Latest / Oxide and Friends / Economics and Open Source

Transcript

Economics and Open Source · Oxide and Friends

• 0:00
  Speaker 1: Alright. Let's take a look. So we are talking about by popular demand, which is to say someone on the Internet, about Joel Spalsky's strategy letter 5 from 2002. But before we do, Adam, I feel we need to talk we just need to wrap some things up from last week,
• 0:19
  Speaker 2: just briefly. Okay. I didn't realize this is a surprise.
• 0:23
  Speaker 3: Right?
• 0:23
  Speaker 1: Little surprise. 1, that was a lot of fun, actually. I really I I that was just it was great to hear that. And so and I I saw, where did, Tim was here? I thought there he is.
• 0:40
  I'm gonna ask Tim to speak. So, Tim, I've invited you to speak. Hopefully, you can join us. So, I started buying a couple of these books, And then Amazon, of course, is just extremely excited that I and is now throwing all these suggestions at me. One of which was this book, Endless Loop, that I got that I love.
• 1:01
  It's so good. This is a history of basic, which I'm Nice.
• 1:05
  Speaker 2: Really That's a great title too.
• 1:07
  Speaker 1: Oh, it's a great title. Well, well, I mean, 3rd time's a charm. This guy's written 2 other histories of basic, which I think is great, including okay and not okay. To the but it's really good. I I I but there was a line in this that I tweeted out over the weekend that I just found I I don't know.
• 1:33
  I don't know if you saw this. Adam, first of all, did you ever code basic?
• 1:37
  Speaker 2: You know what? Like, not seriously. I mean, this is why the endless loop is so appealing to me because, that's basically the basic I coded on the Apple 2e in the computer lab, the, like, print Brian smells go to 10.
• 1:51
  Speaker 1: Right. Okay. So you never actually so
• 1:54
  Speaker 2: No. Not I the closest I got was I did some visual basic, but but very different. Right. So so you've
• 1:59
  Speaker 4: cut a post
• 2:00
  Speaker 1: date just slightly, not by much. Yes. It's only this is like the total, like, the, what, 5 years between difference between us is is where it basic is where it really shows up. So there's this line in the and I think this is just absolutely delightful that the well, 1, per our Martian conversation last time, Kemeny, hope I'm pronouncing that correctly, really interesting guy. Kemeny and Kurtz are the the, inventors of basic, but he's one of the Martians.
• 2:28
  So he is with Von Neumann and Nerdist and so on. Becomes the president of Dartmouth at a really young age. Really interesting kind of story there. Very progressive thinker. Invented the part of the I mean, I always viewed basic as, like, a punishment as opposed to being, like, actually deliberately trying to actually make it accessible.
• 2:50
  It was actually I was not giving you enough credit. But I love the fact that they call the basic that I learned street basic, which I think is it's like, I wanna put that on, like, a resume or something. I I came up on the mean streets of street basic.
• 3:04
  Speaker 2: That's right. You're in a book smarts basic. You have street He's street basic.
• 3:08
  Speaker 1: Yeah. He's like, no. I get and then okay. So, Tim, you are here, and I'm wondering if you can get because you had this response that I just thought was incredible on the the Gibsonian flavor. Can you tell me a little bit about your I'd like you to do an out loud reading of this tweet.
• 3:26
  Is that unreasonable to ask?
• 3:29
  Speaker 2: To demand.
• 3:29
  Speaker 1: Well, yes. It doesn't.
• 3:34
  Speaker 5: Well, I mean, I I have a copy of Neuromancer here in front of me. I I recently read it and I guess it was fresher on my mind.
• 3:41
  Speaker 1: So are you adopting a well, first, would you can I ask you to do an out loud reading?
• 3:46
  Speaker 2: Do you
• 3:46
  Speaker 1: mind if I put you on the spot to do an out loud reading from
• 3:48
  Speaker 6: your tweet?
• 3:48
  Speaker 5: Oh, yeah. Yeah. It was well, you first. It was a response to your
• 3:55
  Speaker 1: tweet. Well, so my tweet is I've I feel I've already shown you mine. I've my tweet is all about street I then I came up learning street basic.
• 4:04
  Speaker 5: Yeah. Mine was basically just, hey. That sounds like Gibson. Pace snapped the ROM into his deck and paused. His ice stayed quiet.
• 4:15
  A new strobe or a new shape flickered in the corner of his vision strobing amber at 60 hertz. He turned to squint at the glowing text. It was street basic, a cowboy's mother tongue. Beautiful. It's just it's just monkey see monkey do.
• 4:32
  It's just, you know, trying to evoke to that style.
• 4:36
  Speaker 1: Okay. Well, so here's my question.
• 4:37
  Speaker 5: Cowboys delightful heckers cowboy. You know?
• 4:40
  Speaker 1: Is do you is that a a, like, a particular passage, or is that written in the style
• 4:48
  Speaker 6: of Gibson?
• 4:49
  Speaker 5: I I just I just I just stole some lingo he uses. Jesus. Case is the name of his character. He talks about ROM and ICE for, you know, intrusion countermeasures and, you know, people aren't hackers. They're console cowboys.
• 5:05
  So, you know, man, it's just the window.
• 5:07
  Speaker 1: It's okay. I'm just telling you it's fucking art, man. That was great. It was so good. It was so literary.
• 5:13
  I I feel that, like, I I'm not really a like, that speaks to me more than Gibson itself, honestly. Not the not the to I I just I feel that it was, terrific. Very, very well written, and it reminds me of the imitation Hemingway itself. I thought so I the imitation can be better than the original, and I think that that is the case here. I'd love to say
• 5:38
  Speaker 5: I I'm actually, strictly speaking, much too young to have grown up with basic, but I did anyways. I was very my family was very, very poor and, had a hand me down Apple 2 that a neighbor gave us, like, in the mid nineties. And, I thought myself basic from one of the Apple books that came with the computer when I was, you know, 8 or 9 years old, and it it felt like magic.
• 6:07
  Speaker 1: So there is there's some there is something great about that that the this is reminds me of Antranik had that line too, Adam, earlier at an earlier space about how not having a lot as a kid, you know, you had you still had this magical machine. Even though it was a hand me down in the mid nineties, it's a decade out of date. Tim, you're still able to make this thing do magical things many years after it's putative prime.
• 6:32
  Speaker 5: Yeah. I mean, occasionally, if someone asks me, why are you a programmer? I tell them, well, I wanted to be a wizard, but this is the closest I could do.
• 6:40
  Speaker 1: I I wanted to book 1st class tickets on Pan Am to Paris. Like, that's the reason that I'm I mean, come on. I that's a, it's a war game reference, kids.
• 6:50
  Speaker 2: Yeah. I got I'm I'm hanging with you with the war game reference.
• 6:52
  Speaker 1: I got it.
• 6:52
  Speaker 7: Okay. Is that that's you know, there there
• 6:54
  Speaker 1: was a time when you might not always, so it's it's good.
• 6:56
  Speaker 8: Yeah. Tell tell
• 6:57
  Speaker 9: me about it.
• 6:57
  Speaker 1: Yeah. I'm relieved.
• 6:59
  Speaker 2: But but you're right. There there is that, it's it's first a, you know, creativity from adversity, but then also just like a simpler era where you could put all these pieces together without it being quite as complicated.
• 7:14
  Speaker 1: But you know, the kids today are building computers out of Minecraft. It's like Yeah. That's true. They recreate that simplicity for themselves, which I think is pretty
• 7:23
  Speaker 2: That's that's right. They they don't have a parallel they don't have a parallel port, and they can't like flash GPIOs, but they do have Minecraft. And it's it's very different, but you're right that it's it's scratching it very similar.
• 7:33
  Speaker 9: I actually disagree with that. I mean, they have Arduinos, which do have GPIO ports, maybe not parallel port but something very analogous to that.
• 7:42
  Speaker 1: Yeah. I know. They're doing that. Yeah. Yeah.
• 7:43
  Speaker 9: They
• 7:43
  Speaker 2: No. They they they do but but it's not, in agreed that they they have access to the same kinds of things. But I think it's it's not as, it's not necessarily in the family PC like it used you
• 7:57
  Speaker 1: have to admit, Adam, now you wanna have, like, Dan's guy So who's GPU based parenting.
• 8:01
  Speaker 4: If if if you guys don't mind, I would like to, tie together, the this this topic of using basic past its prime with an aspect of of, the history of accessible computing that I didn't get to cover a few weeks ago.
• 8:18
  Speaker 10: Yeah.
• 8:18
  Speaker 4: Go for it. So so there were in starting in the late eighties and going up through the early 2000, one of the earliest, PDAs, if you remember that term, that were designed specifically for blind people was a device called the Braille and Speak. And it it was it was a it was a portable yeah, portable device with with a braille keyboard and and, text to speech output. And it was running it was, you know, battery powered, you know, running a z 80 processor. It first came out in 1987, and I think they discontinued it in, like, 2,000 or 2,001.
• 9:00
  But there was, there's a there's a there was a blind kid in, like, 2002 who, was using one of these Braille and speak devices as his way of taking notes and doing assignments in class at school. And this device had a basic interpreter on it, and that was his introduction to programming. So
• 9:27
  Speaker 1: That's pretty great. And I I very much
• 9:30
  Speaker 4: And and and he he became a a protege of mine through a mutual friend. So but, his introduction to programming was on this was on this, PDA designed for blind people using basic.
• 9:47
  Speaker 1: That that is awesome.
• 9:50
  Speaker 2: Yep. Yep. You know, I should say, I should caveat my I I've never written basic with I did a lot of programming on the TI 82 back in the day. Yeah. It has its own flavor of basic.
• 10:02
  Speaker 1: So, Adam, you gotta read this book, man. Because the the the no. The trash 80 is all over this place. I mean, it's it it's, you'll like it.
• 10:11
  Speaker 2: Sweet. Yeah. I'm into it.
• 10:13
  Speaker 1: Alright. So that is, so I just had to get that out.
• 10:18
  Speaker 2: There we go. It's clear clearing the slate. Good now.
• 10:21
  Speaker 1: Clearing the slate. Now only 11 minutes in. So, did you beat this thing when Aarushi came out, Adam?
• 10:31
  Speaker 2: I mean, if I did. I mean, so this came out when in in 2,002? Can I jump?
• 10:35
  Speaker 1: Can I jump back
• 10:36
  Speaker 11: to the previous thing real quick? Sure.
• 10:38
  Speaker 1: The
• 10:39
  Speaker 11: the very first ISP that I worked for, we had a blind user, and he was using I believe it was actually a very, very early version of one of the assistive software programs from Dragon. And I remember going over to his apartment, because he lived 2 blocks away from us, to help him get set up. And the thing I noticed was what it was actually doing was it was reading off all of these SIS calls that were happening. So, like, when a dialogue box popped up, it would be like, dialogue blah blah blah. Button blah blah blah.
• 11:14
  Button blah blah blah. And then when he would click the mouse, it would be like, mouse button down, mouse button up. And so, like
• 11:20
  Speaker 4: That must have been, like, a very early Windows screen reader. It was doing that.
• 11:25
  Speaker 11: But I remember thinking this would actually be
• 11:27
  Speaker 4: Sorry, guys. I I'm we're I'm taking you this space further afield.
• 11:32
  Speaker 1: No. No worries, Matt. No. It's okay.
• 11:34
  Speaker 11: I I remember thinking how great that would be as an introduction to the Windows Sys call. Like, if you were trying to become a Windows programmer, you could just sit there and turn that on and just listen to it all day, and you would just start to pick up what was going on.
