: And then next thing I know is like,
43:39
SPEAKER_03: oh,
43:39
SPEAKER_03: you wrote this,
43:40
SPEAKER_03: like you write about everything,
43:41
SPEAKER_03: which,
43:42
SPEAKER_03: you know,
43:43
SPEAKER_03: we'll link some more and more things,
43:44
SPEAKER_03: but I love writing it.
43:45
SPEAKER_03: But you wrote about like,
43:46
SPEAKER_03: oh,
43:47
I'm stepping down as head of engineering because I'm going back
43:49
to coding.
43:49
SPEAKER_03: Yeah.
43:49
SPEAKER_03: Which was not what I would have expected,
43:51
SPEAKER_03: again,
43:52
SPEAKER_03: from just...
43:53
SPEAKER_01: And I view that as another step in me sort of coming out of retirement,
43:58
SPEAKER_01: right?
43:58
SPEAKER_01: Because I had given up on coding.
44:00
SPEAKER_01: I just, it wasn't worth it anymore.
44:03
SPEAKER_01: Kent Beck had given up on coding.
44:04
SPEAKER_01: A lot of my old buddies and colleagues, right?
44:06
You know,
44:07
it's just like environment setup is just over the top these
44:09
days,
44:10
right?
44:10
SPEAKER_01: And,
44:10
SPEAKER_01: you know,
44:11
SPEAKER_01: just building a simple web app,
44:12
SPEAKER_01: you probably have to use,
44:13
SPEAKER_01: you know,
44:13
SPEAKER_01: 25 different frameworks,
44:14
SPEAKER_01: many of which have incompatible competing,
44:16
SPEAKER_01: you know,
44:17
SPEAKER_03: whatevers.
44:17
Yeah,
44:17
and as soon as you update to the latest React thing,
44:20
all the router is bracing.
44:21
SPEAKER_03: You have to relearn it.
44:22
SPEAKER_01: Who wants that?
44:22
And so at some point you get tired of it and you're just like,
44:25
I'm done,
44:25
man.
44:25
SPEAKER_01: I can't.
44:26
SPEAKER_01: This isn't, it's not worth it, right?
44:27
SPEAKER_01: Yeah.
44:28
SPEAKER_01: And AI completely turned that on its head.
44:29
SPEAKER_01: And I saw it coming as soon as ChatGPT came out.
44:32
I was like,
44:32
oh,
44:32
wow,
44:32
look,
44:33
it can write an actual function that's reasonably good,
44:36
right?
44:36
SPEAKER_01: And then when 4.0 came out,
44:38
SPEAKER_01: then I was able to project forward with exponential growth and say,
44:40
SPEAKER_01: uh-oh,
44:41
SPEAKER_01: uh-oh,
44:42
SPEAKER_01: you know,
44:43
SPEAKER_01: it's coming,
44:43
SPEAKER_01: right?
44:44
And so now I'm getting sort of like more and more fired up with
44:48
each passing month.
44:50
SPEAKER_03: This episode is brought to you by Sonar,
44:51
SPEAKER_03: the creators of SonarCube,
44:53
the industry standard for integrated code quality and code
44:56
security that is trusted
44:57
SPEAKER_03: by over 7 million developers at 400,000 organizations around the world.
45:02
SPEAKER_03: Human-written or AI-generated,
45:04
SPEAKER_03: SonarCube reviews all code and provides actionable code intelligence that
45:07
developers can use to improve quality and security early in the
45:10
development
45:11
SPEAKER_03: process.
45:13
SPEAKER_03: Sonar recently introduced SonarCube Advanced Security,
45:15
enhancing Sonar's core security capabilities with software
45:18
composition analysis and
45:19
SPEAKER_03: advanced static application security testing.
45:23
SPEAKER_03: With advanced security,
45:24
development teams can identify vulnerabilities in third-party
45:26
dependencies,
45:27
SPEAKER_03: ensure license compliance,
45:28
SPEAKER_03: and generate software bills of materials.
45:31
It's no wonder that SonarCube has been rated the best static
45:33
code analysis tool for
45:34
SPEAKER_03: five years running.
45:36
SPEAKER_03: So join millions of developers from organizations like Microsoft,
45:39
SPEAKER_03: NVIDIA,
45:40
SPEAKER_03: Mercedes-Benz,
45:40
SPEAKER_03: Johnson & Johnson,
45:41
SPEAKER_03: and eBay,
45:42
SPEAKER_03: and supercharge your developers to build better,
45:44
SPEAKER_03: faster with Sonar.
45:46
SPEAKER_03: Visit sonarsource.com slash pragmatic security to learn more.
45:51
one thing that has we've talked a lot in the industry and
45:54
everyone's talking about
45:56
it is how ai will first and foremost and like i think like
46:00
experienced developers
46:01
we can get there but how will it impact junior developers and
46:03
you wrote this again
46:05
controversial title the death of the junior developer but
46:09
interesting enough
46:10
SPEAKER_03: When you read closer, a lot of your articles are like this, by the way.
46:13
SPEAKER_03: Like there's a title which you think like, oh, it's the end.
46:16
SPEAKER_03: But a lot of them are wake up calls.
46:18
To me,
46:18
when I read it properly,
46:19
it wasn't like,
46:20
oh,
46:20
there's no more junior developers.
46:22
SPEAKER_03: It was a wake up call saying,
46:23
SPEAKER_03: hey,
46:24
SPEAKER_03: if you're a junior developer,
46:25
SPEAKER_03: you need to get your stuff together quickly.
46:29
And like whatever the junior developers before you were,
46:33
it's not going to work for you.
46:34
SPEAKER_03: So like, what is it that you've seen?
46:38
SPEAKER_03: And like, did something inspire you?
46:39
Did you like see some young titans who were actually just doing
46:43
great with these tools?
46:45
SPEAKER_01: Man,
46:45
you've hit on a question that is just so fundamental and
46:48
foundational to our
46:48
SPEAKER_01: industry right now.
46:49
SPEAKER_01: It's shaking the industry, that question, you know.
46:53
SPEAKER_01: And the answer is,
46:54
SPEAKER_01: I mean,
46:54
SPEAKER_01: the shortest way that I would think about it is AI is not easy to use.
47:00
SPEAKER_01: And the more senior you are,
47:02
SPEAKER_01: the more likely it is you're going to notice when it's being bad,
47:05
SPEAKER_01: when the AI is being naughty,
47:07
SPEAKER_01: right?
47:07
SPEAKER_01: It's just common sense.
47:08
SPEAKER_01: And the AI is very naughty and in very subtle and insidious ways.
47:13
SPEAKER_01: And even as they get smarter,
47:14
SPEAKER_01: and they are getting exponentially smarter,
47:16
SPEAKER_01: and they will be frighteningly smart within a year,
47:18
SPEAKER_01: they will still,
47:19
SPEAKER_01: I mean,
47:19
SPEAKER_01: software is always bigger than they are.
47:21
SPEAKER_01: And they will still do silly things.
47:24
SPEAKER_01: And it's just going to bias towards more senior people.
47:27
SPEAKER_01: But it's not really seniority we learned.
47:29
SPEAKER_01: It's nothing to do with junior and senior.
47:30
It's really more about who is demonstrating the ability to work
47:33
well with AI and
47:35
SPEAKER_01: get good outcomes in software.
47:36
SPEAKER_01: And that could be anyone.
47:37
SPEAKER_01: It could be a product manager.
47:39
So I think there's a big shakeup coming where the roles change
47:42
and everybody
47:42
SPEAKER_01: becomes more focused on what they're building instead of,
47:45
SPEAKER_01: like,
47:46
SPEAKER_01: who's building it.
47:47
SPEAKER_01: Yeah.
47:48
SPEAKER_01: Have you heard about the collapsing the stack stuff from Scott Belsky?
47:52
SPEAKER_03: Can you refresh to know?
47:54
SPEAKER_01: Basically,
47:54
SPEAKER_01: like,
47:55
SPEAKER_01: there's a line of thinking that we've over-specialized and everybody's,
47:58
SPEAKER_01: like,
47:58
SPEAKER_01: incredibly domain expertise specialized.
48:00
And you've got these senior engineers at Google who know
48:02
exactly how the Fuse file
48:03
SPEAKER_01: system drivers work for every version of Linux kernel,
48:06
SPEAKER_01: right?
48:06
SPEAKER_01: Yeah.
48:07
SPEAKER_01: It's like, we don't have that anymore.
48:08
SPEAKER_01: We don't need that.
48:09
SPEAKER_01: That's stupid.
48:09
SPEAKER_01: That's going away.
48:10
But all the specialization is going away because AI is
48:13
democratizing all of it.
48:15
SPEAKER_01: You can't hide that knowledge anymore.
48:16
SPEAKER_03: This is interesting because I just talked with someone,
48:19
SPEAKER_03: I think a week or two ago,
48:21
about what has changed in software engineering even before AI
48:24
and what has changed
48:25
SPEAKER_03: back in early 2000s when you looked at software developers.
48:28
SPEAKER_03: You had the Java developer.
48:30
SPEAKER_03: You had the .NET developer.
48:31
SPEAKER_03: You had the Python developer.
48:33
SPEAKER_03: And these were different people.
48:34
It was the Java developer would not do .NET,
48:37
even though they're pretty similar.
48:38
SPEAKER_03: So on the back end, languages were specialized.
48:43
Fast forward to even before AI,
48:44
like 2015 or 2018,
48:45
when we had a big hiring for.
48:47
SPEAKER_03: Like when I worked at Uber,
48:49
SPEAKER_03: we no longer like Uber was seen as like,
48:51
SPEAKER_03: oh,
48:51
SPEAKER_03: this like completely changed.
48:52
SPEAKER_03: We didn't care if you did Java or .NET or whatever.
48:56
SPEAKER_03: You come, you know, one of those languages, you'll pick up.
48:59
SPEAKER_03: whatever we're doing.
49:00
SPEAKER_03: And at some point, my team was doing Go, Python, Node.js, and what else?
49:05
SPEAKER_03: We're still doing something else, but we're doing all of it.
49:08
SPEAKER_03: So we started to have less specializations.
49:11
SPEAKER_03: So I wonder if this thing started earlier and maybe AI actually just...
49:15
SPEAKER_03: makes it more viable that now,
49:18
SPEAKER_03: until now,
49:18
SPEAKER_03: we've had,
49:19
SPEAKER_03: you know,
49:19
SPEAKER_03: a front-end engineer would not touch back.
49:20
SPEAKER_03: And they might understand the concept of APIs, but now they actually can.
49:24
SPEAKER_03: In fact, when the product manager can actually create pull requests.
49:27
SPEAKER_03: Yeah, we see that now, right?
49:29
SPEAKER_01: Like at SourceGraph,
49:30
one of our UI designers is now sending pull requests for the UI
49:33
instead of asking
49:34
SPEAKER_01: engineers to do it.
49:34
SPEAKER_03: And are they any good?
49:37
SPEAKER_03: They're actually decent?
49:37
SPEAKER_03: Yeah.
49:39
SPEAKER_01: Yeah.
49:40
SPEAKER_01: Look, I mean, it's all over the map.
49:41
It's just like I believe this is the new role for junior
49:44
developers is they're
49:45
going to be mentoring the next layer down of non-technical or
49:49
technical adjacent
49:49
SPEAKER_01: people who are now starting to contribute PRs.
49:52
SPEAKER_01: Right.
49:52
And they'll be the ones who are like helping them fix the
49:55
security issues or
49:56
SPEAKER_01: whatever else they have with their.
49:58
Basically teaching junior developers because they're still
50:00
trained engineers,
50:01
right?
50:01
So they can teach like a UX designer or product manager what
50:05
are the right
50:05
questions to ask the AI about your thing to know whether you're
50:08
done or not yet,
50:09
SPEAKER_01: right?
50:09
SPEAKER_01: You know, give me those kinds of skills.
50:12
SPEAKER_03: I like this because I think we all know things will change.
50:16
SPEAKER_03: I think we're all struggling to like put the finger exactly.
50:19
SPEAKER_03: I mean,
50:19
SPEAKER_03: you clearly have a bunch of conviction,
50:21
which I think is great because I think you need conviction in
50:24
this area of going
50:25
SPEAKER_03: around them.
50:26
SPEAKER_03: And I have been.
50:27
SPEAKER_03: So, you know, you work at Sourcegraph.
50:29
SPEAKER_03: You guys are heavily using your AI.
50:32
In fact,
50:32
you have your own AI tool,
50:33
but you've been using the existing tools from the beginning.
50:37
And most of the stories I'm hearing so far about a
50:40
non-technical person doing
50:42
technical stuff is at AI companies where they're surrounded by
50:45
these people.
50:46
SPEAKER_03: Winster co-founder and CEO Varun, he told me that they have a...
50:51
SPEAKER_03: I think it was a salesperson who had a sales tool,
50:54
SPEAKER_03: and they just kind of vibe-coded it with Windsurf.
50:57
SPEAKER_03: It had no state.
50:58
SPEAKER_03: It was a super simple thing.
50:59
SPEAKER_03: You know, it's not complicated.
51:01
SPEAKER_03: But that person did it.
51:02
SPEAKER_03: And I wonder if,
51:03
SPEAKER_03: you know,
51:03
SPEAKER_03: we might be seeing these type of,
51:06
SPEAKER_03: like,
51:06
SPEAKER_03: kind of AI or just very startup environments,
51:08
SPEAKER_03: like,
51:09
lead the path of what the kind of legacy or larger companies
51:13
will be in 10 years.
51:14
SPEAKER_03: Yeah.
51:15
SPEAKER_01: Absolutely, man.
51:15
SPEAKER_01: We're seeing it everywhere.
51:17
SPEAKER_01: I mean,
51:17
SPEAKER_01: we're seeing companies where marketing teams are writing their own,
51:20
SPEAKER_01: you know,
51:21
SPEAKER_01: outbound,
51:22
SPEAKER_01: you know,
51:22
SPEAKER_01: campaign software.
51:23
SPEAKER_01: You know, we're seeing product teams, you know, bypass vendors.
51:27
They don't have to re-up with their renewal or contracts with
51:29
some crummy vendor
51:30
software because they wrote their own and had somebody from
51:33
engineering just vet it
51:34
SPEAKER_01: and be like,
51:34
SPEAKER_01: okay,
51:34
SPEAKER_01: yeah,
51:35
SPEAKER_01: you can make these two changes and then you can—
51:37
I want to ask you a little bit about that,
51:38
because I'm a bit skeptical about that.
51:41
SPEAKER_03: Have you seen specific examples of what they replace?
51:43
SPEAKER_03: Because,
51:44
SPEAKER_03: for example,
51:44
SPEAKER_03: with Workday,
51:45
SPEAKER_03: you're not going to replace that,
51:46
SPEAKER_03: which has all the compliance,
51:48
SPEAKER_03: a lot of state,
51:49
SPEAKER_03: a lot of regulation,
51:50
SPEAKER_03: a lot of ongoing maintenance.
51:52
That is not what you're going to...
51:54
But what are the things that you've seen?
51:56
SPEAKER_01: Well,
51:56
this was a – so imagine a company with a lot of really bad
51:59
actors coming in and
52:00
trying to crawl over the site and find fan-bad ways to
52:03
basically siphon money out.
52:05
SPEAKER_01: So they have many,
52:06
many different kinds of teams that are looking at different
52:08
kinds of fraud and
52:09
SPEAKER_01: different kinds of attacks.
52:10
SPEAKER_01: And there are lots of kind of bespoke tools.
52:12
And so you get into this long tail of little vendors that offer
52:15
these crummy tools
52:16
SPEAKER_01: that are really expensive.
52:18
SPEAKER_01: Very vertical, domain-specific.
52:20
SPEAKER_01: And so the product team at this one company was like, screw it.
52:22
SPEAKER_01: We're going to ask AI to build it.
52:24
SPEAKER_01: We'll give it the specs.
52:25
SPEAKER_01: We know what it wanted to do.
52:26
SPEAKER_01: And they built it in Python.
52:28
SPEAKER_01: And so it wasn't production software.
52:30
It was software that they use as their investigation trying to
52:33
find bad actors.
52:34
Nevertheless,
52:35
it saved them from a re-up with a contract that was rather
52:38
expensive.
52:38
SPEAKER_01: And it gave them – they were happy, right?
52:41
SPEAKER_01: They had full control over the software.
52:42
SPEAKER_01: They could make it do whatever they wanted at that point.
52:44
SPEAKER_01: So we're starting,
52:45
SPEAKER_01: and this is just one of probably a dozen examples I could give you,
52:48
SPEAKER_01: but we're seeing it.
52:48
SPEAKER_03: But let's carry on that thought, because you and me, we built software.
52:52
SPEAKER_03: We've seen the internal tool that the team has built.
52:54
SPEAKER_03: In fact, Google is famous for building all the internal tools.
52:57
SPEAKER_03: What is the next step, though?
52:59
SPEAKER_03: Can we just move?
53:00
SPEAKER_03: Because we know what's going to happen, right, as experienced users?
53:03
SPEAKER_03: people who are working software.
53:04
So what's happening next year when there's more functionality
53:07
to be added?
53:08
SPEAKER_03: How far might we be able to take it?
53:10
SPEAKER_03: And what's going to be the breaking point?
53:12
SPEAKER_03: Because this happened before AI, right?
53:14
Like one internal developer wrote it,
53:16
and at some point it becomes just a pain,
53:19
right?
53:19
SPEAKER_01: Yeah.
53:19
SPEAKER_01: Look,
53:20
I predict I'm going to go around the record and I'm going to
53:22
predict that there is
53:23
SPEAKER_01: a new role,
53:24
a new category of roles that's going to emerge that are the
53:27
Winston Wolfe that are
53:28
SPEAKER_01: going to come in and fix shit that you broke with AI.
53:31
They're going to be fixers and they're going to come in and
53:33
they're going to be small and large.
53:35
SPEAKER_03: I should call them.
53:36
SPEAKER_03: Let's give us role name.
53:38
SPEAKER_03: Call it Fixers.
53:39
SPEAKER_03: It's a cool name.
53:40
SPEAKER_01: I don't know.
53:41
SPEAKER_01: Fixers sounds pretty good.
53:41
SPEAKER_01: But whatever, right?
53:43
I do think of them as Fixers in the sense that,
53:45
like,
53:45
you've made a horrible mess.
53:47
SPEAKER_01: You've realized that this company that promised the world to you,
53:50
SPEAKER_01: because,
53:50
SPEAKER_01: like,
53:50
something like 60% of all the world's programmers are systems
53:53
integrators.
53:53
SPEAKER_01: They go to big companies that are desperate,
53:55
SPEAKER_01: and they say,
53:55
SPEAKER_01: we can make your systems talk to each other,
53:56
SPEAKER_01: and it'll be really expensive.
53:57
SPEAKER_01: And 70% of the projects fail, but companies go for it anyway.
54:01
SPEAKER_01: That whole economy of...
54:03
Rich countries sending work to poor countries,
54:05
the architects and all that.
54:06
SPEAKER_01: That's all getting turned on,
54:07
potentially turned on its head because we don't know who's
54:09
going to be doing the
54:09
SPEAKER_01: work now,
54:10
SPEAKER_01: the actual implementation,
54:11
SPEAKER_01: right?
54:12
SPEAKER_01: Is it the rich countries that are going to do it for themselves now?
54:14
SPEAKER_01: A lot of economists are looking at this problem right now, right?
54:18
SPEAKER_03: Yeah, but we've seen this with outsourcing.
54:20
SPEAKER_03: Don't forget, like the whole idea with outsourcing.
54:23
SPEAKER_03: From the 90s,
54:24
SPEAKER_03: I kept hearing like,
54:25
SPEAKER_03: oh,
54:26
all the highly paid developer jobs will go to India or Asia
54:29
because it's cheaper.
54:30
SPEAKER_03: And then it happened but also didn't happen, right?
54:32
SPEAKER_01: Yeah.
54:33
SPEAKER_01: That's how,
54:33
SPEAKER_01: I mean,
54:34
SPEAKER_01: look,
54:34
SPEAKER_01: I think,
54:34
SPEAKER_01: look,
54:34
it's going to ultimately be cheaper if a human being needs to
54:37
babysit 10 AIs to get
54:38
SPEAKER_01: a project done.
54:39
It's going to be cheaper to have that human being be in Vietnam
54:42
than in,
54:43
you know,
54:43
in the UK.
54:44
SPEAKER_03: But the reason we have developers sit next to the business,
54:47
SPEAKER_03: because we're sitting next to it,
54:48
SPEAKER_03: you can actually talk to them.
54:50
SPEAKER_03: And that communication, I've seen this, you must have seen this a lot.
54:54
So when I was at Uber,
54:56
we do this round robin and Uber still does it to this day.
54:58
SPEAKER_03: It's like HQ is there in San Francisco and it's very expensive to hire.
55:03
SPEAKER_03: Amsterdam is half the price and India is one third of the price.
55:06
SPEAKER_03: So there's this round robin of like,
55:07
SPEAKER_03: okay,
55:07
SPEAKER_03: let's hire people in the US and like,
55:10
SPEAKER_03: oh,
55:10
SPEAKER_03: it's expensive.
55:11
SPEAKER_03: Let's hire in India.
55:12
SPEAKER_03: OK, we hire for a while.
55:13
SPEAKER_03: Well, it turns out you can get like less experienced people.
55:16
SPEAKER_03: There's communication issues.
55:17
SPEAKER_03: It's kind of breaking down.
55:18
SPEAKER_03: Let's now hire an Amsterdam because it's closer.
55:20
SPEAKER_03: It's kind of midway.
55:21
SPEAKER_03: And then it comes back at somebody.
55:23
SPEAKER_03: Let's hire.
55:23
SPEAKER_03: And it just like every few years it goes to the next one.
55:26
SPEAKER_03: And it's just a repeat.
55:28
SPEAKER_03: Like when they were cutting Amsterdam and now they're actually hiring it.
55:32
SPEAKER_03: I'm like, right.
55:33
SPEAKER_03: Yeah.
55:33
SPEAKER_01: It's been like four years.
55:35
Outsourcing is one of those classic expansion contraction
55:38
cycles that a lot of
55:38
companies just go through periodically along with centralizing
55:41
and decentralizing
55:42
SPEAKER_01: QA or centralizing,
55:43
SPEAKER_01: decentralizing,
55:45
SPEAKER_01: you know,
55:46
SPEAKER_01: TPMs or whatever.
55:47
SPEAKER_01: Like they just like they'll try both and the grass is always greener.
55:51
SPEAKER_01: They can never make up their lines, you know.
55:53
SPEAKER_03: So your new book is – the title is Vibe Coding and –
55:58
It's a heated debate if you should even call it vibe coding
56:01
because of the definition.
56:02
SPEAKER_03: So let's start with what do you define as vibe coding?
56:05
SPEAKER_01: Vibe coding is when the AI writes the code.
56:08
SPEAKER_01: All right.
56:08
SPEAKER_01: There's a reason that that definition is going to win.
56:11
SPEAKER_01: You can't put an if clause in a slogan.
56:14
SPEAKER_01: Use vibe coding as long as you're doing a fine print,
56:18
which is what they're trying to do is they're trying to put a
56:21
condition on it.
56:22
SPEAKER_03: I agree, by the way.
56:23
SPEAKER_03: You can't do that.
56:24
SPEAKER_01: It's out of the bag.
56:26
SPEAKER_03: That's how I've heard people use it as well.
56:28
SPEAKER_03: It's like, you know, like some people use it for prototype.
56:31
SPEAKER_03: The point is like, yeah, you're kind of like, I'm in this vibe.
56:33
SPEAKER_03: I'm telling, I'm letting it go.
56:35
It often is an Asian mode,
56:36
you know,
56:37
where it kind of goes and does stuff.
56:39
SPEAKER_03: But it might also be, I might kind of rein it in, but it's just like...
56:44
SPEAKER_03: You know,
56:44
SPEAKER_03: like vibing,
56:45
SPEAKER_03: like,
56:46
SPEAKER_03: so I think this stuff,
56:47
I think because a lot of people are pointing to like the Andre
56:50
Carpenter's like
56:51
SPEAKER_03: tweet or however he defined it.
56:53
SPEAKER_03: And yeah, I think it'll just come into like whatever.
56:56
SPEAKER_01: The question is, is it giving you a buzz?
56:58
SPEAKER_01: Like for real?
56:59
SPEAKER_01: Because programming can give you a buzz when you get into flow, right?