• 11:50
  Speaker 4: I don't think I'm acquainted with this program that you're talking about, but I I mean, I'm I'm not aware of any Windows screen reader that was quite as raw as what you're describing, but then I did come into that whole, aspect of of accessible computing a little bit late. So anyway Well,
• 12:11
  Speaker 2: I love this idea of, like, an s trace or trust sonar Exactly. Just have on the background. That sounds awesome.
• 12:17
  Speaker 1: And, well, then it also has to be said that because this, Tom, I know or the the the the snoop option to generate, network traffic over dev audio. In fact, I don't even Tom even Tom, do you wanna think about that? That might even have been you. But this was kind of famously Snoop would generate, output over dev audio. Do you remember this, Adam?
• 12:39
  Speaker 2: Yeah. Totally. Yeah. I never used it productively, but there were certain
• 12:44
  Speaker 1: I definitely never used it productively.
• 12:45
  Speaker 9: Thought they could. Exactly.
• 12:48
  Speaker 1: I thought they used it. Guilty party.
• 12:49
  Speaker 9: That was useful. That was genuinely useful.
• 12:53
  Speaker 8: Yeah. I I don't remember that. But
• 12:55
  Speaker 9: I I remember that. We used to do stuff like, you know, pipe the output of ping through said and change the word bytes to a control g or something. And that would be really useful because if you were, like, tracing down a break in thin net, you'd run that on a workstation and start mucking around with Ethernet cables. And when you found the, you know, the segment that admitted had been disconnected, all of the these are these beeps coming from all down the hallway. That was that was incredibly useful.
• 13:22
  Speaker 1: Yeah. This is like automating my role in every home construction project, which is basically to tell my father when the light was on or off, from upstairs or what have you. And the Okay. Yeah.
• 13:33
  Speaker 9: Right.
• 13:36
  Speaker 1: Alright. Well, so the the,
• 13:38
  Speaker 2: So strategy letter 5
• 13:39
  Speaker 1: Strategy letter 5. From Joel.
• 13:41
  Speaker 2: This is Joel. I no. So if I read it back in 2002, I have no memory of it. Do you do you I mean,
• 13:46
  Speaker 4: did you read it?
• 13:47
  Speaker 3: Do you do you remember reading it?
• 13:48
  Speaker 1: I I I definitely don't remember reading it. I read it at the time. Because I was trying to remember, like, trying to remember the the Go Go days of of 2002. This is before blogging more or less. Like, I'm not sure where I would have done it.
• 14:01
  I mean, I guess he just would have, like, put it on his website. But
• 14:05
  Speaker 2: Yeah. Yeah. That's right. But you're right. It was it was before blogging was as broadly as the things have became for sure.
• 14:13
  But this was right up your alley. And not to, like, out you as the economics nerd that you are or were, but, like, this is but I guess I just did. But this is, like, right in in your in your crosshairs.
• 14:25
  Speaker 1: Right? It is. And the I wrote a piece 2 years later that I don't part of this I don't think I read this is because I didn't refer to it in the economics of software piece that I wrote 2 years later. And I feel that, like, there are some truisms in here, but well, actually, let me ask you. What's your take on this?
• 14:44
  Speaker 2: I thought it was very thought provoking, but there were definitely some pretty big, pretty holes in the arguments. You know, in particular, identifying total cost of ownership as being a critical factor, but then ignoring that for for lots of convenient examples.
• 15:00
  Speaker 1: Yeah. And I I feel that he's got a little bit of that Microsoft hangover. I mean, you can take the boy out of Microsoft, but you can Microsoft out of the boy apparently. In the there's a little bit of that pejorative sense that he has towards open source just in general. There's also, like there are couple of lines in here that are just, like he's got a line because I think it's here's the need to like, the big thing that he misses that is, I think, very important for understanding not just software, but or not just software software, but software, is this idea that the cost of goods sold the the the cost to manufacture software is 0.
• 15:40
  And I feel that is an that's an extremely important detail. It doesn't once you've written it, it doesn't cost anything to manufacture.
• 15:48
  Speaker 2: Yeah. I I I totally agree, and I had that written down as a note where, and that the supply is infinite.
• 15:53
  Speaker 8: And it
• 15:53
  Speaker 1: is what's That's part that's right.
• 15:55
  Speaker 2: That's part and parcel of what you're saying. But if you drive up demand, your, like, your your your cogs is is the same. Your, I mean, your distribution is, you know, again, in this back in the day, back in that day, there was a printed CD and, like, a box and shrink wrapped and that kind of stuff. But but even then
• 16:13
  Speaker 1: God, that's old. The Yeah. I we so he so he had this line. It's like, debugged code is not free, whether proprietary or open source. It's like, actually it actually is.
• 16:23
  Sorry. It I mean, it is debugging code is not free. Like, if you've got code that's defective, getting that debugged is not free. But if you have code that's debugged or code that has bugs in it, that if you have software, that software actually is free from, like, a manufacturing perspective.
• 16:41
  Speaker 2: Well, I did I think that does tack into one of his great lines, which was
• 16:45
  Speaker 3: cast Well,
• 16:46
  Speaker 4: and keeping it debugged in a changing environment
• 16:49
  Speaker 1: is you
• 16:49
  Speaker 4: know, so maintenance is also not free.
• 16:51
  Speaker 9: Sure. Right? I mean, like, when he says that it's not free, I think that's what he's referring to.
• 16:58
  Speaker 1: That is definitely what he's implicitly referring to is that it this kind of ongoing maintenance of software, but I also feel that this misses something important about software. And that yes. I mean, we always think about software maintenance, but software actually doesn't need maintenance in any traditional sense. Software doesn't actually wear out.
• 17:21
  Speaker 4: And Especially if it's running on a platform that has managed to be stable. Case in point, the speech synthesizer that I use every day was last updated in 2002.
• 17:33
  Speaker 1: But it it that's a great case in point, actually, Matt, is that the you have something that basically the the things around it haven't changed. And so the hardware will long since wear out, but the software lives on. So on the one hand, yes. Like, improving software is not free. But this is like the paradox of software that's, like, super expensive to develop at some level.
• 17:57
  Speaker 8: But I think part of the problem is that any any software business model relies on ops leading software.
• 18:05
  Speaker 11: Well, and and I would I would even suggest that sometimes it's the data that that software interacts with that is obsoleting the software in terms of, like I have a Macintosh from 1996. It still boots up. It's got a copy of Netscape Navigator on it of that vintage. That copy of Navigator starts up fine, but it is essentially useless to me because it cannot make sense of today's web.
• 18:30
  Speaker 4: And, of course, my preferred speech synthesizer from 2002 isn't keeping up with changes in the language. For instance, it didn't know how to pronounce coronavirus.
• 18:39
  Speaker 1: Oh, interesting. How does it pronounce it?
• 18:45
  Speaker 4: Cor coronavirus or something
• 18:47
  Speaker 1: like that? Right. Of course. Yeah. That that make sense.
• 18:49
  Wait. Sure. But the in the the bowels of software, you have small bits of software that grow larger over time that actually the things around them don't change and that they actually are I mean, I don't know, Adam. I'm sure you've had this this a couple of times where we've had flashbacks to software we've written 20 years ago that's been untouched in the last 20 years.
• 19:14
  Speaker 2: Yeah, absolutely. I mean, in fact, I was debugging some of it today where, you know, kind of offhanded jokes I had made to the D trace source code, you know, 15, 20 years ago, kinda caught me off guard today. So yeah. And and it's just franking along.
• 19:31
  Speaker 1: And I think of that, like, that's an important way important thing to understand because this piece is all about, like, oh, I'm going to, drive down the price of my complimentary goods. And I think it kind of misses the boat in terms of open source's ability to create a larger pie and be able to if I can and he definitely misunderstands Sun in this piece. I don't know, Adam. What would you feel about his Sorry. Just a
• 20:01
  Speaker 8: and, yeah, another sort of meta point, which is, he makes reference to cathedrals and bazaars in the essay. And I don't know is it, did anyone else wonder if, you know, because, you know, Eric Raymond's essay came out in 99 So it's almost like he took the point from Cathedral in the Bazaar and, like, kind of argue the opposite.
• 20:28
  Speaker 1: Oh, yeah. And he for sure. I mean, he's a cathedralist. But so I'm I'm, like, there, I'm kinda sympathetic. I I mean, the bazaar versus cathedral is a false dichotomy.
• 20:37
  Right? But this it's a good point, I mean, in that the the cathedral versus the bazaar is much more, current in 1002 than it is. I don't to what degree do people still know about that? Have you read the is it a piece or a book? I tried, you know That's a
• 20:56
  Speaker 8: it's not
• 20:56
  Speaker 11: It was an essay that became a book collection of essays, I believe.
• 21:00
  Speaker 1: I think right. Okay. Yeah. That's what I thought. They did I I definitely feel I've seen it in both form factors.
• 21:06
  And I feel I mean, that is the that is a huge false dichotomy. I okay. Alright. So hot take, Cathedral and the Bazaar actually sets open source back. That idea.
• 21:16
  Speaker 2: Yeah. Just because it's too reductive.
• 21:17
  Speaker 1: It's too reductive. Yeah. It's it's it's and it it basically says, if you believe in in engineered deliberate software, you are a proprietary cathedralist. And similarly, if you are participating in an open source project, you are in this, like, totally unplanned, shelter bizarre. It's, like, actually not.
• 21:40
  Speaker 2: Oh, and or perhaps it was a dichotomy that was relevant at the time for helping to explain things, perhaps, per that may be overly I
• 21:49
  Speaker 11: think I agree with that. And and not only because I happen to work for a company that runs open source projects and what Cathedral and the Bazaar would call a very cathedral like fashion.
• 22:00
  Speaker 9: I I'm I would actually kinda disagree with the entire thesis there. His example of the cathedral is EMAX. I mean, that is
• 22:08
  Speaker 12: not Oh, I should be honest with that. Okay.
• 22:11
  Speaker 13: I mean, that was that was the example that was positive in the original essay.
• 22:16
  Speaker 12: Right.
• 22:16
  Speaker 13: E max and DCC representatives as examples of cathedral model where the source code was only available each time the software is released, like, there was a point release or a major release. But between releases, they were restricted to to a internal group. Whereas the Bazaar model was everything was developed in the public on the Internet. So, you know, both the Cathedral and Bazaar in that respect are open source. It's just different flavors of
• 22:42
  Speaker 7: open source.
• 22:44
  Speaker 8: I I I do believe that one of the fundamental advantages of Linux was the development model, you know, willingness to take contributions from anywhere.
• 22:54
  Speaker 1: I so I agree with you, Tom. I think that the the to me, a fundamental advantage was that it was not forkophobic, in that it encouraged people to go do experiments and was did not insist on and I realized that, like, I need to go maybe I need to go it kills me to, like, to just because Eric Raymond is Anyway, I I I how can I how can I consume this without giving me any royalties? Because I do realize that maybe I have falsified into it that it was kind of a proprietary versus open model. But that is funny that it that e max is the canonical cathedral. So fine.
• 23:40
  I believe in all the cathedrals.
• 23:42
  Speaker 11: I was wondering if maybe there was also it's been a long time since I read it, but
• 23:46
  Speaker 3: I'm wondering if maybe there was implicit contrast
• 23:48
  Speaker 11: with the way the BSDs did things where there was a group that had commit bits Yeah. And anything that happened had to go through that group.
• 23:55
  Speaker 9: That that really is is the the critical distinction in the Linux world.
• 23:59
  Speaker 4: Like And you were working with things like CVS or later subversion that didn't have the same, again, forkophilic.
• 24:07
  Speaker 9: No. That that wasn't that wasn't how much it. I mean, the the thing that really contrasted Linux versus as as the canonical bizarre was this kind of, like, we take from everything. If you talk to some of the folks who were involved in both the BSD community and the early Linux community, they'll talk about, hey. Linux was great because we didn't have to satisfy the gods of BSD before we could get something, you know, into the distribution.