57:01
SPEAKER_01: You can get an actual buzz going.
57:03
SPEAKER_01: And you know what?
57:04
SPEAKER_01: It is insanely addictive.
57:06
SPEAKER_01: Cloud Code and Friends,
57:07
SPEAKER_01: Sourcegraph,
57:08
SPEAKER_01: AMP,
57:08
SPEAKER_01: you know,
57:08
SPEAKER_01: try them out because,
57:10
SPEAKER_01: wow,
57:11
SPEAKER_01: they're like a dopamine hit.
57:12
SPEAKER_01: It's like a slot machine.
57:14
SPEAKER_01: They're literally addictive.
57:15
I mean,
57:16
Ken Beck told me the same thing,
57:17
and I've experienced the same thing.
57:18
SPEAKER_03: Like, I have this side project, which I just don't like to touch.
57:21
So I try to build my APIs on the side and not pay vendors when
57:25
I can,
57:25
but it's just a hassle.
57:26
SPEAKER_03: It's somewhere on AWS, and it's a hassle to, like, remember how I deploy.
57:31
SPEAKER_03: But...
57:33
SPEAKER_03: With Windsurf,
57:34
I just built a small API on how people can claim perplexity and
57:39
CAGI codes if
57:40
SPEAKER_03: they're paid subscribers to the newsletter.
57:42
SPEAKER_03: And I connected with an MCP server.
57:44
SPEAKER_03: I connected my database so I can just talk to my database.
57:46
SPEAKER_03: And I asked them, oh, how many people have requested codes?
57:49
SPEAKER_03: And they're like, oh, today, there's the last 10 days.
57:52
SPEAKER_03: Oh, nine days ago, there were 20, 30, 1,000, 2,000, 3,000.
57:56
SPEAKER_03: I'm like, hold on, what is going on?
57:58
SPEAKER_03: That doesn't look normal.
57:59
SPEAKER_03: I asked, can you analyze the patterns
58:02
unusual patterns and then it figured it told me how you know
58:05
like there's the same
58:06
email with different cases and i needed to code a fix for this
58:10
but i was about to
58:11
have dinner and usually like if i don't have like 30 minutes to
58:13
code or an hour it
58:15
SPEAKER_03: doesn't make sense i had like 10 minutes
58:18
SPEAKER_03: And in that 10 minutes, I got, like, a fix done.
58:21
SPEAKER_03: I went and had dinner.
58:22
SPEAKER_03: I actually was, you know, present on a dinner.
58:24
SPEAKER_03: And I came back and I got back into it.
58:26
SPEAKER_03: And in a total of 30 minutes,
58:27
SPEAKER_03: I did stuff that would have taken me,
58:29
SPEAKER_03: like,
58:30
SPEAKER_03: even if I had the hands on,
58:31
SPEAKER_03: like,
58:31
SPEAKER_03: two hours easily.
58:33
And I felt like,
58:34
hold on,
58:34
I'm no longer worried about,
58:35
like,
58:36
falling out of the flow.
58:38
So,
58:38
like,
58:38
there is a lot of new stuff that it does make you more
58:42
productive.
58:42
SPEAKER_03: And, you know, as an experienced developer, like, it's amazing.
58:45
SPEAKER_03: Yeah.
58:46
Now I understand why Ken Beck is saying in 52 years he's never
58:50
felt this good about
58:52
SPEAKER_03: or this excited about writing code.
58:54
A lot of your listeners listening to us right now have no idea
58:57
what you're talking about.
58:59
Because they haven't actually tried the terminal app versions
59:02
of these things,
59:03
SPEAKER_01: like SourceGraph AMP and Cloud Code and Codex from OpenAI or Klein,
59:08
SPEAKER_01: right?
59:10
SPEAKER_01: And by the way,
59:10
SPEAKER_01: Klein is going to start taking on real,
59:12
SPEAKER_01: real,
59:12
SPEAKER_01: real importance,
59:13
being able to run local models as soon as local models reach
59:16
where Cloud Sonnet is
59:17
SPEAKER_01: today.
59:17
SPEAKER_01: Because Cloud Sonnet is very viable if you keep it on the rails.
59:21
SPEAKER_01: Because look, let's face it.
59:22
SPEAKER_01: The reason people are screwing this up and saying,
59:23
SPEAKER_01: this doesn't work,
59:24
SPEAKER_01: and I don't understand why AI works,
59:26
SPEAKER_01: and all these stories are BS,
59:27
SPEAKER_01: is
59:27
It's because it's very difficult to wrap your head around the
59:30
fact that you can't
59:31
SPEAKER_01: get an answer out of the AI.
59:32
SPEAKER_01: All you can do is converge on an answer together with it.
59:36
SPEAKER_01: Even if it's an agent running off and doing things,
59:38
you're still doing it together and you're going to eventually
59:40
converge on the right
59:41
SPEAKER_01: answer,
59:41
SPEAKER_01: hopefully,
59:42
SPEAKER_01: most of the time.
59:43
SPEAKER_01: Sometimes you have to go try a different model, right?
59:45
And you will very quickly learn the limits of their sort of
59:47
cognitive ability and
59:48
SPEAKER_01: that will be the constraints that you have to work within.
59:51
SPEAKER_01: And it's not easy, man.
59:52
SPEAKER_01: It's not easy.
59:52
SPEAKER_01: People expect it to be easy.
59:54
SPEAKER_01: They want it to be handed to them.
59:55
SPEAKER_03: Well,
59:55
SPEAKER_03: and also people,
59:56
SPEAKER_03: I think there is this,
59:58
SPEAKER_03: I'm trying to put a finger on it,
59:59
SPEAKER_03: but like the first time I used ChatGPT,
1:00:01
SPEAKER_03: it was magic.
1:00:02
SPEAKER_03: It was like, you know, mind blown.
1:00:04
SPEAKER_03: I think most of listeners have had this experience.
1:00:08
The first time I connected my MCP server,
1:00:11
my database,
1:00:12
in my case,
1:00:13
it was Windsurf.
1:00:14
SPEAKER_03: It could have been Cursor.
1:00:15
SPEAKER_03: It could have been anything else.
1:00:17
SPEAKER_03: And I solved something with, you know, the agent.
1:00:20
SPEAKER_03: I kind of guided it, but I...
1:00:22
SPEAKER_03: I was just a bit lazy,
1:00:23
SPEAKER_03: and I knew what I wanted to do,
1:00:24
SPEAKER_03: and I kind of stopped it and got it done,
1:00:25
SPEAKER_03: and it got it done so much faster.
1:00:27
SPEAKER_03: It was magic.
1:00:28
SPEAKER_03: But I have a feeling that with ChatGP, the magic faded after a while.
1:00:34
SPEAKER_03: It was magic initially, but then it's work, and I think somehow we...
1:00:38
A lot of people either get disappointed after the magic doesn't
1:00:41
continue.
1:00:42
And my most surprising conversation was with Simon Willison,
1:00:45
who has been the creator of Django.
1:00:47
SPEAKER_03: He is a super productive developer.
1:00:50
SPEAKER_03: He writes so much code written for AI.
1:00:53
SPEAKER_03: And he told me that this thing is hard.
1:00:55
SPEAKER_03: And in two and a half years of nonstop using it, he keeps learning.
1:00:58
SPEAKER_03: And to me, there's this contradiction.
1:01:01
SPEAKER_03: It feels so easy, but it needs so much work.
1:01:05
SPEAKER_03: What is going on?
1:01:05
SPEAKER_01: Yeah, that is a really weird contradiction, isn't it?
1:01:08
SPEAKER_01: It feels like it's making your life incredibly easier,
1:01:11
SPEAKER_01: and yet it's very,
1:01:12
SPEAKER_01: very non-trivial to keep the thing on the rails.
1:01:14
SPEAKER_01: It's like a toddler with a chainsaw, right?
1:01:17
SPEAKER_01: Like, seriously.
1:01:18
SPEAKER_01: Okay, let me tell you why.
1:01:18
SPEAKER_01: I'll tell you one reason.
1:01:19
SPEAKER_01: This is from Jason Clinton.
1:01:21
SPEAKER_01: He's the CISO at Anthropic, and he was kind enough to share with us.
1:01:24
After I whined at Gene Kim's engineering leadership conference
1:01:28
a few weeks back,
1:01:29
SPEAKER_01: I whined that Claude had deleted all my tests and said,
1:01:31
SPEAKER_01: your tests are all passing now.
1:01:33
SPEAKER_01: Which is true.
1:01:33
SPEAKER_01: They had passed away.
1:01:35
SPEAKER_01: Like, they were gone, dead.
1:01:36
SPEAKER_01: It deleted it?
1:01:37
SPEAKER_01: It deleted them.
1:01:38
SPEAKER_01: And it's like, all tests pass now.
1:01:39
SPEAKER_01: And it's like, well, goddammit, right?
1:01:41
And so Jason told us,
1:01:43
well,
1:01:44
what happened was Claude was trained on a reward function.
1:01:48
SPEAKER_01: And it wasn't trained not to hack that reward function.
1:01:51
SPEAKER_01: And so it will cheerfully hack it.
1:01:52
SPEAKER_01: And so that's the state of the art today is it will tell you it's done.
1:01:56
SPEAKER_01: And what you have to do is say, no, you're not.
1:01:58
SPEAKER_01: And send it back to the drawing board.
1:02:00
SPEAKER_03: Ken Beck was literally saying the same thing.
1:02:01
He calls it a genie,
1:02:02
which is it grants your wish,
1:02:04
but sometimes in unexpected ways.
1:02:06
SPEAKER_01: Exactly.
1:02:06
SPEAKER_01: It's a monkey's paw sometimes, right?
1:02:08
SPEAKER_01: Yeah.
1:02:09
SPEAKER_01: You got to be really careful how you phrase things.
1:02:10
SPEAKER_01: You know how you know the moment you know you're a modern programmer?
1:02:13
When you come down and sit down in front of your computer one
1:02:15
day and realize you
1:02:16
don't have any instances of any IDEs open and you're writing
1:02:19
more code than you
1:02:19
SPEAKER_01: ever have in your life.
1:02:21
SPEAKER_01: So everybody listening in,
1:02:22
SPEAKER_01: if you've got an IDE open and you're looking at source code,
1:02:24
SPEAKER_01: you're doing it wrong.
1:02:26
SPEAKER_01: Isn't that funny?
1:02:27
SPEAKER_01: Man, people are going to be freaked out about this.
1:02:29
SPEAKER_03: So until AI really took off,
1:02:33
SPEAKER_03: AI coding tools,
1:02:34
one of the hottest topics that I discussed and I think was in
1:02:37
everyone's mind is
1:02:37
developer productivity and specifically the question of whether
1:02:40
should we measure
1:02:41
PRs per developer or not,
1:02:43
because at Uber they were doing it and it was helpful in some
1:02:45
ways.
1:02:46
But I recently talked with a startup who is doing a developer
1:02:49
productivity tool.
1:02:50
SPEAKER_03: They're launching a new startup and I told them,
1:02:52
SPEAKER_03: they're like,
1:02:53
SPEAKER_03: oh,
1:02:53
SPEAKER_03: we're thinking of measuring PRs or not measuring it.
1:02:55
SPEAKER_03: I'm like, hold on.
1:02:57
SPEAKER_03: I think you're doing this wrong.
1:02:58
SPEAKER_03: Like,
1:02:58
SPEAKER_03: if we're looking ahead,
1:03:00
SPEAKER_03: like,
1:03:00
SPEAKER_03: the question is not,
1:03:01
SPEAKER_03: like,
1:03:01
SPEAKER_03: if developers are doing Hamlet and PRs,
1:03:03
SPEAKER_03: like,
1:03:04
SPEAKER_03: you will be able to do however many you want.
1:03:06
SPEAKER_03: But we need to think about, like, what will productivity look like?
1:03:10
SPEAKER_03: Because now,
1:03:11
SPEAKER_03: looking at the output of,
1:03:12
SPEAKER_03: like,
1:03:12
SPEAKER_03: how much code doesn't tell me anything,
1:03:14
SPEAKER_03: what would tell me something is if I sat next to someone,
1:03:17
SPEAKER_03: for example,
1:03:18
SPEAKER_03: are they actually reviewing the code before it goes into the code base?
1:03:21
SPEAKER_03: Are they challenging...