• 24:33
  Speaker 1: Just don't invent a ramp and layered violation, apparently.
• 24:38
  Speaker 11: Well, there was also the whole era when there were the AC patches to the Linux kernel and there were a lot of people who were like, yeah. The the mainline Linux kernel is great, but if you wanna get anything done on modern hardware, you need the AC patches. And then later on, they got merged back into the mainline kernel.
• 24:55
  Speaker 1: Well, and I guess that that's right. What I kinda meant is that there there was this, contentment to have things live down that I think is actually very healthy. I think it's it's it's good for people to be able to go off and do, experiments and be able to go develop things on their own and be able to take the system in different directions. I would argue also personally, obviously, that because there's so much that is only 80% complete, and, in in Linux in particular. And there's a lot of things that are that would benefit from being better thought out.
• 25:26
  Speaker 4: By the way, Brian, to answer your your passing, question or comment about how you could read the essay without giving Eric Raymond any royalties. The original essay is available on his website.
• 25:39
  Speaker 1: A web but is he does that website have ads on it? So, you know, I gotta ask him to
• 25:43
  Speaker 8: call us.
• 25:44
  Speaker 9: No. We're talking about
• 25:45
  Speaker 12: Eric Raymond. His website does not have ads.
• 25:48
  Speaker 1: That that's fair. Fair fair point. Fair point. Alright. Yeah.
• 25:52
  Okay. I guess I can go safely and send it. So we we but one of the things
• 25:58
  Speaker 4: And if you don't wanna give him any traffic, I'm sure you can pull it the Internet archive.
• 26:02
  Speaker 1: Yeah. There we go. That that that's how I'll do it. I can I can figure out a way to do it? The so I I also wanna give I mean, I think Joel credit for being there's certain elements in here that I think are prophetic.
• 26:16
  I do think that that it's prophetic in terms of seeing open source as being in a company's commercial best interests. I feel like that was not something that people really that was that was, not a widely held view in 2002.
• 26:33
  Speaker 2: Yeah. Absolutely. I mean, I think that his, although in the early days, it felt
• 26:39
  Speaker 11: like before or after SCO had sued IBM?
• 26:43
  Speaker 4: Before, I think.
• 26:46
  Speaker 1: I believe that that is going to be after, but close. Right?
• 26:56
  Speaker 11: I was Like, it's post Halloween memos, but I don't remember how far post
• 27:01
  Speaker 1: They lost its file in 2003 according to the state.
• 27:05
  Speaker 4: Memo the original Halloween memo was 19.98.
• 27:12
  Speaker 1: So the Halloween memo may, that merits, I think, an explanation. This is the because the how we memo is the is the
• 27:27
  Speaker 11: memos or if it was, or if it was leakage of verbal statements that he made, but it was the kind of origin, at least in era, of the Microsoft viewpoint that open source is a cancer, the GPL is a cancer. You know, open
• 27:45
  Speaker 4: source Steve Ballmer actually came out and said that in like 2000.
• 27:49
  Speaker 1: Right. Exactly. I wanna say that.
• 27:51
  Speaker 11: Yeah. It it was Steve Ballmer that said it and it was it was it set the tone for quite some time.
• 27:59
  Speaker 1: Well and I guess and I guess that's prophetic too and that they correctly realized. I mean, this is this is a a a kind of a a moronic thing to say, but, open source is a really big deal, actually. You know, like, they they they I mean Yeah. The the the just the deeper I get into my career, the more I and especially and then I kept said this before here, but, like, going back and reading tales of software development in the nineties makes you realize how far we've come in a world of like, I mean, for example, having to pay for your compiler is insane. Right?
• 28:31
  We've we have not had to do that for a generation. And, you know, no one is ever gonna pay for a compiler ever again, more or less. And
• 28:41
  Speaker 4: Then again, sometimes you read oh, there you read comments on, like, Hacker News about how we, developers, are entitled cheapskates that aren't willing to pay for quality tools.
• 28:54
  Speaker 1: We have got the I mean, honestly, the highest quality development environment I've ever used in my career is the one I'm using right now, and it's a 100% open source. I mean, it we are getting I mean, the tooling, I feel, has just gotten better and better and better. And there are I mean, this is I I kinda hate, like, if you're on Mike Eric Raymond and now Naval. But the, you know, you just tweet about, like, the thing about open source is that problems only need to be solved once. And that is there's a lot of truth to that, actually.
• 29:24
  And that is really, really powerful.
• 29:27
  Speaker 8: The other thing, though, is that, you know, these tools are really good because they're built by people who want to use them, not by people who want to sell them. And so you satisfy yourself, and it makes it really good.
• 29:40
  Speaker 1: That is definitely true.
• 29:42
  Speaker 12: I will accept shoddy open source tooling that I can patch myself of a proprietary tooling that I can't fix when it annoys the shit out of me any day?
• 29:50
  Speaker 1: Well, this is what I you know, we point internally, you know, obviously, at Oxide, we definitely believe in open source software. But, you know, we actually really believe in high quality software. And, actually, the the proprietary software that's high quality doesn't bother me nearly as much as the as the proprietary software that's that needs to be fixed. For and then the problem is that there's very little of that proprietary software that's high quality.
• 30:14
  Speaker 2: Well, on that topic, Joel talked about was OpenOffice or StarOffice, which is a great example, I think, of the opposite where it got no traction because it was not anywhere near the the quality of the proprietary alternatives.
• 30:29
  Speaker 1: Well and I think yeah. That's and I'm glad you you mentioned that, Adam, because that kinda gets us to, like, the big thing that I think that he and, I mean, in his defense, like, no one really saw coming. The degree to which we would use software as a service to even edit our documents. Right? The fact that we use Google.
• 30:47
  I mean, I don't
• 30:48
  Speaker 2: Yes. Yeah. Yeah. I think it's okay to
• 30:50
  Speaker 1: to admit that. Right, Kai? You wanna save space? Can I talk about Google Docs here? But I I and and that I would never again install I certainly wouldn't and that I never, in my life going to install proprietary software on a desktop machine to edit a document.
• 31:04
  That's never gonna happen again.
• 31:05
  Speaker 11: It's interesting that you mentioned that in the context of SaaS though because for quite some time now, Richard Stallman and the and the FSF have been on a bit of a crusade against SaaS to the point of, as they do so often, soft, was it service as a
• 31:24
  Speaker 9: software?
• 31:25
  Speaker 4: Service as a software substitute.
• 31:27
  Speaker 11: Yeah. And so they really consider SaaS just to be another proprietary presentation of software because you don't get the source code. You can't modify it. You can't do anything
• 31:38
  Speaker 4: but And you can't even hack around it to in the ways that you might be able to if it were running locally on your machine.
• 31:47
  Speaker 1: Sure. I mean, yes. But right? I mean, the I I this is where I I think that the FSF sometimes loses the the the forest for the trees, where, the the sure. But the to me, open source is about actually creating the components more than the actual products.
• 32:07
  And it's like, yeah. With the when I'm consuming a service, I actually am consuming a product. And I, it is actually more important to me that I that I can get to those underlying components because I can actually build on those components. Give me the Lego bricks so I can go build my own thing. I actually don't need Gmail to be open source.
• 32:24
  It's personally. But I don't know. Maybe I'm maybe that's that that No.
• 32:28
  Speaker 2: No. I think I think you're right. And and when it comes for a different software, I'm good.
• 32:32
  Speaker 11: Of the people from the Free Software Foundation would go back and point out that we, the FSF, are talking about software. You're talking about open source, and that's not the same thing.
• 32:44
  Speaker 1: Yeah. I think we're getting hit a bit by the Twitter space as delays too. The, yeah. The sorry, Adam. I think you were what were you saying?
• 32:53
  Speaker 2: It's alright. I'll I'll
• 32:54
  Speaker 12: I'll step back for a second.
• 32:56
  Speaker 1: The, yeah, the distinction between well, I I I I I get where where they're coming from. I think that the the the problem is that the FSF has lost its efficacy because they've, I I think that they end up picking the the wrong fights and then ignoring the ones that are much more important. Right? Like, can we get an amicus brief in Oracle v Google placed? But the, I don't think I I wanted to to kinda ask your take on it.
• 33:24
  Did you see have you read any of the strategy letters, by the way?
• 33:27
  Speaker 2: I read some as a result of this one, but I I yeah. No. If I had, I didn't remember the previous one.
• 33:33
  Speaker 1: So how about you? I had a well, I mean, obviously, I remember reading his, like, 20 questions to ask your future employer or whatever. Do you the what is this?
• 33:42
  Speaker 2: I don't remember that one.
• 33:43
  Speaker 1: Oh, no. No. There's, like,
• 33:43
  Speaker 4: a test. The 12 point Joel test.
• 33:46
  Speaker 1: Yes. The Joel test. Thank you. Yeah. This is like, do they use source code control?
• 33:51
  Do they, I mean, I think this is arguably the the the most famous thing he's ever written. The and I it got now I what is the last time you've read the I don't know if you've read the Jolt test recently. So you've you surely have seen the Jolt test, Adam. If you saw it, you'd recognize it.
• 34:09
  Speaker 2: I'll go look for it now.
• 34:10
  Speaker 1: Yeah. The
• 34:12
  Speaker 4: I can't rattle off all 12 from memory, but I I have read it.
• 34:18
  Speaker 1: And, you know, I mean, it's like some of it is like, do you have the best tools that money can buy? You know, do
• 34:24
  Speaker 2: So guess $0.
• 34:25
  Speaker 1: $0. Exactly. The so I think I know remember doing that, but I had not read his other kind of strategy letters. The he's got a piece that, definitely believes that Gmail is not the future model circa 2,007, which is interesting. Because I'm like, I I don't know that I I I don't don't know that I necessarily would've disagreed with that in 2,007, but it's clearly
• 34:48
  Speaker 9: Well, whatever what I recall from 2007. So that was fun.
• 34:54
  Speaker 4: If you're talking about strategy letter 6 from 2,007, what I recall from that one was that he thought that the current generation of web applications like Gmail were kind of akin to, the DOS generation of PC apps in that they, they couldn't very, that they couldn't interoperate with each other very well. And he thought that, he thought that something akin to windows was going to come along and be implemented on top of the web platform to allow these, these, and of course this being 2,007 ajax or web 2.0 or whatever you wanna call it was still pretty hot. And he he was he was trying to predict where, where that generation of web apps was gonna go. And I, I think, I think he was wrong because, and, and well, I I think what really happened was was that, the, you know, web apps got upstaged by, by mobile, native apps. Although not I mean, they're that's that's not to say that web apps are irrelevant, that that that that's also a stupid false dichotomy.
• 36:11
  But that that's what I recall from that letter.
• 36:14
  Speaker 11: Wasn't that the era when the Chromebook really came into being as a concept? Because 2000 Chromebook was 2,010,011. But I remember back in, like, 2,000 7 ish, people were saying the web browser is going to be the new operating system. Everything Yeah. Built on the web browser is the platform.
• 36:32
  Speaker 4: Oh, Netscape is predicting that as far back as the late nineties.
• 36:36
  Speaker 12: Why does everything have to be the operating system of of something?
• 36:39
  Speaker 7: Well, Emax was the only one.
• 36:41
  Speaker 9: It it
• 36:43
  Speaker 1: I agree, Josh. Like, you think
• 36:45
  Speaker 12: Like, x Xfinity key Xfinity who I'm not gonna say too many bad things about, but but certainly like the, it's not this modem you sold me is not the operating system of my home.
• 36:59
  Speaker 1: We Don't call it.
• 37:00
  Speaker 12: Don't call it that.
• 37:00
  Speaker 1: I did. Well, if you want the the example, the name of Ultra of this is WeWork. WeWork was making an operating system. For buildings? The the the yes.
• 37:08
  They view we are the Particularly though.
• 37:09
  Speaker 6: We are the
• 37:10
  Speaker 12: Not a building management system, which is a thing that exists already.
• 37:13
  Speaker 1: Right. No. Not a building management system. That would be very pedestrian. We are building an operating system for the way you live, Josh.
• 37:18
  Speaker 8: So Everybody wants to be a platform. Right? So operating system is just a way to say that.
• 37:23
  Speaker 1: Everybody wants to be a platform. Look, Tom, why is that? Why does everyone wanna be a platform?
• 37:27
  Speaker 12: At least it's an ethos, I guess.
• 37:29
  Speaker 8: Yeah. There's a lot of lock in and, you know, being the center of the ecosystem.
• 37:34
  Speaker 1: And so that's my read too is that people want to be a platform for purely out of their own capitalistic rapaciousness. That it's there's not, like, a better reason than that. Like, I Well, and
• 37:46
  Speaker 12: and the email thing, like, talking about looking back then from back then, g the the thing that happened with email was less about, the success of this one web application and more about how a few large actors managed to make it almost impossible to run your own email. Actually, like the the, the structure of email as a federated system is extremely difficult to participate in unless you are already Google.
• 38:14
  Speaker 4: And I think that was a casualty of the spam arms race.
• 38:17
  Speaker 9: Yeah. That was spam. That was a fair number. Sure.
• 38:20
  Speaker 7: Sure. Gmail.
• 38:21
  Speaker 12: But between between Google and Microsoft with outlook.com, it is effectively and has been for quite some time structurally, basically impossible to ensure deliverability unless you are one of those people like organ.
• 38:34
  Speaker 2: Alright. And Brian, you were you were asking about the sun, shout out in this in this, blog post, which is part of the reason I thought you were you were kinda kicking it my way on on Twitter the other day.
• 38:46
  Speaker 1: I was. I accidentally muted everyone a second ago. Sorry about that. I didn't mean to mute everybody. Sorry, Adam.
• 38:49
  Go ahead.
• 38:50
  Speaker 2: No. No problem.
• 38:51
  Speaker 12: Sorry. I can I can tell when you want me to stop?
• 38:53
  Speaker 1: Exactly. I just I just
• 38:56
  Speaker 2: But, the like, what was your take having lived through, you know, from what when did you join Sun? 96 through 2009. Did you what did you think of Joel's take on on Sun and open source?
• 39:11
  Speaker 1: Well, obviously, a gross caricature. I mean, the the viewing Sun I mean, he's, I think, genuinely perplexed by Sun's motivations. Because, like, Sun's a hardware company. Why are they pushing Java? Oh, they're pushing Java because of their hatred.
• 39:24
  It's an emotional self destructive act out of their emotional hatred of Microsoft. And he's like, come the fuck on, dude. Like, we are I mean, come on. That is just no. The I I do feel that a core belief that I think Sun was not the only company that had this belief, but a core belief of Sun was about making the pie larger, that it was in everybody's interest to make the pie larger.
• 39:50
  And to me and, Tom, I would love to actually get your perspective on this because you lived it. But to me, truly one of the most iconoclastic things that Sun did was porting NFS to their rivals platforms, porting it to Alteryx and to Irix back in the mid eighties. Right, Tom? Late eighties, early nineties?
• 40:11
  Speaker 8: Oh, yeah. Mid mid to late eighties. It was on everything that moved. But And
• 40:16
  Speaker 9: Sun didn't do those ports. They just released the source code as the reference implementations and hosted these, like, interoperability tests or something.
• 40:25
  Speaker 8: But we we we had actually a substantial consulting group that that would do a lot of the work.
• 40:30
  Speaker 1: Yeah. It's Sun did that's right. That was my understanding. Yeah. Yeah.
• 40:33
  Tom, you could speak to it. But it's like it definitely happened because Sun was actively encouraging it and putting resources behind it. And it was a Pretty different
• 40:41
  Speaker 12: pretty different to what Microsoft then did with SMB. A little different.
• 40:45
  Speaker 2: Different. A little bit.
• 40:46
  Speaker 1: A little bit.
• 40:47
  Speaker 8: But, you know, the Sun and Microsoft were in talks for a while to converge even even back very early on. And we we looked at the SMB protocol and all that stuff about printers. It was like, well, this will never fit in in UNIX.
• 41:03
  Speaker 1: Yeah. And the op locking. Oh, the op locking. Which is it it feels like it's not necessarily a bad idea. It's just fraught with peril.
• 41:11
  Speaker 8: Well, locking and networking don't get along at all.
• 41:15
  Speaker 1: They, yes. Words to live by. Locking and networking definitely don't get along. So I felt like that piece, Adam, that part of it was, I thought, pretty misplaced because I think it, what actually, maybe, Tom so taking that NFS as an example, what is going through the mind of Sun as they are pushing NFS? To because it's like maybe I maybe I am retrofitting my own view on a historical sun, but my view was the belief in open systems was if we can get everybody collaborating on this protocol, we can get more people using networking computing.
• 41:46
  And if we get more people to use network computing, that's a bigger pie that we can go compete for.
• 41:51
  Speaker 8: Right. But but it's also you know, one one of Joel's columns is all about switching costs. I forget which one. But, you know, it it made it easy to insert sun into some other environment, you know, that was dominated by mini computers or whatever. If you if you could have NFS and then, you know, just lots more walls came down.
• 42:13
  Speaker 1: Yeah. This is something that I remember that that, McKinley had a piece talking about the the need for an off ramp. That was important for technology to have an off ramp, which I actually I mean, I think it's important for a I mean, this this is what people that use technology don't want to be locked in. We don't wanna be locked in. Right?
• 42:32
  I don't wanna be locked in. You don't wanna be locked in. When we, as technologists, make technology based choices, we are we don't wanna be locked in. We wanna know, at least in the abstract, that we can go take ourselves elsewhere relatively easily.
• 42:48
  Speaker 8: Yeah. But but similarly, you know, you'd you'd have companies that say, oh, we're an IBM shop. And Sun could turn around and say, well, look. We have NFS for the mainframe. You should try this out.
• 42:58
  So, of course, they'd have to buy a few workstations to try it out.
• 43:02
  Speaker 2: Well, I think that was, you know, Joel says that, you know, Java and write once, run anywhere didn't make sense. But I think through that lens, it was it was lowering the the barriers to that to get that IBM shop. And you do that when you're in a point of strength when your hardware is better than other folks.
• 43:20
  Speaker 8: Yeah. And and Java succeeded at becoming a platform. Sun failed at monetizing it. So that that's the crux of Sun and software is that Sun can never monetize software. It was always hard.
• 43:34
  Speaker 12: Microsoft has a pretty
• 43:35
  Speaker 4: I wonder if I wonder though if that was because they gave too much away. And and and maybe they maybe they were pressured to give too much away because we've been conditioned by open source to expect our development tools for free.
• 43:51
  Speaker 8: Well, they they kind of well, the other thing is they they never had they never had anything above the platform.
• 43:57
  Speaker 9: Yeah.
• 43:57
  Speaker 8: And this is where Apple shines. Apple has all kinds of apps and its appliance and blah blah blah.
• 44:04
  Speaker 9: Yeah.
• 44:05
  Speaker 8: Where was always, look at this great platform, and but we're done.
• 44:09
  Speaker 9: Adan was all about making standards but not providing reference implementations, and that was an issue in the early Java days. I remember being very enthusiastic about Java the first time I saw it. I was like, this is pretty cool. You know, I mean, compared to, like, c plus plus, it was plus plus circle, like, 1997 or something. I mean, it was a pain with Deepgram.
• 44:28
  And, you know, then they were saying, like, hey. We're gonna have, like, Java spaces and, you know, application servers and stuff. You're like, cool. Where do I get these things? It's like, go talk to WebLogic.
• 44:37
  They'll sell you one for half half a $1,000,000. Like, no. I'm not gonna do that.
• 44:42
  Speaker 8: Right. Right.
• 44:44
  Speaker 11: Well, I remember looking at some Java apps back in the day when people were, you know, actually intending to build Java desktop applications. And the only thing you could really get away with was to build them with what was it? AWT or whatever? The I guess the advanced windowing toolkit and it's Abstract windowing
• 45:02
  Speaker 4: toolkit. Although or as my high school programming teacher, in 97 called it the awful windowing toolkit.
• 45:09
  Speaker 11: Yeah. It looked like TWM from 1986, you know So, like, aesthetically, if you if your experience of Java was apps that looked like that, you were like, now something else.
• 45:21
  Speaker 1: Hey. Just a quick mechanical note. I know we got more folks requesting to speak than we've got. We could 4 g two hundred spaces only allows us to have a finite number of speakers. It's it's set at 12 right now.
• 45:31
  So we're at our cap. So just to, if we can get more folks just allow us to get more folks in there. Little little PSA. Sorry. I didn't mean to I didn't mean to to cut anyone off.
• 45:47
  But the
• 45:47
  Speaker 4: is there a way to relinquish my slot as a speaker, short of leaving the space and coming back in?
• 45:56
  Speaker 1: Yes. I think I don't know if you can relinquish it. I can I can I can relinquish you?
• 46:03
  Speaker 12: In the in the 3 dot menu at the top, there is a switch to listening thing.
• 46:07
  Speaker 1: Oh, there you go. But not everyone all at once. So sorry. The the the yeah. Folks that have been been, requesting
• 46:18
  Speaker 7: that. Microsoft Microsoft have been
• 46:20
  Speaker 12: pretty idle minded on this stuff for a long time though. Like, I mean, for for a very long time, dot net was called cross platform, and by which I think they meant, I don't know, both kinds of windows. Like it like NT and the other one. I don't know. It's hard to say.
• 46:38
  Speaker 9: Well, I think they meant hardware platforms. Like, their their whole deal was IL will run on alpha and activity 6 and, you know Yes. Mix for
• 46:46
  Speaker 1: all that. Right. But all by cross platform, I mean, all the platforms on which we run, of course.
• 46:53
  Speaker 12: Right. But they would say those things with a straight face. And people would like, so can I run it on UNIX? And it's like, what is that?
• 47:00
  Speaker 1: That's not that that that's actually not what we meant. Exactly.
• 47:02
  Speaker 8: Yeah. Then they they managed to convince most of the world that anything from Microsoft was open systems. Right?
• 47:09
  Speaker 1: Yeah. I mean, do they manage to give us a little bit? I don't know if they if they
• 47:13
  Speaker 8: Well, not
• 47:14
  Speaker 1: did or not.
• 47:14
  Speaker 8: Not as tech guys, but an awful lot of the world thought, oh, PC equals open.
• 47:21
  Speaker 1: That is true that is true that they thought that, that, which is kind of paradoxical because I don't necessarily anyway. Yeah. This I think you're right, Tom. That they definitely think that. I
• 47:30
  Speaker 7: mean Of
• 47:31
  Speaker 9: course, to the mainframe world, anything non mainframe was open systems.
• 47:35
  Speaker 7: Right.
• 47:36
  Speaker 1: It that's also true. Right? Everything was was open by comparison. So I think this this question about the about Sun and and the that, I guess, the monetization of Java is interesting. I I for whatever it's worth, Arthur Van Hoff just take this as a statement of fact.
• 47:51
  Arthur Van Hoff told me in in, what, in 1997 that every e ten k that was sold was sold because of Java. Now I obviously disagreed
• 48:00
  Speaker 7: with this. And in fact, I
• 48:02
  Speaker 1: may have disagreed with this in a relatively colorful way that may have alienated the rest of the people at the dinner that I was at. But, that was definitely, his perspective. Was that sorry, Jason. Go ahead.