1:03:23
this the ai instead of just buying the lgtm you know looks good
1:03:27
to me and sending
1:03:28
it back and i'm not sure how like you know this is this is
1:03:31
going a little bit to
1:03:32
engineering leadership but there is going to be this big
1:03:33
question of like what does
1:03:35
actually i'm going to ask you this like fast forward to two
1:03:37
years let's assume
1:03:38
these tools evolve or you know they will not be worse but they
1:03:41
will be better what
1:03:42
do you think a really productive software engineer will look
1:03:45
like in terms of what
1:03:46
SPEAKER_03: they do not what they measure just what they do
1:03:48
SPEAKER_01: Yeah.
1:03:49
SPEAKER_01: First of all, I got to share Kent Beck's toboggan analogy.
1:03:52
He's like,
1:03:53
using these agents is like being on a sled going down like a
1:03:56
ski slope.
1:03:57
SPEAKER_01: You're going really fast.
1:03:59
SPEAKER_01: You're not really in control.
1:04:01
SPEAKER_01: You can steer it.
1:04:04
SPEAKER_01: And unfortunately, that is the state of the art right now.
1:04:06
SPEAKER_01: That's what software engineers who are embracing this.
1:04:08
SPEAKER_01: And they're spending thousands of dollars a week, right?
1:04:10
SPEAKER_01: Yeah.
1:04:10
SPEAKER_01: Which is why clients are going to become so important,
1:04:12
SPEAKER_01: why local inferencing is going to be so important.
1:04:14
The only way for VibeCoding to become truly sustainable is for
1:04:17
it to be local.
1:04:18
SPEAKER_03: Hold on.
1:04:18
SPEAKER_03: I'm going to stop you there.
1:04:19
SPEAKER_03: You're saying they're spending $1,000 a month.
1:04:21
SPEAKER_03: A week.
1:04:22
SPEAKER_03: A week.
1:04:23
SPEAKER_03: Who's spending it?
1:04:24
SPEAKER_03: Because now what I'm reading is like,
1:04:26
SPEAKER_03: oh,
1:04:26
we're not really going too much over with like the $100 Cloud
1:04:30
Pro subscription.
1:04:31
SPEAKER_03: Yeah.
1:04:31
I personally get a bill from Anthropic for $220 about every day
1:04:35
and a half or two days.
1:04:36
SPEAKER_01: It's absolutely insane.
1:04:37
SPEAKER_03: I'm desperate for local inference.
1:04:39
SPEAKER_03: As a professional developer.
1:04:40
SPEAKER_03: Yeah.
1:04:41
SPEAKER_03: And you're seeing this with, like, teams that you're working with.
1:04:44
SPEAKER_03: Like, you have some insight into a lot of other engineering teams, right?
1:04:47
SPEAKER_01: Well, yeah.
1:04:48
SPEAKER_01: We have people using AMP.
1:04:49
SPEAKER_01: We know how many tokens they're spending.
1:04:50
SPEAKER_03: Mm-hmm.
1:04:51
SPEAKER_01: They're token pigs, man, these agents.
1:04:53
SPEAKER_01: They solve problems.
1:04:54
All the problems you've ever heard about with AI,
1:04:57
they solve by just brute forcing it.
1:04:59
SPEAKER_01: Oh, I hallucinated something.
1:05:01
SPEAKER_01: Let me fix it.
1:05:02
SPEAKER_01: Oh, that was a hallucination too?
1:05:03
SPEAKER_01: I'll fix it again.
1:05:04
SPEAKER_01: And they keep going until they get it right at your expense.
1:05:07
SPEAKER_01: But it's still way faster than you could have done.
1:05:10
SPEAKER_01: So you can't not program this way.
1:05:11
SPEAKER_03: But this thousand of dollars,
1:05:13
SPEAKER_03: are vendors swallowing it or companies are actually being built for this?
1:05:17
SPEAKER_03: Publicly, I haven't heard too much chatter about this.
1:05:20
Maybe it's because it's mostly indie devs sharing on social
1:05:23
media and corporate
1:05:24
SPEAKER_03: devs,
1:05:24
SPEAKER_03: they might not just care.
1:05:25
SPEAKER_01: No corporate devs.
1:05:26
SPEAKER_01: Look, you know who's using these coding agents right now in corporations?
1:05:30
SPEAKER_01: The CTOs.
1:05:32
For some reason,
1:05:32
we've noticed a pattern where the CTOs are all the ones who
1:05:35
kind of get it.
1:05:36
SPEAKER_01: Right.
1:05:36
SPEAKER_01: The global network of CTOs, they get it.
1:05:39
SPEAKER_01: They understand what's happening and they understand the terrible,
1:05:41
SPEAKER_01: terrible economic tradeoff they're going to face,
1:05:43
which is how many engineers do you fire in order to pay for the
1:05:46
rest of them to
1:05:46
SPEAKER_01: have AI?
1:05:47
SPEAKER_01: Because it's very, very, very expensive right now.
1:05:50
SPEAKER_01: This is why I keep bringing up Klein and local inferencing,
1:05:52
because you're going to find real fast that as soon as you
1:05:54
start running four
1:05:54
SPEAKER_01: agents,
1:05:55
SPEAKER_01: you will feel like Poseidon,
1:05:57
SPEAKER_01: like navigating the seas.
1:05:58
SPEAKER_01: Right.
1:05:59
SPEAKER_01: You'll feel like a deity.
1:06:00
SPEAKER_01: Right.
1:06:00
SPEAKER_01: Right.
1:06:01
SPEAKER_01: How productive you are.
1:06:01
SPEAKER_01: 20,000 lines of code a day I've written.
1:06:04
SPEAKER_01: OK.
1:06:05
SPEAKER_01: Like for an entire week sustainably.
1:06:06
SPEAKER_01: OK.
1:06:07
SPEAKER_01: But it will cost you.
1:06:08
SPEAKER_01: You'll have to do a bank heist.
1:06:10
SPEAKER_03: Yeah.
1:06:10
SPEAKER_03: But where does all these lines of code go?
1:06:13
Because,
1:06:13
you know,
1:06:14
one example that stuck with me recently,
1:06:16
it was on Twitter.
1:06:17
SPEAKER_03: I'll have to credit whoever it was.
1:06:20
SPEAKER_03: But they told their agent,
1:06:21
SPEAKER_03: like,
1:06:22
SPEAKER_03: look,
1:06:22
SPEAKER_03: I want you to solve this problem,
1:06:24
SPEAKER_03: which is like,
1:06:25
SPEAKER_03: let's not do two things at once.
1:06:27
SPEAKER_02: Right.
1:06:28
SPEAKER_03: basically locking.
1:06:29
SPEAKER_03: And the agent spun up a new Redis server,
1:06:33
SPEAKER_03: added a new service that implemented optimistic or pessimistic locking.
1:06:37
SPEAKER_03: It was like 4,000 lines of code.
1:06:39
SPEAKER_03: And it was a Rails project.
1:06:41
SPEAKER_03: The person was like, hold on, maybe don't do all that.
1:06:45
SPEAKER_03: And then it kind of went on and did something in Redis.
1:06:48
SPEAKER_03: And in the end,
1:06:49
SPEAKER_03: because this person knew Redis,
1:06:50
SPEAKER_03: it just needed to use a keyword that does the locking.
1:06:53
SPEAKER_03: And then it kind of told, just do this.
1:06:56
SPEAKER_03: But the point is, you know, these agents can write a lot of code.
1:06:59
SPEAKER_03: And I'm wondering about two things.
1:07:01
SPEAKER_03: One, like how sustainable is it?
1:07:02
Because we've seen junior developers even before AI just like,
1:07:05
you know,
1:07:06
like spitting out code.
1:07:07
SPEAKER_03: And then like what's going to happen with maintainability?
1:07:10
SPEAKER_03: And is it good code?
1:07:10
SPEAKER_03: Is it the code that you actually want?
1:07:12
Because I'm also hearing that people are using agents are
1:07:15
writing the first thing,
1:07:16
but they're going back and they're kind of changing it to keep
1:07:18
their coding style
1:07:19
SPEAKER_03: or like to tidy it up and that kind of stuff.
1:07:21
SPEAKER_01: Yeah,
1:07:22
SPEAKER_01: look,
1:07:22
SPEAKER_01: I mean,
1:07:23
SPEAKER_01: the answer is you can do all of this as a professional engineer today,
1:07:27
and you can get a gazillion PRs through if your organization is
1:07:32
willing to speed up
1:07:34
SPEAKER_01: the bottlenecks that emerge when you start generating code at that rate.
1:07:37
SPEAKER_00: Mm-hmm.
1:07:38
And some organizations are and some organizations aren't
1:07:41
willing to let that speed up.
1:07:42
SPEAKER_01: And you're going to start seeing them separate very quickly.
1:07:44
SPEAKER_01: And of the ones who decide to do it,
1:07:46
you'll see some of them turn into train wrecks that become very
1:07:48
public potentially.
1:07:49
SPEAKER_01: And then you'll see some of them succeed.
1:07:51
You really want to be in the I tried it and I succeeded
1:07:53
category,
1:07:54
I think.
1:07:55
SPEAKER_01: And that means you're going to have to take some risks.
1:07:58
The only advice I would give people,
1:07:59
I would say,
1:08:00
look,
1:08:01
look,
1:08:01
because our book is 300 pages.
1:08:03
SPEAKER_01: How do you write 300 pages about vibe coding?
1:08:06
SPEAKER_01: Can it really be that hard?
1:08:07
SPEAKER_01: And the answer is Gene and I spent five months.
1:08:10
SPEAKER_01: We wrote the book in a month after spending five months doing deep,
1:08:13
SPEAKER_01: deep,
1:08:14
deep dive researching on how do you push the LLM and VIVE
1:08:17
coding in different ways
1:08:19
and found a bunch of anti-patterns and found a bunch of
1:08:21
patterns and found that
1:08:22
SPEAKER_01: it's extremely hard.
1:08:23
SPEAKER_01: It's non-intuitive.
1:08:25
SPEAKER_01: Nobody's born knowing how to do it.
1:08:26
SPEAKER_01: It's completely new to humanity.
1:08:28
They have these sort of human-like but non-human distinctly
1:08:31
different helpers.
1:08:32
And the best advice that I can possibly give you is give them
1:08:36
the tiniest task,
1:08:38
SPEAKER_01: the most molecularly tiny segmented task you can give them.
1:08:42
SPEAKER_01: And if you can find a way to make it smaller, do that, okay?
1:08:45
SPEAKER_01: At a time,
1:08:46
keep real careful track with them on what they're working on at
1:08:49
all times,
1:08:50
SPEAKER_01: and then own every line of code that they ultimately commit.
1:08:54
And if you follow those rules,
1:08:55
then you'll be astoundingly productive without causing.
1:08:58
SPEAKER_01: But,
1:08:58
SPEAKER_01: man,
1:08:59
SPEAKER_01: dude,
1:08:59
I've already personally caused so many nightmares because
1:09:02
Claude hacking its reward
1:09:03
SPEAKER_01: function and saying,
1:09:04
SPEAKER_01: hey,
1:09:04
SPEAKER_01: your tests are all done,
1:09:05
SPEAKER_01: right?
1:09:06
SPEAKER_01: So, I mean, like, this is not easy.
1:09:08
SPEAKER_01: And it's not going to get any easier.
1:09:09
SPEAKER_01: That's the painful part, man.
1:09:11
SPEAKER_01: And that's what people are struggling with.
1:09:12
SPEAKER_01: The AIs will get smarter and they won't hack the reward function anymore,
1:09:16
SPEAKER_01: but they'll have some other problems.
1:09:18
SPEAKER_00: OK,
1:09:18
and there's always going to be another problem and it'll never
1:09:20
be ready enough for
1:09:21
SPEAKER_01: somebody to come in and just like it just works.
1:09:24
SPEAKER_01: That's what everybody is asking for and what they want.
1:09:27
And you hear on Hacker News,
1:09:28
I mean,
1:09:28
anybody says I've been successful with AI.
1:09:30
SPEAKER_01: Everyone says, well, I tried it and I wasn't successful.
1:09:32
They're not realizing that you can today,
1:09:35
but it's not it's not a freebie.
1:09:37
SPEAKER_01: It's a tool that you have to learn how to use.