• 48:17
  Speaker 9: Well, I could see from part of it because the early versions of Java and Solaris were horrible. They were terrible. If you try to do IO, it would issue a system call, a read or write for every single byte.
• 48:31
  Speaker 1: Well yeah. And I think that the that I I do feel that, like, the the the failure is the failure to kinda connect all the dots. I feel that, like, you've got the, you've got the potential to build a great system, one that's open, one that customers believe in. But you do need to connect all the dots, and you need to actually there's a lot of execution involved to get all that working.
• 48:52
  Speaker 8: I mean, the line I think Sun was slow slow to get Java going on the server side. They spent way too much time on mobile phones and stuff like that.
• 49:00
  Speaker 1: Yeah. Yes. Amen.
• 49:02
  Speaker 2: The the line that definitely resonated was what he described sun as a loose cannon. And, that that sounded right. And and it also made me think
• 49:11
  Speaker 11: I think they really over promised with it as well because I remember how early it became a joke that Java was write once, debug everywhere.
• 49:22
  Speaker 8: Yeah.
• 49:23
  Speaker 12: Yeah. But, hopefully, that's just software, though. Right? I mean
• 49:28
  Speaker 11: Yeah. But they had actually explicitly made the promise that because Java was bytecode compiled, you could write your Java app and it would run everywhere on every architecture. And, you know, it was gonna be this magical new world of cross platform software written compiled the byte code and it just it didn't it didn't pan out. And if they hadn't promised it, no one would have expected it. But they did, and people did, and then it was disappointing.
• 49:54
  Speaker 9: What's the directory separator on MVS?
• 49:59
  Speaker 8: Who says they have directories?
• 50:00
  Speaker 1: Exactly.
• 50:02
  Speaker 11: Anybody who work with Stratus BOS?
• 50:06
  Speaker 8: No.
• 50:08
  Speaker 1: Okay. This I
• 50:09
  Speaker 13: I mean, a lot I mean, a lot of the challenges that people ran into with Java cross platform were not differences. It was because of, web application server differences, like WebSphere versus Tomcat was the, you know, common thing. And then the other thing that they hit was often, like, cross database support where you'd write an application, that was meant to support, like, Oracle and Postgres and MySQL. And, of course, such undertaking is more difficult in practice than than just write once run anyway.
• 50:42
  Speaker 12: On the on the other hand, you you can run Jira on FreeBSD or Lumos, and that's never been built for those platforms. So,
• 50:52
  Speaker 1: like Yeah.
• 50:52
  Speaker 13: Exactly. That's insane. From from the Jira from the Jira and Confluence perspective I'm talking here, the challenges with cross OS development and deployment were very rarely due to, operating system differences and were way more often due to web application server and database differences. Those were the things that bit Jira and Confluence, early on. The only, kinda caveat to that would be that there is differences in performance for operating systems when you're running something like Bitbucket server, where, shelling out to Git can cause very different performance under Windows versus other platforms, and that has been a cause of problem.
• 51:39
  Speaker 7: Hey. Can I this is James Todd? I actually worked on the Tomcat team back at Sun. Oh. Sun a long time.
• 51:45
  Just kinda curious about folks' comments about reference applications not being made available. Admittedly, Sun didn't, leverage everything's Java, but I'm still doing Java to this very day. I totally love it. I think Ant and Tomcat kinda changed the world quite a bit personally. I mean, not trying to pat my back or anything, but how did Sun not make reference implementations readily available?
• 52:06
  Speaker 9: It wasn't that they didn't make them readily available. It's that, like, at the time they were not as mature as the alternatives. And Oh, yeah. And Sun, like, I remember going to kind of ironically, it was probably early 2,001, like, January or something. Going to the Sun office in the World Trade Center and to see a presentation about Java.
• 52:31
  And, you know, basically the sales engineers there were like, look, we're not interested in producing something that you're gonna run your production applications on. Like, we're engineers first.
• 52:41
  Speaker 1: I
• 52:41
  Speaker 7: agree. Actually, I I think it it's shiny and awesome. Frankly, changed the world. I was about ready to read how to do, you know, Borland and X Windows and all sorts of stuff. So I think Java, like, liberated a ton of things.
• 52:52
  Frank, I'm really fascinated with the past. I love the oxide threads and conversations. But the fit and finish, I mean, Sun should have owned a number of businesses like Veritas, EMC, etcetera. Amen. Amen.
• 53:03
  Right. Right.
• 53:03
  Speaker 8: Totally. Totally.
• 53:05
  Speaker 1: I mean, and and, James, this is how I mean, that was the deck on which we got them to basically fund us for Fishworks. Right? Is that we Sun had invented NFS. And when we started Fishworks in 2006, we had 0.1% share of the NAS Eiler space. It's like, that's embarrassing.
• 53:25
  And I always felt like the problem was just not quite getting to that. And, I mean, Sun had a hard time putting these incredible pieces together into a fit and finished product.
• 53:38
  Speaker 8: Yeah. And Yeah. There there's a related one, near and dear to my heart from the late eighties. Sun Sun was the number 3 router company.
• 53:47
  Speaker 1: Who would Right.
• 53:49
  Speaker 8: Nobody knew that we made routers.
• 53:51
  Speaker 7: Yeah. Tom, your your retrospective and stuff is just fascinating. I mean, I worked on the Joxa team for a while, by the way, which I think was Genie next week. We should have just got our shit together. Right?
• 54:00
  But anyway, it just felt like we were kind of in a little bit of a bubble celebrating things. And then people were coming along with us, but then we didn't go with them. I'm not sure what that gap was, but I'm not sure. But, anyway, I'll step down. Thanks for letting me speak again.
• 54:14
  This is awesome.
• 54:15
  Speaker 1: Yeah. James, thanks for the comments. I I I the end man, Joxta and JD, you were I I remember at some point we had I'm doing we were drinking coffee out of, like, Juxta cups. Do you remember this?
• 54:26
  Speaker 9: Juxta is totally I
• 54:28
  Speaker 7: I mean, is, you know, being a peer to peer sort of guy and loving decentralization, I
• 54:33
  Speaker 8: wish a hell of a
• 54:34
  Speaker 7: lot more Facebook, for example, decentralized. You know? The technology Sun had in the day 1 and continue to build and evolve were spot on. Clearly. Right?
• 54:45
  Like in the IDE that, you know, net beans, again, very Frankenstein y to your earlier point, just amazing. I actually had Jack's to work in within net beans at one point. Probably nobody knows that I did that right before I left. But any anyway, it had all the gems and jewels and frankly, reference implementations are readily made available and the world's kind of moved on without Sun. But within Sun, I'm not sure what it was.
• 55:07
  I don't think it was engineer infighting. I don't think it was marketing and product conflicting with each other, but the execution was clearly lacking on many fronts.
• 55:17
  Speaker 8: Well, we we didn't really have any end user DNA. We were we were all about, you know, being developers and helping developers.
• 55:24
  Speaker 7: Tell me your comment about Apple, that just resonates with me so much. I mean, you know, I I actually I started build maps kind of looking at what Apple was doing as far as why. Meaning, they hide all the details. Right? So but yeah.
• 55:38
  Speaker 4: On on the topic of of, not having any end user DNA, signed specifically.
• 55:44
  Speaker 1: Let's just go.
• 55:45
  Speaker 4: You guys no. You guys were you guys were talking about, AWT, GUI earlier. And Joel Spolsky in one of his other, one of his more rambling articles, I think, called, lord lord Palmerston on programming. He was he was talking about various cross platform GUI options, and and he he compared Sun's, GUI implementations in Java, AWT and swing to, star Trek aliens who had been watching humanity through a telescope and knew what human food was supposed to look like, but didn't know what it was supposed to taste like. And and his his point being that they never really got the gooey got their gooey's to feel like like a native GUI should.
• 56:40
  Speaker 7: Yeah. I don't I don't differ with that. The thing is that's a server company just going that far far and trying to get there. Right? I mean, AWS is terrific.
• 56:47
  So
• 56:48
  Speaker 8: Well, it Reminds me of the plot of Galaxy Quest.
• 56:51
  Speaker 1: Right? Well, actually, what's funny is that that exact and that almost to the word analogy was used to me to explain the source code of AIX that someone had been one of our customers had had a source license to AIX, and they described AIX as Unix as if written by a an alien species from another planet given POSIX but no other context.
• 57:16
  Speaker 9: I've never heard a
• 57:17
  Speaker 13: I mean UNIX?
• 57:19
  Speaker 1: Right. Exactly.
• 57:20
  Speaker 13: That cross cross cross OS, GUI is inherently an extremely difficult problem to to solve where if you're trying to write once and make it run-in, you know, on Mac and Windows and to look native and feel native to both of those platforms. That's still not really a solved problem today, in my opinion. And you can see that in the, like, in the proliferation of of Electron as a solution to that problem where they can run around it.
• 57:47
  Speaker 4: Mozilla's Zuul got pretty close back in the day, at at least at least on Windows and and desktop Linux. I I can't vouch for how how good it might have been on Mac, but it was certainly better to do it was certainly possible to do better than swing.
• 58:08
  Speaker 13: It was definitely it was still extremely difficult to get right, even if you're native to that to that development environment, and Xero was extremely difficult to work in. I imagine most people would read the initial documentation bulk and walk away and try and do something a little bit easier, in my opinion.
• 58:28
  Speaker 1: But, Ian, hasn't the browser become that effectively? I mean, isn't the browser that
• 58:32
  Speaker 13: Yeah. Exactly. That's what I'm saying is, like, the the the browser and by extension Electron has been the the the, like, the facto solution over the past, I don't know, 4 years or so, which, like, in my opinion, also just a bit of a an end run around the problem where it's like, well, we can't solve this problem. Let's just do a window, and then inside that is another browser, browser and then you can do whatever the hell you want inside of that. And it'll look like a web app.
• 59:00
  It won't look like a name.
• 59:01
  Speaker 12: The the success
• 59:02
  Speaker 13: It'll look like a web app every week.
• 59:04
  Speaker 12: The the success of that approach, though, highlights the fact that people actually don't give, like, 2 hoots about native controls for the most part. Like, so all the stuff about, like, Java doesn't look right, nor does any Electron application. They all look like the Electron application. They don't look like the native platform very much at all.
• 59:22
  Speaker 7: I kinda got excited about, I think it was JSF, Java server faces, which the fall, and, frankly, I stepped away from UI stuff at at that point beyond Swayn. I mean, anybody take a look at that at all? But, again, I I totally went to big data.
• 59:37
  Speaker 4: Oh, that that that's a server side web framework. Right?
• 59:39
  Speaker 7: Yeah. JSF.
• 59:40
  Speaker 4: I got the declared. The thing the thing that I got excited about it at one point was, the SWT toolkit from IBM, which kind of takes the the, wx widgets approach of just of of mostly doing rappers on top of the actual platform. Well, it's kind of like AWT done better. I I guess it's really
• 1:00:01
  Speaker 7: what it is. AWT being, like, alpha and then along comes, you know, beta. And then at least at least it gets started. And I think there is something out there today. 1 really wanna look into it.
• 1:00:10
  So If you're right, mom, we can actually relinquish my speaker, if others need it. I again, thoroughly enjoyed this. And thanks for hosting all this. This is awesome.
• 1:00:19
  Speaker 1: Oh, you the the JavaScript's you got. Yeah. No. You're all good. The the so, the but how's it like the the browser having kind of become the the the de facto way that we we interface effectively graphically with these applications, I feel like the the substrate on which we build is virtually all open source.
• 1:00:37
  Is is there a proprietary what is the proprietary alternative to React or Angular or what have you. Is there one? I can't imagine there I mean, haven't they all died? Isn't it all open source? This is That's fair.
• 1:00:51
  That'd be
• 1:00:52
  Speaker 9: This is open source.
• 1:00:53
  Speaker 1: But, you know, are the maybe on these mobile ecosystems. But it feels to me that, like, all that stuff is open source, which is actually a pretty big step in the right
• 1:01:01
  Speaker 9: direction. Probably the only real analogs are native and native platform stuff like Microsoft Foundation classes or whatever. That's probably been up in source at this point.
• 1:01:11
  Speaker 4: Oh, and then Microsoft Microsoft keeps trying to I mean, they they they've had, like, 4 or 5 different GUI toolkits now that are supposed to be the native way of doing Windows apps. First, you had Win 32, then you had MFC, then Windows forms, then WPF, universal Windows platform. And now when UI is supposed to be the new new thing. And and and most developers are are ignoring most third party developers seem to be ignoring what Microsoft is doing and saying either we'll we'll keep going with our old win 30 2 code or we'll use Electron.
• 1:01:53
  Speaker 12: I I ironically ironically, like
• 1:01:55
  Speaker 4: Hell, even some teams in Microsoft use Electron.
• 1:01:58
  Speaker 12: Ironically, WPF Microsoft Teams. WPF from a programming perspective, like, at least in its original implementation, basically was like swing that they were gonna pretend was not swing. Something something lawsuits, but like
• 1:02:10
  Speaker 4: It was swing that it was swing that required for its time, a a relatively high end GPU.
• 1:02:17
  Speaker 9: And the I the whole idea of native applications is kinda going away to some extent because Right. You know, outside of highly, highly specialized, very vertical things, it's just not necessary anymore. Like, who cares? You know, like run it in a web browser. It doesn't matter.
• 1:02:33
  I I you know, I can take a performance hit on my absurdly powerful, you know, desktop machine, which itself is kind of going away because I can do most of the stuff. Like, right now, I'm I'm doing this on my phone. Right? I can't even do this on my desktop. That's why I wanted to.
• 1:02:49
  Speaker 8: Whoever thought the Internet would be fast enough for Google Docs?
• 1:02:54
  Speaker 4: And and as far as accessibility is concerned, just because, I'm I'm sure someone is wondering when I think about this, It's it's probably I I would say it's generally easier to get accessibility right in a web app than in a native app, at least on Windows.
• 1:03:14
  Speaker 9: Well, I would hope so. I mean, you know, if nothing else, like, you have structured markup that the application could be sending to the end user. Whether they do or not, I think is, you know, maybe not so much anymore, but that's the theory anyway. I I did wanna make there is an interesting analogy to be made when we talked about sort of Sun and the Java days back in the, you know, sort of late nineties, early 2000, and the idea of reference implementations. Maybe the analogy is more that what the way that I my impression of the way that Sun was looking at Java was the way that the x 11 people looked at x and, like, the Athena toolkit.
• 1:03:49
  They were like, look, we're gonna give you, you know, here's a bunch of standards. We're not gonna, like, you know, we're gonna give you mechanism, not policy. That was always kind of the mantra in the x eleven world. And it kinda felt like that's what the Sun folks were doing with Java. They were very interested in the standards and owning those.
• 1:04:06
  They were not so interested in, like, the actual baked in software ecosystem.
• 1:04:11
  Speaker 1: Yeah. I also just I
• 1:04:12
  Speaker 8: didn't remember that mantra? Yeah.
• 1:04:15
  Speaker 1: I feel we just did our did as a as a a company. We did just a very poor job of going and connecting at all to actually deliver something of value to the end user. I mean, Dan, I just had my head in my hands when you're describing. Yeah. First of all, you I the the WTC office, I I mean, I was in that office.
• 1:04:32
  Would have been the same time you were, basically, in early 2001. Thankfully, no one was in that office on the morning of 9:11, and the only son employee lost was actually on one of the one of the aircraft. But the, just I can just feel like I'm in the room with you when they're saying, like, oh, no. No. This is not for production apps.
• 1:04:51
  This is not for like, what are we doing here? Are, like, are we talking about you trying to sell me a like, aren't you at the end trying to make a product that I am going to pay money for? And I feel that, like, a lot of things that we were doing were not didn't make that impossible. In fact, it would have been easier, but it did require us to connect more dots than we we tend to connect.
• 1:05:11
  Speaker 2: Well, Son is running the X server. Right?
• 1:05:13
  Speaker 9: I mean, I I I got a wonderful X server from Son, but I but I didn't get Motif. You know?
• 1:05:19
  Speaker 1: You're right.
• 1:05:21
  Speaker 2: Well, this is what I feel like was the biggest gap in this essay was, this notion of, of our rational actors in this environment. And, my favorite one of my favorite anecdotes from Sun Along These Lines is, Jonathan Schwartz stood up in front of the audience, pointed to OpenSolaris downloads
• 1:05:40
  Speaker 13: and the
• 1:05:41
  Speaker 2: number, and said, we know where each of those is from, and each one of those is a potential buyer of the ZFS appliance because they're all downloading open Solaris for ZFS. And I just thought the the sequence, the, like, mental gymnastics to get from one place to the other made no sense. In other words, there there might have looked like there was an open source strategy, but that is very generic.
• 1:06:07
  Speaker 1: Well and so yeah. I mean, this is very on point, Adam, because I I mean, I believed in the abstract in that strategy, but then there was no real execution on it. And in particular, there was a company that had adopted effectively everything that Sun had put out in terms of software and wanted to also buy Sun hardware and to run all the Sun software on. And they couldn't get the hardware to show up. And do you know where I'm going with this, Adam?
• 1:06:39
  No. Oh, so this is that company wrote a blog entry called The Sun Does Not Shine On Me, and that's Jason Hoffman at Joyant. And in The Sun Does Not Shine On Me, he Jason describes all and we definitely have to get the web archive to get this one. This is long before I was at the company. But describes how enthusiastic they were to adopt all this on software and how heartbreaking it was that he couldn't make hardware.
• 1:07:01
  He's like and actually, like, someone from Dell actually saw this and immediately made gear show up and, solve all my problems for me from a go to market perspective. And I I think you know who that person was from Dell, Adam.
• 1:07:19
  Speaker 2: Oh, really? Was that Steve?
• 1:07:20
  Speaker 1: It's Steve.
• 1:07:21
  Speaker 4: That's awesome.
• 1:07:22
  Speaker 2: So idea. That's amazing.
• 1:07:23
  Speaker 1: Yeah. So it's so our boss, my cofounder, Oxide CEO, Steve Tock, was the that was the person at Dell who and Steve's got terrific stories about the inside of making that happen where he basically you know, in classic, like, scrappy go to market fashion, diverted a box that was headed towards another customer, apologized to them, diverted it to to Jason, and got a huge win at at at Joiant. And, ultimately, he's I kinda don't think Steve is here. If he is gonna please go volunteer. But the he couldn't get Dell to take Facebook seriously as a customer and left Dell because, Dell did not Facebook as a relevant Dell customer in, sort of in, like, 2,009.
• 1:08:09
  Speaker 2: So so open source successful in breaking down those barriers, but then, you know, then it down to execution.
• 1:08:15
  Speaker 1: That was it.
• 1:08:16
  Speaker 2: Broken down the barrier.
• 1:08:17
  Speaker 1: That was it to me. It's just like and actually, like, it there's a certain degree to which that was the moment we were doomed. Because the all of the strategy in the world cannot make up for poor execution. And I felt like we had there are a bunch of strategic things that have been done correctly. And now is the time to, like, hey.
• 1:08:37
  Good news, everybody. All we need to do is take the purchase order and make the gear show up and and support that customer and stand by them, and we couldn't do it. Like, we
• 1:08:46
  Speaker 8: Oh, I I heard I heard that Sun totally lost Wall Street after 2,001 because of failure to execute in the aftermath of the attack.
• 1:08:56
  Speaker 1: So that's a what And so
• 1:08:58
  Speaker 8: They they just could not get shipments together in a hurry.
• 1:09:01
  Speaker 1: Well, so there were a bunch of things. And so alright. So the I do have to tell this story. I'd be so, when I spent a lot of time going back and forth to New York in, from 2001 to 2006. And there were a bunch of things.
• 1:09:15
  I don't know how much of that was 911 after my phone. There's a at that point now, Linux on x86 is, like, indisputably performing quite a bit better than Solaris on Spark, and there are a bunch of other issues that are going on. And Sun is kind of losing the plot at the same time where we've got a lot of exciting things happening with the CFS and DTrace and so on. And then Andy Bechtelsheim and Kalea and the x86 servers, and there's a whole bunch of reason for optimism. And but at the same time, like, huge struggles.
• 1:09:42
  And I'm going out to New York a lot. And I, I at one point, the one of the reps was just like, hey. I just wanna thank you for, you know, you I know you're out here a lot, and I appreciate it, and we appreciate it. And it's great to have you out here with the accounts, with our customers. I'm like, you know, this is a problem.
• 1:09:57
  Of course. I mean, of course, I'm only out here a fraction of the time that Jonathan is out here. And he's like, yeah. No. Actually, Jonathan has never been out here.
• 1:10:06
  I was like, what? And I Wow. So, the and Jonathan would not travel. And I I will I'll leave it at that. Jonathan refused to travel.
• 1:10:22
  And I don't think you can be a CEO of a global 2,000 company and refuse the trial. Because I think this is, like, just, you know, life advice. When you have customers, you need to go visit the customer where the customer is, and you need to go I mean, Adam, you and I did this a lot going in, like
• 1:10:43
  Speaker 12: a lot. Yeah.
• 1:10:44
  Speaker 1: And I know you did this a ton at at Delphix as well in terms of, like, actually going out to the customer site, being with them, seeing their problems through their eyes. I mean, I mean, like I said, you did a ton of that. Like I said, I post Sun as well.
• 1:11:01
  Speaker 2: Yeah. Yeah. No. In in invaluable in getting getting the real, I don't know, the the real dirt and hearing the real anger from customer.
• 1:11:07
  Speaker 1: Well, because in in like, an executive briefing center is a home game. And if you only play home games, you are actually only seeing those customers that are already, like, your big advocates advocates and so on. You're not actually seeing so, Tom, during those kinda years, I was I don't know. Adam was with me on a couple of these occasions. We go to customers that were just like, oh my god.
• 1:11:26
  They were so hostile because they weren't already, like, kick Sun out. And you're thinking, like, I just need to, like I've got, like, one minute, and in that one minute, I need to buy 2 minutes. And in the 2 minutes, I'll buy 5 minutes. And that was always with, like, a DTrace demo. And I always loved, like, demoing DTrace in those days.
• 1:11:42
  And, Adam, I know you had a lot of these experiences too where, like, there's this predisposition of being, like, I really wanna kick you out, and you're making it hard because this is actually sounds kind of interesting. And then you get to the demo, and they're like, I just remember being with the Chicago Mercantile Exchange in particular, and we are standing in the hallway outside like, they won't let us into their office, and they all have their arms crossed. They basically being like, we're about to call security. And I'm kind of, like, explaining d trace. And finally, the guy has this, like, huge sighs.
• 1:12:10
  Like, alright. Come in. He's just like, alright. Woo hoo. And we sat down to give him a demo.
• 1:12:19
  He's like, oh, god. Yeah. This looks pretty good. Yeah. This looks
• 1:12:22
  Speaker 12: I'm so unhappy right now.
• 1:12:24
  Speaker 1: I'm so like, you've made my life so much more complicated. Like, it would have so much simpler. But I and I do feel and I actually it's kind of ironic because we actually got to the Spalsky piece, Adam, from this tweet that was going around about the about the asteroid headed to Sun called Linux on x86. And I think it's more complicated than that. I think it's Yeah.
• 1:12:47
  Speaker 2: I saw that as well and agreed. It's it's it's a multifaceted layered onion of dysfunction.