1:09:39
SPEAKER_03: So in the book,
1:09:40
you use an example of when you kind of turned the page of like
1:09:43
actually believing
1:09:43
SPEAKER_03: this stuff,
1:09:44
SPEAKER_03: which was around your game that you have been building for.
1:09:47
SPEAKER_03: I remember actually when you retired,
1:09:49
you announced that you're working on this game and you were
1:09:52
making some progress
1:09:53
SPEAKER_03: and releasing it.
1:09:54
SPEAKER_03: What happened there in terms of using AI to get back to the game?
1:09:59
SPEAKER_03: And what was the outcome?
1:10:00
SPEAKER_03: Where are you with that game right now?
1:10:02
SPEAKER_03: And what is the game for those who don't know?
1:10:04
SPEAKER_01: Oh, the game's called Wyvern.
1:10:06
SPEAKER_01: It was a hobby game I started in 1995.
1:10:09
SPEAKER_01: It's a massively multiplayer, like, you know, RPG online.
1:10:13
SPEAKER_01: But it's 2D, all 2D sort of pixie sprite graphics.
1:10:17
Super,
1:10:17
super high speed animation,
1:10:19
though,
1:10:19
with like spells flying around and stuff.
1:10:20
SPEAKER_01: It's a lot of fun, man.
1:10:21
SPEAKER_01: People love it.
1:10:22
SPEAKER_01: They have a soft spot for it.
1:10:23
SPEAKER_01: People continue to play the game for literally decades.
1:10:26
SPEAKER_00: Oh, wow.
1:10:27
And I have volunteer contributors working on it right now
1:10:29
who've been working on it
1:10:30
SPEAKER_01: for years and years and years.
1:10:31
SPEAKER_01: So labor of love, for sure.
1:10:33
SPEAKER_01: During that time when I said I was working on it during COVID,
1:10:36
I got it on Steam and I got a bunch of cloud overhauls done and
1:10:39
modernized it.
1:10:39
SPEAKER_01: And it was all really fun.
1:10:41
But the player base got so excited about it and they asked for
1:10:44
so much features,
1:10:45
right?
1:10:45
SPEAKER_01: They asked for so much work from me that I suffocated me.
1:10:53
SPEAKER_01: as the owner of the game, right?
1:10:55
SPEAKER_01: And so I gave up, and that's when I was, like, really done coding.
1:10:58
SPEAKER_01: And then AI has come back and put it all back on the table for me.
1:11:01
SPEAKER_01: I realized,
1:11:02
SPEAKER_01: oh,
1:11:02
SPEAKER_01: my God,
1:11:02
SPEAKER_01: like,
1:11:04
this thing can churn through my bug backlog that the players
1:11:07
had asked me to go
1:11:07
SPEAKER_01: fix,
1:11:08
SPEAKER_01: right?
1:11:09
SPEAKER_01: And I'll have time to spare, right?
1:11:12
And this is why,
1:11:12
I mean,
1:11:13
like,
1:11:13
this is why people are coming out of retirement right now.
1:11:15
SPEAKER_03: And then,
1:11:15
SPEAKER_03: so on that game,
1:11:16
SPEAKER_03: you went back and you started to implement,
1:11:18
SPEAKER_03: like,
1:11:19
SPEAKER_03: certain features with AI?
1:11:20
SPEAKER_03: Yeah.
1:11:21
SPEAKER_01: Yeah.
1:11:21
So the thing is,
1:11:23
I've been working on Sourcegraph Codi,
1:11:25
coding on Codi for quite some time.
1:11:27
SPEAKER_01: And then the agents came out.
1:11:28
SPEAKER_01: And I was like, you know what?
1:11:29
SPEAKER_01: I'm going to try it on a – because all we had was a brand new code base.
1:11:33
SPEAKER_01: I want to try it on a crummy old legacy code base.
1:11:36
SPEAKER_01: 30 years old is pretty crummy and pretty legacy.
1:11:37
SPEAKER_01: It really is, man.
1:11:38
SPEAKER_01: It was bad.
1:11:39
SPEAKER_01: So that's what I've been doing is I've been doing different things.
1:11:41
SPEAKER_01: I've been doing cleanups.
1:11:42
SPEAKER_01: I've been adding tests.
1:11:43
SPEAKER_01: I've been doing migrations.
1:11:44
SPEAKER_01: All the things that a larger company would need to do.
1:11:46
SPEAKER_01: Yeah?
1:11:47
SPEAKER_01: Yeah.
1:11:47
Because I have lots of experience with those at Amazon and
1:11:49
Google and so forth.
1:11:50
SPEAKER_01: And so you can scale it up.
1:11:52
SPEAKER_01: You can say, okay, I'm doing it for Wyvern.
1:11:54
And this is what the experience you're going to get as a
1:11:56
developer in a year and a
1:11:57
SPEAKER_01: half,
1:11:57
SPEAKER_01: two years,
1:11:58
SPEAKER_01: working on a giant enterprise code base,
1:12:00
SPEAKER_01: right?
1:12:01
SPEAKER_01: And the answer is it's going to be real different.
1:12:02
SPEAKER_01: It's going to be a lot of fun.
1:12:04
SPEAKER_01: It's going to be really hard still.
1:12:07
SPEAKER_01: And it's just a completely different role.
1:12:08
SPEAKER_01: You don't write code anymore.
1:12:10
SPEAKER_01: You build software.
1:12:11
So on this game,
1:12:13
but just going back,
1:12:15
you're describing the AI what to do.
1:12:19
It turns out the code,
1:12:20
you look at it,
1:12:21
you test it,
1:12:22
and then you push it out?
1:12:23
SPEAKER_01: It is a very complicated process that's way too long to talk about here.
1:12:28
SPEAKER_01: It is built inherently on a foundation of distrust.
1:12:31
SPEAKER_01: You cannot trust anything the LLN gives you.
1:12:35
SPEAKER_00: Anything.
1:12:35
And that means multiple safeguards and guardrails and sentries
1:12:38
and security and practices.
1:12:40
And you have to train yourself to say the right things and do
1:12:42
the right things and
1:12:43
SPEAKER_01: look for the right things.
1:12:44
SPEAKER_01: And it is not easy.
1:12:45
And it has reinforced my belief that people who are really good
1:12:50
developers are
1:12:50
going to thrive in this new world because it takes all of your
1:12:54
skill to keep these
1:12:55
SPEAKER_01: things on the rails.
1:12:56
SPEAKER_03: Do I hear it correctly that what we're kind of saying,
1:12:59
SPEAKER_03: because at first I might have misunderstood you first.
1:13:01
SPEAKER_03: It's like, all right, you know, like companies, you should invest in it.
1:13:04
SPEAKER_03: You should do it because otherwise you'll be left behind.
1:13:06
SPEAKER_03: But it might be a little bit like what we've seen with,
1:13:08
SPEAKER_03: let's say,
1:13:09
SPEAKER_03: early Google,
1:13:09
SPEAKER_03: you know,
1:13:10
like Google was building out all their platforms and they're
1:13:13
not really making a
1:13:13
SPEAKER_03: secret or let's say Amazon is a better example.
1:13:16
They were like building all these internal APIs that talk to
1:13:19
each other,
1:13:19
which no one did.
1:13:20
It seemed like a lot of work to do and it didn't seem why you
1:13:24
shouldn't just stick
1:13:24
SPEAKER_03: with what you have.
1:13:25
SPEAKER_03: But, you know, 20 years later, Amazon actually, like, built AWS.
1:13:29
SPEAKER_03: They have an organization that actually everyone talks with APIs.
1:13:32
SPEAKER_03: And some companies are still have not figured out,
1:13:35
SPEAKER_03: you know,
1:13:35
SPEAKER_03: like we can look at,
1:13:36
SPEAKER_03: for example,
1:13:36
SPEAKER_03: Google.
1:13:37
SPEAKER_03: So what we might be saying is,
1:13:38
SPEAKER_03: like,
1:13:38
SPEAKER_03: look,
1:13:38
SPEAKER_03: this future is coming,
1:13:39
SPEAKER_03: but it's going to be a lot of work.
1:13:41
SPEAKER_03: Like, start now because you will need to figure out so many things.
1:13:44
SPEAKER_03: And it's not just going to be a... That's right.
1:13:46
The call to action is absolutely not give agents to all of your
1:13:49
developers.
1:13:50
SPEAKER_01: That would be...
1:13:52
That would be an apocalyptic event for your company in more
1:13:55
ways than one.
1:13:56
But what you should do is you should start getting some of your
1:13:59
developers together
1:14:00
SPEAKER_01: to understand what is going to have to change in your company.
1:14:04
And I don't just mean the technology and the IT stuff and
1:14:07
deployments and monitoring.
1:14:08
SPEAKER_01: I mean, like, the business processes.
1:14:10
What's going to have to change if suddenly code generation is
1:14:13
no longer the bottleneck?
1:14:14
SPEAKER_01: Because it's historically always been the bottleneck.
1:14:17
SPEAKER_01: And so we've allowed everything else to just kind of, like, coast.
1:14:19
SPEAKER_01: Yeah.
1:14:21
And this is why I really wanted to talk about your game because
1:14:24
I think this was
1:14:24
really helpful for me because what I'm trying to understand is
1:14:27
what does it look
1:14:28
SPEAKER_03: like when we use these?
1:14:29
SPEAKER_03: And I'm glad that you said that it wasn't that,
1:14:31
SPEAKER_03: I don't know,
1:14:32
SPEAKER_03: all your bugs are now suddenly fixed magically.
1:14:34
No,
1:14:34
it's going to be years and years of work,
1:14:35
but I'll be going 100 times faster,
1:14:38
so it's fun.
1:14:38
SPEAKER_03: Yeah, but by the time you finish.
1:14:40
SPEAKER_01: Yeah.
1:14:40
SPEAKER_03: Yeah.
1:14:41
SPEAKER_03: And in the book,
1:14:42
SPEAKER_03: a thing that I liked,
1:14:43
SPEAKER_03: again,
1:14:44
SPEAKER_03: you made a prediction about how jobs will be impacted.
1:14:47
SPEAKER_03: And I kind of thought,
1:14:48
SPEAKER_03: we exchanged emails earlier,
1:14:51
SPEAKER_03: and I kind of thought you would be saying there will be fewer jobs.
1:14:54
SPEAKER_03: In the book, you actually say the opposite.
1:14:56
SPEAKER_03: You said that you think there will actually be a lot more developer jobs.
1:15:01
SPEAKER_03: Why do you see this?
1:15:03
SPEAKER_03: But what will change?
1:15:03
SPEAKER_03: They're all going to be the same things as today, right?
1:15:05
SPEAKER_03: Yeah.
1:15:07
It's so hard for people to get their heads around because
1:15:09
what's happening is
1:15:10
SPEAKER_01: we're,
1:15:11
SPEAKER_01: you know,
1:15:12
SPEAKER_01: commoditizing the creation of software,
1:15:13
SPEAKER_01: just like digital cameras commoditized photography.
1:15:17
SPEAKER_01: Right.
1:15:17
SPEAKER_01: Everybody can take nice professional pictures now.
1:15:20
SPEAKER_01: And that was inconceivable back in the 80s.
1:15:22
SPEAKER_01: Inconceivable.
1:15:22
SPEAKER_01: Yeah.
1:15:23
Yeah,
1:15:23
I mean,
1:15:24
how much would have these things cost,
1:15:25
like,
1:15:25
you know,
1:15:26
just 20 years ago?
1:15:27
SPEAKER_01: And by the way, everybody crapped all over digital photography for years.
1:15:30
SPEAKER_01: Oh, yeah.
1:15:30
SPEAKER_01: And they were like, it'll never, it'll never.
1:15:32
SPEAKER_01: There was a lot of it'll nevers being thrown around.
1:15:34
SPEAKER_03: Well, and Kodak went bankrupt on not believing it.
1:15:37
SPEAKER_03: They actually buried their own digital camera.
1:15:39
SPEAKER_01: Yeah, yeah, yeah, yeah, yeah.
1:15:40
SPEAKER_01: So, like, we're in that situation again.
1:15:42
SPEAKER_01: Everybody's like, AI will never.
1:15:44
SPEAKER_01: They are wrong.
1:15:45
SPEAKER_01: AI will ever.
1:15:46
SPEAKER_01: It will get to where all the places that you think you're...
1:15:49
SPEAKER_01: You don't think that it's going right now.
1:15:51
And what's going to happen is your mom will be able to create
1:15:53
software,
1:15:54
okay?
1:15:55
SPEAKER_01: Your boss will be able to create software.
1:15:58
SPEAKER_01: Somebody at McDonald's will be able to create software.
1:16:00
SPEAKER_01: Like,
1:16:00
SPEAKER_01: literally,
1:16:01
SPEAKER_01: we're going to find all the Ramanujans,
1:16:02
SPEAKER_01: you know,
1:16:02
SPEAKER_01: the undiscovered real geniuses in the world,
1:16:05
SPEAKER_01: right?
1:16:06
SPEAKER_01: My friend Brendan Hopper,
1:16:07
SPEAKER_01: he's the head of technology,
1:16:09
SPEAKER_01: CTO for technology at Commonwealth Bank of Australia.
1:16:12
He's got some amazing hypotheses about how AI is going to bring
1:16:16
out a meritocracy.
1:16:18
SPEAKER_01: Because AI is a spotlight.
1:16:19
SPEAKER_01: It shines on all the work that people are doing.
1:16:22
SPEAKER_01: And you can't hide shoddy work anymore.
1:16:24
SPEAKER_01: The AI will detect it.
1:16:25
SPEAKER_01: If you're hoarding knowledge...
1:16:28
SPEAKER_01: Like you're an engineer who hoards knowledge to keep your job security.
1:16:32
SPEAKER_01: That's gone now.
1:16:34
SPEAKER_01: The AI knows everything you know now.
1:16:35
SPEAKER_02: To be honest, there are always these stories about doing so.
1:16:38
SPEAKER_02: I never really believed that.
1:16:40
SPEAKER_01: It happens, but it's a rare edge case.
1:16:41
But there's other common edge cases where people manipulate the
1:16:44
system to try to
1:16:45
SPEAKER_01: benefit whatever they want instead of what's best for the system.
1:16:48
SPEAKER_01: The AI is eventually going to highlight that.
1:16:50
SPEAKER_01: And so all the people with merit,
1:16:51
SPEAKER_01: meaning the people who are good at using AI to get important things done,
1:16:54
SPEAKER_01: I guess,
1:16:56
SPEAKER_01: are going to bubble to the top.
1:16:57
And there are going to be an astounding number of jobs because
1:17:00
creating software is
1:17:00
SPEAKER_01: so much more empowering than creating pictures.
1:17:03
SPEAKER_01: If anybody can create a video, so what?
1:17:04
SPEAKER_01: But if everybody can create software, that's mind-blowing.
1:17:07
So you know what I think is going to happen is I think big
1:17:09
companies are going to
1:17:10
SPEAKER_01: shed a lot of jobs.
1:17:11
SPEAKER_01: I think a lot of people are not going to work for big companies.
1:17:13
SPEAKER_01: There are going to be a bazillion startups.
1:17:15
See,
1:17:16
one thing that I'm not 100% on this is big companies are highly
1:17:21
profitable.
1:17:22
SPEAKER_03: And I could see them shedding certain jobs, but then replacing it.
1:17:26
SPEAKER_03: But they will want to keep their edge.
1:17:29
SPEAKER_03: Like,
1:17:29
SPEAKER_03: you know,
1:17:30
SPEAKER_03: they will,
1:17:30
SPEAKER_03: of course,
1:17:30
SPEAKER_03: want to try to increase profitability,
1:17:32
but they're happy keeping it at level and making and having
1:17:36
enough reserve to like
1:17:37
SPEAKER_03: fight off the startups.
1:17:38
SPEAKER_03: Right.
1:17:38
SPEAKER_01: Absolutely.
1:17:39
SPEAKER_01: I mean, there's that balance will always be there, that tension.
1:17:43
SPEAKER_01: So,
1:17:43
SPEAKER_01: but I mean,
1:17:44
SPEAKER_01: I just,
1:17:44
I feel like right now the calculus is not looking in favor of
1:17:47
big companies bulking
1:17:49
SPEAKER_01: up any further.
1:17:49
SPEAKER_01: Like I don't see big companies getting bigger.
1:17:51
SPEAKER_03: Well, in fact, I just, we were doing a Google deep dive.
1:17:54
SPEAKER_03: I saw that Google peaked as headcount in 2022.
1:17:57
SPEAKER_03: It's been kind of like going slowly a little bit down.
1:18:00
SPEAKER_03: It was like 188,000 or something.
1:18:01
SPEAKER_03: So actually like it,
1:18:03
SPEAKER_03: and this is Google we're talking about,
1:18:04
SPEAKER_03: which is profitability and revenue keeps going up.
1:18:06
SPEAKER_03: So.
1:18:07
SPEAKER_01: Yeah,
1:18:07
right now companies are discovering the easy solution is you
1:18:10
can do the same that
1:18:11
SPEAKER_01: you've been doing for cheaper by,
1:18:13
SPEAKER_01: you know,
1:18:13
SPEAKER_01: losing some headcount and doing some stuff with AI,
1:18:15
SPEAKER_01: right?
1:18:16
And I think the more ambitious ones are going to do,
1:18:18
they're going to be more ambitious.
1:18:20
SPEAKER_03: So you've done your game.
1:18:22
SPEAKER_03: I want to ask you about a metaphor that I've been thinking about,
1:18:25
SPEAKER_03: and I asked you to poke some holes in it,
1:18:29
SPEAKER_03: the ones you see.
1:18:30
SPEAKER_03: Game development.
1:18:31
SPEAKER_03: And game development for...
1:18:33
If you think back of what the biggest barrier of entry used to
1:18:36
be to build a nice,
1:18:37
SPEAKER_03: like,
1:18:38
SPEAKER_03: cool game,
1:18:39
SPEAKER_03: it was initially building the 3D engine.
1:18:41
SPEAKER_03: You know, this is why Doom was massive, Wolfenstein.
1:18:44
SPEAKER_03: They built the engine, and then they kind of built the game around it.
1:18:47
SPEAKER_03: But, you know, like, that was, like, 90s.
1:18:49
SPEAKER_01: That guy's my next door neighbor, by the way.
1:18:51
SPEAKER_01: Michael Abrash, the one that made Doom fast.
1:18:53
SPEAKER_01: Really?
1:18:53
SPEAKER_01: And Quake.
1:18:54
SPEAKER_03: Wow.
1:18:54
SPEAKER_03: Wow.
1:18:55
SPEAKER_03: And over time,
1:18:56
SPEAKER_03: you know,
1:18:57
SPEAKER_03: now we actually have software,
1:18:59
SPEAKER_03: Unity and Unreal Engine,
1:19:02
SPEAKER_03: which take care of the engine so you can now focus on the games.
1:19:05
SPEAKER_03: And what this has resulted in,
1:19:07
SPEAKER_03: I've now interviewed a few people,
1:19:10
SPEAKER_03: very small teams can also make really,
1:19:11
SPEAKER_03: really cool games.
1:19:13
If you actually want to build a game,
1:19:14
I actually did a Unity tutorial,
1:19:16
I could build a game.
1:19:17
SPEAKER_03: I mean,
1:19:17
SPEAKER_03: I would need to put in the work,
1:19:19
SPEAKER_03: but it's no longer like it can look professional and all these things.
1:19:22
SPEAKER_03: And if I look at how the game industry has evolved,
1:19:26
SPEAKER_03: I'm following a little bit of the news.
1:19:28
SPEAKER_03: AAA studios are mostly struggling, not all of them.
1:19:30
SPEAKER_03: You know,
1:19:30
SPEAKER_03: GTA 6 is still doing great and some of them and the EA Sports,
1:19:33
but some traditionally massive studios are struggling because
1:19:37
it doesn't work that
1:19:38
SPEAKER_03: we throw a bunch of money and we get a bestseller.
1:19:40
SPEAKER_03: There's a lot more indie games, way more than ever.
1:19:43
SPEAKER_03: They're having trouble consistently...
1:19:46
doing so i'm wondering if we might see something similar
1:19:49
because again like like
1:19:50
there the game engine was central to all all of this and now
1:19:53
everything that is not
1:19:54
the game engine is really important marketing story all those
1:19:57
things in software
1:19:58
engineering coding like being able to code was the bottleneck
1:20:02
and now you know that
1:20:03
will to some extent be removed but software engineering is
1:20:07
still everything around
1:20:08
SPEAKER_03: this still still remains
1:20:09
SPEAKER_01: That for sure.
1:20:10
SPEAKER_01: That is absolutely true.
1:20:12
SPEAKER_01: So,
1:20:12
SPEAKER_01: yeah,
1:20:12
SPEAKER_01: we're going to see we're going to see a lot more software get created,
1:20:15
SPEAKER_01: period,
1:20:15
SPEAKER_01: just like a lot more small software.
1:20:18
And we're going to see more indie games and we're going to see
1:20:20
more stuff bubble up
1:20:21
SPEAKER_01: that's high quality.
1:20:23
Somebody's going to find a way to organize it all like the App
1:20:25
Store organized,
1:20:26
you know,
1:20:26
apps.
1:20:27
SPEAKER_03: Maybe we'll see a new startup for this.
1:20:29
SPEAKER_01: Man,
1:20:29
SPEAKER_01: dude,
1:20:30
SPEAKER_01: I'm telling you,
1:20:30
SPEAKER_01: man,
1:20:31
SPEAKER_01: almost every time I talk to anybody about this,
1:20:33
SPEAKER_01: we come up with a couple of new billion-dollar ideas,
1:20:36
SPEAKER_01: right?
1:20:36
SPEAKER_01: I mean,
1:20:37
it's like this is another reason I think there's going to be so
1:20:39
many jobs,
1:20:40
SPEAKER_01: is that this will create legitimate,
1:20:42
SPEAKER_01: real,
1:20:43
SPEAKER_01: actual GDP productivity.
1:20:45
SPEAKER_01: Nothing fake about it, nothing artificial.
1:20:47
SPEAKER_01: It will create real value.
1:20:48
SPEAKER_01: It's going to be an explosion of value, right?
1:20:51
SPEAKER_01: It's going to take...
1:20:52
a couple of tipping points for the AI to reach this sort of
1:20:54
mass market ability for
1:20:56
SPEAKER_01: people to be able to use it to create reliable software.
1:20:59
SPEAKER_01: But we're no more than two years away from that, man.
1:21:01
And it's going to be like this incredible proliferation of just
1:21:05
cool shit for you to try.
1:21:06
SPEAKER_01: There's going to be too much, actually.
1:21:07
SPEAKER_01: You're going to have to have AI to help you find your way through it.
1:21:10
SPEAKER_03: So in those two years,
1:21:12
SPEAKER_03: whether a listener is a less experienced engineer,
1:21:16
SPEAKER_03: especially if they're an experienced engineer,
1:21:18
what would your advice be to prepare best to make the most of
1:21:24
either being an AI
1:21:24
SPEAKER_03: engineer,
1:21:25
SPEAKER_03: working with these tools,
1:21:26
SPEAKER_03: figuring them out?
1:21:27
SPEAKER_03: What is the tactic?
1:21:28
SPEAKER_03: What is the advice that you give to engineers working,
1:21:31
SPEAKER_03: let's say,
1:21:31
SPEAKER_03: a source graph where you're at,
1:21:33
SPEAKER_03: who are around you?
1:21:34
SPEAKER_01: Yeah.
1:21:35
SPEAKER_01: So, you know, what's the guy that wrote the movie The Room?
1:21:39
SPEAKER_01: Tommy Wiseau.
1:21:40
SPEAKER_01: I think that's his name.
1:21:42
SPEAKER_01: Somebody asked him on Twitter.
1:21:43
SPEAKER_01: They were like, hey, man, I want to start writing a screenplay.
1:21:45
SPEAKER_01: What should I do?
1:21:46
SPEAKER_01: And he said, start.
1:21:47
SPEAKER_01: Right.
1:21:48
SPEAKER_00: Yeah.
1:21:48
SPEAKER_01: I mean,
1:21:48
SPEAKER_01: like,
1:21:49
SPEAKER_01: for starters,
1:21:50
SPEAKER_01: if you're saying,
1:21:51
SPEAKER_01: oh,
1:21:51
SPEAKER_01: I don't know,
1:21:52
SPEAKER_01: buddy,
1:21:52
SPEAKER_01: I am not ready,
1:21:53
SPEAKER_01: blah,
1:21:53
SPEAKER_01: blah,
1:21:53
SPEAKER_01: blah.