• 1:12:55
  Speaker 1: It's just really important to be customer centric. You know? You gotta really like, if you when you're and, you know, this is something that I always, like I I just have really admired about. I mean, particularly on Amazon does very well. It's Amazon's super customer centric, and you've got to be that customer centrism is so important and, obviously, not at all costs, but you there's a lot you can go do, I think, that Sun did not go do.
• 1:13:20
  Speaker 3: I guess, I got a quick question. I've tried to jump in a couple Yeah.
• 1:13:23
  Speaker 1: Sorry, Ben. Go ahead.
• 1:13:24
  Speaker 3: So no. No. Don't worry about it. And I'm really glad you brought up the Amazon side of things. Or Amazon as a customer centric sort of company because it leads into the the this question I have, which is, like, you know, I've only read about some microsystems.
• 1:13:39
  I'll betray my age here. I'm a little bit younger.
• 1:13:42
  Speaker 1: My daughter thought it was a brewery. And so as long as you knew it was a computer company, we're in good shape.
• 1:13:49
  Speaker 3: But I've I've read a little bit about it, and, I guess the it's hard to see it, you know, outside of the context that we're in now. And it's sort of like, we're talking about Jupyter Open Source or not open source, but software and hardware and where do you make the money and all that stuff. And I'm seeing it in the context of now of, like, the open core model and what's going on with Elasticsearch and whatever weird name Amazon made up that I can't remember. And how, like, you're they're trying to make money off, like, the the last 6 was making money, still are, but now they're trying to make more money by trying to keep Amazon from taking that money. And I guess it's sort of the flip side of the like, there are companies that figured out how to make money off of open source, and they're on to the next problem.
• 1:14:34
  But I'm wondering why during the Sun Microsystem, hey days when they were really trying to figure out how to monetize this great open source stuff that they had. Did they try something akin to OpenCore, or was it sort of not feasible for some reason?
• 1:14:49
  Speaker 1: If for sure. I mean, Sun had a bunch of different models under one roof, effectively. So the what Java was pursuing was very different than what the what Solaris was pursuing was different than what Spark was. It was not very coherent. In that regard, Sun performed a lot of experiments for us, so we can see a bunch of different things.
• 1:15:11
  I mean, I think that, you know, my belief, and this is kind of in this 2004 piece that my belief was always that you that you want to use open source as a way to drive demand for the thing that you indisputably monetize? What will people pay for? And people will pay for a service, any service. Right? And, you know, whatever it is, people will pay for a service because someone else is doing something for them.
• 1:15:38
  And I, you know, I have always felt that Amazon could be much more forthcoming with open source than it has been historically and not damage itself because people are still gonna pay for Amazon because the the the service is so valuable. And I They clearly are
• 1:15:53
  Speaker 12: not paying for the user interface.
• 1:15:57
  Speaker 1: Josh may have been dealing with the AWS user interface quite but they're not paying for the serial console. We know that. Right,
• 1:16:02
  Speaker 12: Josh? Well, or the web thing or, like, almost any the command line tooling, the, like I mean, the client just it's you know? I I hear about this customer obsession, but I am
• 1:16:11
  Speaker 9: I'm I'm waiting I'm
• 1:16:12
  Speaker 12: waiting to see it.
• 1:16:13
  Speaker 1: That's for sure.
• 1:16:15
  Speaker 2: The
• 1:16:17
  Speaker 1: but I I feel that the and open source serves as a way of driving all of that complimentary demand. And I don't know that Sun was not that coherent about it. I don't know, Adam. What do you think? I I just feel like there was not it was too
• 1:16:34
  Speaker 2: No. For for sure. I mean, some's not that coherent about it, but I think that, you know, even we talk about Elastic. I don't know that that's a great model or and and that OpenCore is is is is that great of a model, but agreed that you're you need to drive people towards the thing that you incur incontrovertibly are adding value that they will pay for. And it's been a lot of takes up, but I don't think we've we've solved that in a writ large, or in a way that can be applied generically.
• 1:17:02
  Speaker 8: Yeah. Sun Sun was selling hardware, but the hardware was was falling behind. Yes. Yes. The software the software had more political power, but was never monetized.
• 1:17:13
  Speaker 1: That's right. And I think that the and we needed the we needed to use that to deliver a a better system. And that's the thing that I find I mean, to me, God's own open source model is when you are using open source to drive hardware sales because that that is like and Apple could be completely open source and would not lose a single sale. I mean, to the contrary, Apple would be much more dominant, I think, if they were much more forthcoming. They believe their Syncracy is very important for their success.
• 1:17:44
  Source, wouldn
• 1:17:52
  Speaker 4: wouldn't and and that would basically make hack what we currently call Hackintoshes legitimate, and everyone would just run macOS on their PCs. Right?
• 1:18:03
  Speaker 12: No. I don't think so.
• 1:18:05
  Speaker 1: Yeah. I I Sorry.
• 1:18:07
  Speaker 8: Well, there there is a period of time when the Macintosh hardware was pretty lackluster. And it's only with the m one that all of a sudden it's like, wow. It's great stuff.
• 1:18:18
  Speaker 3: Yeah. And I guess I'll I'll
• 1:18:19
  Speaker 2: put this on there.
• 1:18:20
  Speaker 9: As much as I mean, I can tell you that I I have a Mac that like the software because I care about the hardware.
• 1:18:26
  Speaker 1: Yeah. That's right. Vinh, sorry. Go ahead. Vinh?
• 1:18:31
  Speaker 3: Oh, yeah. Sorry. I forgot that that's my Twitter username.
• 1:18:34
  Speaker 1: Yeah. Sorry. Yeah. You you listen, man. You gotta keep track of your alts here, you know?
• 1:18:38
  Speaker 3: Yeah. Just this is my 10th Slack bucket. No. I'm kidding. I guess the question that comes up when people talk about open source, it's like especially in regards to free software, it's like, what why are you doing this and for who?
• 1:18:51
  Like, the comment that was made of, like, oh, if Mac OS was open source, everyone would would run it on their laptops and Apple would be, you know, painless or whatever. It's like that's my mom's not gonna do that and I'm not gonna do that for her. I tried that with Linux one time. I was awful. Right?
• 1:19:08
  Like, I got her a desktop and put Linux. I was like, that's this is gonna be better. And that was disrespectful to her, to be honest. So,
• 1:19:17
  Speaker 1: like, how am I right? Yeah. She forgiven you? She'd been, like, you know, I have not I wasn't able to print for 3 years because of you and your card.
• 1:19:25
  Speaker 12: No. No Christmas no Christmas card this year. That's right.
• 1:19:28
  Speaker 3: An iPhone and an Apple Watch really fix that one up.
• 1:19:31
  Speaker 4: Right. No Christmas card because she couldn't print it.
• 1:19:34
  Speaker 3: Oh. But I guess my my point being here is, like, for free software, it's like, oh, who's it free for? Is it free for the end user to do whatever they want, or is it costly, not free, for the end user because they don't know what they're doing? Right? Is it for the experts, or or is it for the consumers?
• 1:19:50
  Right? And and that's that's sort of the the lens that I try to think
• 1:19:55
  Speaker 4: about this. I guess I guess, the Brian, correct me if I'm wrong, but I'm guessing a reason why Joyant was comfortable with making all of smart data center open or Triton, open source was because I guess you guys figured that the real the real value that you were providing and the thing that that people would that companies would be willing to pay for was the service of of helping them get it set up in in their data center and
• 1:20:29
  Speaker 12: and Why couldn't critically, you had a kernel engineer you could call up who would fix the thing that you had happened to you that was terrible with a hypervisor or whatever. Like, that was,
• 1:20:39
  Speaker 7: you know
• 1:20:39
  Speaker 4: So I guess you We had
• 1:20:40
  Speaker 12: I guess you weren't worried about Deeply complex interactions from a support perspective. Like, I mean, just it's amazing. Like, you sell these turnkey things that totally work for you on one machine and for one customer, and then the next customer comes along, they're like, my network's funny. And, like, like, in a way that no one could have predicted. And it takes 16 hours of support to, like, figure out what's wrong with their thing.
• 1:21:03
  Like, that's worth a lot of money.
• 1:21:05
  Speaker 1: It is worth a lot of money. And so, to that point, our, as you can imagine, when we were contemplating open sourcing at all at Jordan, I mean, we're obviously talking to our customers about it and getting their perspective. And one one of our largest customers is just like, I just wanna know that I can continue to pay you. I'm like, yes. Of course, you continue to pay us.
• 1:21:23
  Sorry. Yes. You can give me this.
• 1:21:25
  Speaker 12: By all means.
• 1:21:27
  Speaker 11: Right.
• 1:21:27
  Speaker 1: And so we which was telling you is, like, I I just wanna make sure that, like, you know, you're gonna continue to give me know, this great support and this great product that I've had. I'm like, no. Yeah. Absolutely. What we're he said, oh, then, yes.
• 1:21:37
  Great. Awesome. I can't wait.
• 1:21:39
  Speaker 6: So, Brian, you made 2 comments, one one earlier when I thought I was gonna speak up, but I didn't because the the topic moved. But you you you spoke about, Sun being interested in growing the pie and and just more Did anyone discuss it? The vision for a hardware company. I'm still here. Yeah.
• 1:22:01
  Speaker 1: Yeah. Maybe you cut out just for a second after growing the pie.
• 1:22:04
  Speaker 6: Yeah. Okay. So they're growing the pie thing, and then there was the comment about, open source and and hardware together making a lot of sense. What was the thinking at Sun at the time around Linux, and and why wasn't Sun one of the first, companies to to embrace Linux? Why was it IBM?
• 1:22:22
  You know, that that seems like it could have been a very different future for them if they'd backed the Linux horse and and made sure some hardware worked great on Linux and and, provide brilliant servers that, you know, never never give you any trouble. It just seems like a
• 1:22:37
  Speaker 12: Well, I fit. Or is it
• 1:22:39
  Speaker 6: that we we put our own stuff and it's better and and, we want you play with our open source?
• 1:22:43
  Speaker 1: Well, I mean, that question's obviously a trap. But, the, I mean, our it was better. I mean, that's honestly I mean, if you want my honest opinion, it's like we what we saw and still see, honestly, is, like, there are still tons of things that in that that I mean, it's, you know, whether it's service management, fault management, file systems, observability, debugability.
• 1:23:05
  Speaker 6: So so bring bring bring those capabilities
• 1:23:08
  Speaker 1: to Linux.
• 1:23:08
  Speaker 6: Right? Like, grow grow
• 1:23:09
  Speaker 5: the flow.
• 1:23:10
  Speaker 6: And You can't you can't participate with the Unfortunately, you can't
• 1:23:14
  Speaker 12: force people to be other than who they already are.
• 1:23:18
  Speaker 1: Yeah. I mean, if you look at like I mean, right now right now, ZFS on Linux has been ported to to Linux for how long? And it still is it it has been made clear. It's never going to be adopted. It's a it is a into mainline.
• 1:23:33
  It is a rampant layering violation.
• 1:23:35
  Speaker 12: Well, and and and they they have no compunction about breaking it if it's convenient. Right.
• 1:23:41
  Speaker 1: So I mean
• 1:23:41
  Speaker 8: But but on the other hand, Ubuntu fixes it all up and it's trivial to install. Right.
• 1:23:46
  Speaker 1: That's right. That's right. And so which is great. So terrific. And and, you know, good on them.
• 1:23:53
  The I mean, I do think that so, I mean, I I don't know. That would not have solved Sun's issue. Let me put it that way. That is not that I do not view that as as
• 1:24:04
  Speaker 6: I'm gonna disagree with you on on on one, though, Brian. Like like, if if if Sun had been participating all the years, wouldn't they have a lot more to say about the direction? I mean, that's kind of how it goes with with open source. Those who do get to decide. Right?