1:21:54
SPEAKER_01: Shut up.
1:21:55
SPEAKER_01: Okay, that's done.
1:21:57
SPEAKER_01: You're done whining.
1:21:58
SPEAKER_01: Okay, go learn it right now.
1:22:02
I had the privilege of speaking with Dario Amadei privately for
1:22:05
30 minutes about
1:22:06
SPEAKER_01: three weeks ago,
1:22:08
SPEAKER_01: four weeks ago.
1:22:09
SPEAKER_01: He invited me to come chat with him.
1:22:11
SPEAKER_01: And I got to hear his sort of unvarnished view of the very,
1:22:15
very near future from somebody who could arguably be considered
1:22:19
one of the best
1:22:20
SPEAKER_01: informed people in the world.
1:22:21
SPEAKER_03: Yeah.
1:22:22
SPEAKER_01: And Dario,
1:22:23
SPEAKER_01: you know,
1:22:23
his vision of the future is a little bit more bleak than he
1:22:26
lets on publicly.
1:22:27
SPEAKER_01: And he and Jason Clinton,
1:22:28
SPEAKER_01: his CISO,
1:22:29
SPEAKER_01: are both saying statements that are quite dire,
1:22:31
SPEAKER_01: like there will be badged AI employees by the middle of 2026.
1:22:35
SPEAKER_01: Competing with you, right, basically is the implication there.
1:22:38
SPEAKER_01: And other implications like the Moore's Law of AI,
1:22:41
SPEAKER_01: how it gets four times smarter every 18 months.
1:22:44
So if you do the math three years from now,
1:22:47
if they're IQ 10 today,
1:22:48
they'll be IQ 160.
1:22:50
If you want to choose some sort of rough measure of what 16
1:22:52
times smarter means.
1:22:54
SPEAKER_01: And it'll be too much for people.
1:22:56
Dario told me,
1:22:57
he said,
1:22:58
look,
1:22:58
he said,
1:22:58
society is like an immovable force,
1:23:00
right?
1:23:00
SPEAKER_01: An immovable object and tech and AI are an unstoppable force.
1:23:04
They just won't stop and they're going to collide and it's
1:23:07
going to be ugly.
1:23:08
Because it's going to push society harder than society wants to
1:23:12
be pushed,
1:23:13
SPEAKER_01: harder than society is willing to be pushed.
1:23:15
SPEAKER_01: And we're already seeing signs of it.
1:23:17
We're seeing people revolting against AI,
1:23:19
putting up the I'm sick of it,
1:23:20
right?
1:23:21
SPEAKER_01: He posted I'm sick of it.
1:23:21
SPEAKER_01: He never mentioned AI in the post.
1:23:22
SPEAKER_01: It was really brilliant.
1:23:23
SPEAKER_01: I love the post, by the way, the guy that wrote the I'm sick of it.
1:23:26
He's speaking for a generation of people who are tired of
1:23:29
hearing about this shit.
1:23:30
SPEAKER_01: But unfortunately, you are never going to stop hearing about it.
1:23:34
It is the way things are going to be done and in the very,
1:23:37
very,
1:23:37
very short order.
1:23:38
SPEAKER_01: And so my advice to you is get off your ass and learn it now, now, now.
1:23:43
SPEAKER_01: Start vibe coding.
1:23:43
SPEAKER_01: Figure it out.
1:23:44
SPEAKER_01: There's a lot to learn.
1:23:45
SPEAKER_01: There's a lot of weird instincts you're going to have to learn.
1:23:47
SPEAKER_01: A lot of stuff is not going to work the way you expect it to.
1:23:51
But you start now and you'll be ready because Dario calls 2026
1:23:54
the end game.
1:23:56
SPEAKER_01: And he says it without a hint of drama.
1:23:58
SPEAKER_01: He says it casually.
1:23:59
SPEAKER_01: Oh, yeah, 2026 is the endgame.
1:24:01
SPEAKER_01: You understand?
1:24:01
SPEAKER_01: That's how big this is going to be.
1:24:02
SPEAKER_01: And the first ones to fall, the first jobs are software engineers, right?
1:24:07
So you need to be on top of it to take advantage of the new
1:24:09
jobs that arise,
1:24:10
SPEAKER_01: which are software engineer V2,
1:24:12
SPEAKER_01: which use AI and get amazing things done.
1:24:15
You have to be one of them or you're going to get kicked out of
1:24:17
knowledge work altogether.
1:24:19
SPEAKER_03: Yeah,
1:24:19
SPEAKER_03: well,
1:24:20
SPEAKER_03: this is going to be part of like,
1:24:21
SPEAKER_03: I think it's clear that it's going to be,
1:24:24
SPEAKER_03: it reminds me a bit of the cloud where,
1:24:25
SPEAKER_03: you know,
1:24:26
SPEAKER_03: these days,
1:24:26
SPEAKER_03: like,
1:24:26
SPEAKER_03: yeah,
1:24:27
SPEAKER_03: every company uses a cloud,
1:24:28
SPEAKER_03: either private or public.
1:24:30
SPEAKER_03: And about 15 years ago,
1:24:32
SPEAKER_03: it was like AWS and I talked with banks,
1:24:34
SPEAKER_03: banks were like,
1:24:34
SPEAKER_03: we will never use it.
1:24:35
SPEAKER_03: We will never onboard.
1:24:36
SPEAKER_03: We will always have our data centers.
1:24:38
SPEAKER_01: Oh, yeah.
1:24:39
SPEAKER_03: And,
1:24:39
SPEAKER_03: you know,
1:24:39
there was a time where I think it was very valuable to get AWS
1:24:42
certifications and
1:24:43
SPEAKER_03: you get hired and get a salary bump.
1:24:45
SPEAKER_03: So I feel there are levels where,
1:24:48
SPEAKER_03: like,
1:24:48
I think it's clear to me that AI as infrastructure will be in
1:24:52
every single tech
1:24:53
SPEAKER_03: company.
1:24:54
And of course,
1:24:54
it will be in every single non-tech company and government and
1:24:56
all.
1:24:57
SPEAKER_03: It will happen.
1:24:58
SPEAKER_03: I don't see the time frame.
1:24:59
So I think we might disagree a little bit on how that is,
1:25:02
but it will happen.
1:25:03
SPEAKER_03: And I think your advice is absolutely solid.
1:25:05
SPEAKER_03: Like, get started now.
1:25:07
SPEAKER_03: In fact,
1:25:08
What I'm seeing now,
1:25:09
and again,
1:25:09
this was just this conversation with Jambi.
1:25:12
SPEAKER_03: Jambi said that she saw ChatGPT come out.
1:25:16
SPEAKER_03: She was at Coda.
1:25:17
Coda spun up in a few months,
1:25:18
an AI team,
1:25:19
and she said,
1:25:20
I'd like to be on that team.
1:25:21
SPEAKER_03: And they said, thank you, but no thank you.
1:25:23
SPEAKER_03: You don't have the experience.
1:25:24
SPEAKER_03: And then she thought for a while, like, I'm too late.
1:25:27
SPEAKER_03: You know, there's people been doing it for five years since Transformers.
1:25:30
SPEAKER_03: What can I do?
1:25:31
SPEAKER_03: And then she just went to hackathons.
1:25:32
SPEAKER_03: She just hacked on the side.
1:25:34
SPEAKER_03: Five months later, she was one of the best at the company.
1:25:37
SPEAKER_03: And she got on the team early on.
1:25:39
SPEAKER_03: And I think there's this thing of,
1:25:41
SPEAKER_03: like,
1:25:41
SPEAKER_03: I would suggest the listeners maybe,
1:25:43
SPEAKER_03: you know,
1:25:43
SPEAKER_03: like,
1:25:44
SPEAKER_03: put away the doomsday thing.
1:25:45
SPEAKER_03: But the point is, this thing is happening.
1:25:47
SPEAKER_03: And as you said, now is the best time.
1:25:49
SPEAKER_03: Like, yeah.
1:25:50
SPEAKER_03: Learn it and also do get motivation.
1:25:52
SPEAKER_03: Like I do think the industry will change a lot.
1:25:55
Like we'll probably look back at this time at something big
1:25:57
happened and we're in
1:25:58
SPEAKER_03: the middle of it.
1:25:59
SPEAKER_01: We are in the middle of it.
1:26:00
SPEAKER_01: And you know what?
1:26:00
The funny thing is,
1:26:01
I mean,
1:26:02
the grass really is greener on the other side here.
1:26:04
SPEAKER_01: Like it is so fun.
1:26:05
SPEAKER_01: Right.
1:26:06
It's so it's so I'm having so much fun not coding,
1:26:10
but fixing my bugs and adding features.
1:26:13
SPEAKER_01: I love it.
1:26:13
SPEAKER_03: But I also feel sometimes you are coding,
1:26:15
SPEAKER_03: you know what you expect and you correct it,
1:26:18
SPEAKER_03: so there's a lot of metacoding happening.
1:26:21
SPEAKER_01: Oh, yeah.
1:26:21
SPEAKER_01: I read 100,000 lines of code a day.
1:26:23
SPEAKER_01: Yeah.
1:26:23
SPEAKER_01: It ain't easy, right?
1:26:24
SPEAKER_01: Yeah.
1:26:24
I mean,
1:26:25
it's exhausting because if you're not reading it,
1:26:27
then stuff's slipping by you.
1:26:28
SPEAKER_00: Yeah.
1:26:29
You'll eventually figure it out that,
1:26:31
you know,
1:26:31
you want to try to catch things early.
1:26:32
SPEAKER_01: Yeah.
1:26:33
SPEAKER_01: But, man, it's like it's a different ballgame.
1:26:35
SPEAKER_01: And I love it.
1:26:35
SPEAKER_01: I'm having so much fun.
1:26:36
SPEAKER_01: And Gene Kim,
1:26:37
SPEAKER_01: my amazing co-author,
1:26:39
SPEAKER_01: who's,
1:26:39
SPEAKER_01: you know,
1:26:40
he's an author and researcher who I think probably knows
1:26:42
everybody in the entire
1:26:44
SPEAKER_01: world who's everybody.
1:26:45
SPEAKER_01: Yeah.
1:26:45
And he and I are both just unbelievably excited about
1:26:49
VibeCoding because despite
1:26:51
SPEAKER_01: the doom and gloom sound of what's happening,
1:26:53
SPEAKER_01: the only reason it's doom and gloom is people don't like change.
1:26:56
SPEAKER_01: They don't want to change the way they're working.
1:26:59
SPEAKER_03: I think so.
1:26:59
SPEAKER_03: And I've been guilty of this earlier.
1:27:02
Like when I saw this big change come at first,
1:27:05
I was like,
1:27:05
oh,
1:27:05
this is not great.
1:27:06
And,
1:27:06
you know,
1:27:06
when people are saying it'll eliminate jobs,
1:27:09
I didn't like the message.
1:27:10
SPEAKER_03: It just felt like very threatening.
1:27:11
SPEAKER_03: I think as software engineers, we're kind of used to us.
1:27:16
SPEAKER_03: us automating a bunch of jobs like customer support and,
1:27:20
SPEAKER_03: you know,
1:27:20
SPEAKER_03: like,
1:27:20
SPEAKER_03: oh,
1:27:21
SPEAKER_03: here's a cost savings of like we need fewer customers.
1:27:23
SPEAKER_03: And we never fired customer agents.
1:27:24
SPEAKER_03: We just didn't hire as much.
1:27:26
And I think this is the first time in history where our work is
1:27:30
kind of threatening us.
1:27:31
SPEAKER_03: But what I came to realize is talking to you,
1:27:33
SPEAKER_03: talking to Ken Beck,
1:27:35
SPEAKER_03: seeing my experiences,
1:27:36
SPEAKER_03: if you are a good software engineer and you are open to learning,
1:27:39
SPEAKER_03: and using these things and adding into your toolbox,
1:27:42
SPEAKER_03: you will be a better and more in-demand one.
1:27:43
SPEAKER_03: That's what I'm seeing from people who started to use this.
1:27:47
SPEAKER_03: They're now being hired as AI engineers.
1:27:48
SPEAKER_03: AI engineer is actually a software engineer who is able to use,
1:27:53
SPEAKER_03: but understand the non-deterministic part.
1:27:54
SPEAKER_03: They're going deeper into ML.
1:27:56
SPEAKER_03: So I think it's like...
1:27:58
SPEAKER_03: In some ways, it's ironic.
1:27:59
We might have had some stagnation for like 10 or 15 years where
1:28:02
you could do the
1:28:02
SPEAKER_03: same thing and be more successful.
1:28:04
SPEAKER_03: And, you know, staff engineers just it was more about managing people.
1:28:08
SPEAKER_03: And I think for the first time in 15 years, we're shaken up.
1:28:11
SPEAKER_03: And to be a great software engineer, you need to learn.
1:28:14
SPEAKER_03: You need to let your ego go,
1:28:15
SPEAKER_03: which,
1:28:15
SPEAKER_03: you know,
1:28:16
SPEAKER_03: I think that's something you've always done really well.
1:28:18
SPEAKER_01: Yeah,
1:28:19
SPEAKER_01: I mean,
1:28:20
why get your identity tied up in something that's actually kind
1:28:23
of fragile,
1:28:24
SPEAKER_01: as it turns out?
1:28:25
SPEAKER_01: Look, the way I think about it, man, software is always so big.
1:28:28
SPEAKER_01: Remember when they were building the second Death Star?
1:28:30
SPEAKER_01: I think it was in Empire Strikes Back, and it was half done.
1:28:33
SPEAKER_01: How big was that freaking thing, right?
1:28:36
SPEAKER_01: That's a typical enterprise software project right there.
1:28:38
SPEAKER_01: It's a good visualization of it, right?
1:28:40
So what if you have these robots that are 20 times as
1:28:42
productive as a human?
1:28:43
SPEAKER_01: Yeah.
1:28:44
You're still going to take freaking years and years and years
1:28:46
to build it.
1:28:47
SPEAKER_03: And there will be architects overseeing it.
1:28:49
SPEAKER_01: Right?
1:28:49
SPEAKER_01: Yeah, exactly.
1:28:51
You're going to be very grateful that you have the help of
1:28:53
these robots that are 20
1:28:54
SPEAKER_01: times faster than human data coding or 100 times faster.
1:28:57
SPEAKER_01: You're still building Death Stars, and it still takes years.
1:28:59
SPEAKER_01: Yeah.
1:29:00
SPEAKER_01: So there's still jobs.
1:29:01
SPEAKER_01: Yeah.
1:29:01
SPEAKER_01: They're just different.
1:29:03
SPEAKER_01: Traumatic events can increase your neuroplasticity.
1:29:06
SPEAKER_01: And you said we've been stagnating.
1:29:08
SPEAKER_01: Many of us have been stagnating.
1:29:09
SPEAKER_01: The reason I retired is I felt like I was stagnating.
1:29:11
SPEAKER_03: Yeah,
1:29:12
SPEAKER_03: I was thinking,
1:29:13
SPEAKER_03: I'll be honest,
1:29:13
SPEAKER_03: like now,
1:29:14
SPEAKER_03: like my publication,
1:29:15
SPEAKER_03: The Programmatic Engineer,
1:29:16
SPEAKER_03: covers,
1:29:17
SPEAKER_03: you know,
1:29:18
SPEAKER_03: like the trends happening.
1:29:19
SPEAKER_03: And I was just talking with my brother, like he's also in tech.
1:29:22
SPEAKER_03: He's the founder of Crap Docs.
1:29:24
SPEAKER_03: And I was talking about how looking back,
1:29:26
SPEAKER_03: like if AI did not happen,
1:29:28
SPEAKER_03: what would we be talking about?
1:29:29
SPEAKER_03: Is it how to more efficiently move monoliths to microservices?
1:29:33
SPEAKER_03: We've been talking about it for a few years.
1:29:35
SPEAKER_03: How to measure developer productivity even a little bit better.
1:29:38
SPEAKER_03: How to scale teams better so that how can you manage 10 teams?
1:29:42
SPEAKER_01: Can we switch to memory-safe languages like REST?
1:29:45
SPEAKER_03: Yes.
1:29:46
SPEAKER_03: And I'm like, it was getting a little bit boring.
1:29:49
SPEAKER_03: So, you know, like, I think this is a good takeaway.
1:29:52
SPEAKER_01: Yeah, we were incremental improvement mode.
1:29:54
SPEAKER_01: Yes, exactly.
1:29:55
SPEAKER_03: And this is a step change.
1:29:57
SPEAKER_03: Yeah.
1:29:57
SPEAKER_03: Absolute step change.
1:29:58
SPEAKER_03: So close off with some rapid questions, if you're okay with that.
1:30:01
SPEAKER_03: Sure.
1:30:02
SPEAKER_03: With all this AI stuff here, what is your favorite programming language?
1:30:06
SPEAKER_03: Or do you even have one?
1:30:08
SPEAKER_01: Wow, my favorite programming language?
1:30:10
SPEAKER_01: Oh, my gosh.
1:30:11
SPEAKER_01: I don't even care anymore.
1:30:13
SPEAKER_03: I'm so happy.
1:30:14
SPEAKER_03: What used to be?
1:30:15
My favorite programming before all this AI stuff made it kind
1:30:18
of unnecessary.
1:30:18
SPEAKER_01: Yeah.
1:30:19
SPEAKER_01: I really like TypeScript.
1:30:21
SPEAKER_01: Maybe I shouldn't, but there's something about it.
1:30:23
SPEAKER_01: I mean,
1:30:24
SPEAKER_01: it's just so flexible and expressive,
1:30:26
SPEAKER_01: and I think probably I would have to give it to TypeScript.
1:30:30
And what is an AI tool related to coding that you like and an
1:30:33
AI tool that has
1:30:34
SPEAKER_03: nothing to do with coding?
1:30:36
SPEAKER_01: Okay, an AI tool for coding?
1:30:39
SPEAKER_01: You should try SourceGraph AMP.
1:30:40
SPEAKER_01: It just came out yesterday.
1:30:41
SPEAKER_01: I mean, come on, man.
1:30:42
SPEAKER_01: That's what I've been using.
1:30:43
I actually turn all the permissions off and just let it run,
1:30:46
but don't do that.
1:30:48
SPEAKER_01: But it's so good.
1:30:49
SPEAKER_01: It feels so good.
1:30:51
SPEAKER_01: Yes.
1:30:52
SPEAKER_03: And then... Until there's an rm-rf.
1:30:57
SPEAKER_01: I've gotten pretty good at sandboxing.
1:30:59
SPEAKER_01: But I think I'm probably going to switch to Docker containers.
1:31:02
SPEAKER_01: Anyway, as for an AI tool that's not related to coding.
1:31:06
SPEAKER_01: Yeah, boy, I tried operator.
1:31:07
I really want something like operator that works,
1:31:10
if that makes any sense.
1:31:11
SPEAKER_01: So hopefully some very soon upcoming version of it.
1:31:15
SPEAKER_01: But it couldn't do something simple like edit my Google Doc for me.
1:31:18
It would look at it for literally 20 minutes and then just
1:31:21
delete a paragraph.
1:31:23
SPEAKER_03: I think that's a good example of we will have software explosion there.
1:31:27
SPEAKER_03: Someone will have to build it who's going to build it.
1:31:29
SPEAKER_03: We know who's going to build it.
1:31:31
SPEAKER_03: And what's the book recommendation that you had outside of your own book?
1:31:36
SPEAKER_03: Read Sapiens, man.
1:31:37
SPEAKER_03: It's such an awesome book.
1:31:38
SPEAKER_03: Well, Steve, this was great.
1:31:40
SPEAKER_03: I'm glad.
1:31:41
SPEAKER_03: I feel we went on a roller coaster.
1:31:43
SPEAKER_03: We went high, then low, and then we ended up high again.
1:31:47
SPEAKER_01: Yeah, well, you know, change can be scary, right?
1:31:49
SPEAKER_01: But this is a very positive change, in my opinion.
1:31:52
And I think it's good to just,
1:31:54
like,
1:31:55
I like that we,
1:31:56
let's just,
1:31:57
you know,
1:31:58
name what it is.
1:31:59
SPEAKER_03: It is change, and it is a big change.
1:32:00
SPEAKER_03: And I think for,
1:32:01
SPEAKER_03: I think what makes it scary for a lot of people,
1:32:03
SPEAKER_03: including,
1:32:04
SPEAKER_03: you know,
1:32:04
SPEAKER_03: my generation,
1:32:05
SPEAKER_03: I have not seen this change.
1:32:07
SPEAKER_03: Like, people who have been around the dot-com busts.
1:32:10
SPEAKER_03: might have seen it.
1:32:11
SPEAKER_03: When I talked to Grady Booch,
1:32:12
SPEAKER_03: he actually told me,
1:32:13
SPEAKER_03: like,
1:32:13
SPEAKER_03: oh,
1:32:13
SPEAKER_03: actually,
1:32:13
SPEAKER_03: Ken Beck was saying,
1:32:14
SPEAKER_03: we've seen this change,
1:32:15
SPEAKER_03: like,
1:32:16
SPEAKER_03: when we moved to microprocessors,
1:32:18
SPEAKER_03: for example.
1:32:19
SPEAKER_03: Like,
1:32:19
SPEAKER_03: apparently,
1:32:20
SPEAKER_03: it was a huge thing,
1:32:21
SPEAKER_03: and everyone's world tripped because they're so much faster now.
1:32:23
SPEAKER_03: They were going to, you know, change everything.
1:32:26
SPEAKER_03: And then Ken Beck said,
1:32:27
SPEAKER_03: like,
1:32:28
SPEAKER_03: yeah,
1:32:28
SPEAKER_03: everything changed,
1:32:28
SPEAKER_03: and then,
1:32:29
SPEAKER_03: like,
1:32:29
SPEAKER_03: in some ways,
1:32:30
SPEAKER_03: nothing changed.
1:32:31
SPEAKER_03: Yeah.
1:32:32
SPEAKER_01: That's a good point.
1:32:33
SPEAKER_01: Everybody suddenly had a computer one day.
1:32:35
SPEAKER_01: I was there for that.
1:32:36
SPEAKER_01: And before that, nobody had a computer and it was inconceivable.
1:32:38
SPEAKER_01: Right.
1:32:39
So everybody being able to create software is a really
1:32:41
interesting step in that direction.
1:32:42
SPEAKER_03: Well,
1:32:42
SPEAKER_03: because back then,
1:32:43
SPEAKER_03: right,
1:32:43
SPEAKER_03: as I understand,
1:32:44
SPEAKER_03: as a programmer,
1:32:45
you had to go to work to these companies which had these
1:32:48
massive computers and
1:32:49
SPEAKER_03: whatever.
1:32:50
SPEAKER_03: So it was only very privileged.
1:32:52
SPEAKER_03: And then suddenly anyone could do it.
1:32:53
SPEAKER_03: Yeah, that's right.
1:32:54
Or who had the money,
1:32:55
who had like,
1:32:56
you know,
1:32:56
rich parents or whatever savings.
1:32:59
SPEAKER_01: PCs were the beginning of the big boom.
1:33:01
SPEAKER_01: So we are at the beginning of a big boom.
1:33:03
SPEAKER_01: There's a lot of money to be made.
1:33:04
SPEAKER_03: And PCs turned out to be pretty great for us software engineers.
1:33:07
SPEAKER_01: Yeah.
1:33:08
SPEAKER_01: All right, Steve.
1:33:09
SPEAKER_01: This was great.
1:33:10
SPEAKER_01: This was awesome, man.
1:33:11
SPEAKER_01: Thanks.
1:33:12
I hope you enjoyed this interesting and entertaining
1:33:14
conversation with Steve.
1:33:16
SPEAKER_03: Steve remains a prolific writer,
1:33:17
SPEAKER_03: and you can read more of his rants linked in the show notes below.
1:33:21
SPEAKER_03: For more in-depth reading about developer tools,
1:33:23
SPEAKER_03: the engineering culture at Sourcegraph,
1:33:25
SPEAKER_03: or the impact of AI on software engineering,
1:33:27
SPEAKER_03: check out the Pragmatic Engine Deep Dives also linked below.
1:33:30
SPEAKER_03: If you've enjoyed this podcast,
1:33:32
SPEAKER_03: please do subscribe on your favorite podcast platform and on YouTube.
1:33:35
This helps more people discover the podcast,
1:33:37
and a special thanks if you leave a rating.
1:33:39
SPEAKER_03: Thanks, and see you in the next one.