• 1:24:20
  And and if you have a history of
• 1:24:24
  Speaker 2: doing that over time, might
• 1:24:25
  Speaker 6: the story not Well, I
• 1:24:26
  Speaker 9: I mean, you you you I I had never worked at Sun, but I think that it's also important to contextualize this and the historical context of what's happening at the time. Like, Linux was viewed very much as a toy for like, well into the 2000. You know, in 2006 or 2007 was when it started becoming kinda real to run, like, enterprise y type things on Linux. At least that was my impression. And, you know, like, something to bear in mind is that there was a there was a proposal within Sun back in, like, the early nineties to open source the BSD based on us version 4.
• 1:25:02
  Speaker 3: Dan from Unfortunately, it never went anywhere.
• 1:25:03
  Speaker 1: You know who that's from. Right?
• 1:25:05
  Speaker 9: Oh, yeah. Oh, I know who that's from.
• 1:25:07
  Speaker 1: That's why that's Larry Maguire.
• 1:25:09
  Speaker 9: I do too. Yeah.
• 1:25:10
  Speaker 8: But on the on the Linux front, you know, Sun bought Cobalt for $2,000,000,000. Right?
• 1:25:14
  Speaker 1: A $2,000,000,000 Confederate dollars, Tom. Those were that's that's all, inflated, Sunw Sunw stock price circa 1999. Yeah.
• 1:25:22
  Speaker 9: Yeah. Yeah. Yeah.
• 1:25:23
  Speaker 12: Jeez. But I hope they sold quickly.
• 1:25:26
  Speaker 8: But then but then it was pretty much killed by the, you know, antibodies because they ran the toy.
• 1:25:33
  Speaker 9: I mean, you know, the the writing was on the wall for Sun well before the year 2000, unfortunately. I remember around 1997, 1998, talking to a system administrator who pointed to 2 PCs, and he's like, look. These things are half as good as a spark station, but they cost a quarter as much.
• 1:25:52
  Speaker 1: Yeah.
• 1:25:53
  Speaker 9: Like, give it another 2 years, you know, like, it's it's just not gonna be a comparison. It
• 1:25:58
  Speaker 12: feels like the biggest the biggest product missed product direction would have been to have paved the Earth with x86 machines with good management. Like, Like, because the I mean, the the LOMs and stuff were were better than many of the competing things, but they just want many units being sold. Like
• 1:26:20
  Speaker 3: That's right. I I think a lot
• 1:26:22
  Speaker 2: of people get confused by this
• 1:26:28
  Speaker 1: hardware. And then
• 1:26:28
  Speaker 12: missing the missing the x86 boat?
• 1:26:31
  Speaker 8: That's right. Yeah.
• 1:26:32
  Speaker 12: I mean,
• 1:26:32
  Speaker 8: I I left in 94, and a part of that was the Spark roadmap was looking really bad even that early.
• 1:26:41
  Speaker 1: Yes. SMP, saved Spark in that. Spark had under underperforming, but wildly parallel systems. And that's S and P bought us. And I I know Adam, do you I mean, Adam and I had this moment during the boom, when we just seen the pricing on some boxes.
• 1:26:59
  Do you yeah. Obviously, you remember this, Adam.
• 1:27:01
  Speaker 2: Yeah. I remember. I I wasn't gonna toot my own stupid Prussian horn though.
• 1:27:05
  Speaker 1: You should toot your own stupid Prussian horn. Go ahead.
• 1:27:07
  Speaker 2: Well, so so we were I think I was an intern even at
• 1:27:09
  Speaker 3: the time.
• 1:27:10
  Speaker 2: I'm not sure if it was an intern or I just joined. It was, you know, 2,000 or 2,001. And we had just released it. It was like an 8 core or 8 8 socket.
• 1:27:19
  Speaker 1: Right? Yep. 8. Yep.
• 1:27:20
  Speaker 2: Yep. System. And it cost like a $100,000?
• 1:27:22
  Speaker 1: Yeah. VAD. Yeah.
• 1:27:25
  Speaker 2: And I said, you know, $100,000 seems like a lot of money. We're like, yeah. I don't know. It seems like a lot of money. I said, well, couldn't you buy, like, a a 100 computers?
• 1:27:34
  Like like, you know, x86 piece of shit computers and, like, string them together or something. And at that moment, we should have, like, pulled the fire alarm and, like, forced an all hands on that topic, but instead we just, like, went to lunch.
• 1:27:46
  Speaker 12: Even if he even if he only bought even if he only bought 50 with ECC memory.
• 1:27:50
  Speaker 1: Like That's right. That's right.
• 1:27:52
  Speaker 8: Well, and
• 1:27:52
  Speaker 1: and the thing is, like, I mean, Adam, you and I both remember exactly where we were when we had this conversation. We remember this conversation vividly. I remember you being like, hey. Just explain it to me because I'm sure there's a reason for this.
• 1:28:03
  Speaker 2: Right. I'm new here.
• 1:28:03
  Speaker 1: I'm new here. And why would someone do that? And I remember thinking, like, why would someone do that? That does seem weird.
• 1:28:10
  Speaker 7: I don't know.
• 1:28:11
  Speaker 2: Go figure. I'm sure someone's thinking about it. Right?
• 1:28:13
  Speaker 1: Someone's thinking about my pay grade. Meanwhile, I think you and I both want the time machine to go back to our ourselves and be like, no. Like, it's like Bill and Ted. It's excellent adventure for x86 Economics.
• 1:28:23
  Speaker 2: Take take at that thread a little bit and you could either save the company, or become a billionaire, or or die trying. That's
• 1:28:31
  Speaker 12: The sad thing the sad thing was that the Opteron wave happened. Right? And Sun was there. They had Optron boxes. When AMD 64 happened, and and x86 grew up to become a real 64 bit architecture with a future, like, and and, a whole new, like, CPU product line emerged, Sun had some of the first and best servers that contained those things.
• 1:28:57
  And then I guess they sold 8 of them and stopped. I don't know. Like, they were not expensive comparative, really, compared to, like, HP boxes at least. I mean, I don't know.
• 1:29:07
  Speaker 9: Yeah. I I really
• 1:29:09
  Speaker 8: wanna go go the low margin game that that, Dell and HP were doing.
• 1:29:13
  Speaker 1: Right. But that's Yeah.
• 1:29:14
  Speaker 9: But that's and that's but that's that's
• 1:29:16
  Speaker 12: good night, Irene. I mean
• 1:29:18
  Speaker 9: But that but that but that was the thing. I mean, I I I remember at the UltraSpark announcement that, you know, local sun resaler booked a, you know, room at some restaurant. A bunch of us went to see this thing and McNeely gets up on stage. And I remember him saying this. And if you thought that DEC Alpha was the fastest microprocessor on the market, it's not, it's the UltraSpark one.
• 1:29:39
  Right? And it it like, my impression of that was that, gee, all the, like, weird risk workstation manufacturers are 1 upping each other. And that's not all that useful because Intel is quietly making gains with the Pentium Yes. And they're just cutting the margins razor thin and making it up on volume. What are you guys doing?
• 1:29:59
  You're gonna you're gonna get buried.
• 1:30:01
  Speaker 1: Well, the margins were not razor thin on the end. They have their own razor thin for the CPUs For Intel. Intel's always had pretty bad margins. But the, the it I do think that that Josh is right that in 2005 ish, there was a lot of reason to believe that a renaissance was possible because I do think that there are a bunch of those things were righted. And Tom so I had been through rough times before, 1991, and it kind of emerged.
• 1:30:26
  And I did feel that it was possible, but I it is also true that there was not really deep interest in running a business at the top of the company. And that is the
• 1:30:35
  Speaker 12: It's pretty unfortunate because the X2100 was, like, pretty good
• 1:30:40
  Speaker 1: I know.
• 1:30:40
  Speaker 6: At the time.
• 1:30:41
  Speaker 12: Like, they were, you know, cheap and fast and had a long that worked.
• 1:30:47
  Speaker 8: Like, I
• 1:30:48
  Speaker 9: I I'm not
• 1:30:49
  Speaker 12: I honest to god, I wish that I had a rack of x 21 100 right now compared to the gigabytes and tie ins and stuff, like, that we've got in in the in the corner. I mean, jeez.
• 1:31:01
  Speaker 1: Alright. So I should Josh, it is it is worth explaining an act of particular cruelty that we have inflicted upon you personally. So Josh is that oxide, then sheared oxide. Went it down against just oxide. Hello, man.
• 1:31:11
  And and the, but we have needed I mean, you've gotta develop the software before you actually have the computer built. So poor Josh has had to go stand up the gigabytes and the tie ins. And, Josh, I think we should start a computer company. I don't know about you. I think that I
• 1:31:25
  Speaker 12: I'm ready. Sign me up.
• 1:31:28
  Speaker 1: So you had to suffer with it. It's like, yes. I know. Like, I know it's the thesis of the company that the c things are terrible. By by the way, they're actually terrible, and I'm actually having to deal with it.
• 1:31:37
  Speaker 12: As far as I can tell, when this computer boots up that I have, it engages a USB NIC driver to talk to the BMC, and then asks asks for the boot posture that I have set in the BMC, which I did via the API. 60% of the time, it accepts the new information, and the other times it doesn't. But, like, but it doesn't even print a message to say that it didn't. It merely boots the wrong thing, and there's no way to know.
• 1:32:05
  Speaker 1: But someone's gonna use it.
• 1:32:07
  Speaker 12: Reboot it, it might work. Maybe? I don't know. I just it's just
• 1:32:12
  Speaker 9: Must be using UDP over USB
• 1:32:14
  Speaker 1: over
• 1:32:15
  Speaker 12: So like, as a customer, Spark didn't have any of those problems. It was just extreme like, it was just extremely slow. That's really the at the end of the and too expensive. It was Yes. Slow and expensive, but seriously, it had no other problems.
• 1:32:29
  Like
• 1:32:29
  Speaker 1: it it will and, excuse me. There was and then there there was the the occasionally cash barrier. So we it's it's, like, slow,
• 1:32:35
  Speaker 12: extra cost
• 1:32:35
  Speaker 7: of it. Like, not very reliable.
• 1:32:37
  Speaker 12: So, like, light ultra spark 3, ultra spark 4, t 1, t 2 era stuff didn't have that specific problem. But, like like, seriously, it's just, like, BIOS stuff and even and UEFI things and and ASP and and AMI and like, it's just all such trash.
• 1:32:55
  Speaker 1: I'm so sorry, Josh. I I'm so sorry. And I feel that, like, you because you know what we're building, it's like you you must be awake during the surgery. You must feel like I I can feel all of the pain now much more sharply.
• 1:33:07
  Speaker 4: Well, it's just I mean, just the it's just
• 1:33:10
  Speaker 12: the it's like the thing is that, again, the x2100 that Sun, like, could have made more of didn't have these problems, but it had a fast optron in it. And it was cheap.
• 1:33:20
  Speaker 1: So Oh, you couldn't make them show up. That's the problem. Could sorry. I know.
• 1:33:25
  Speaker 13: Hey, folks. Yeah.
• 1:33:25
  Speaker 9: Sorry. You're
• 1:33:26
  Speaker 10: right. Corliss here. I I just wanted to chime in because I was actually a Sun customer at Cisco, around 2,000, etcetera. And I I just wanted to say that, like, we stuck by Sun for a long while. And the reason being is because we're like, we were one of the earliest biggest websites, you know, before, like, these these massive search systems got started, like, you know, Facebook, all the social media companies.
• 1:33:49
  But just as a company website, we're pretty darn, you know, high end there. And so we use the SUNS because they could support vertical scaling. Like, you can get, you know, 8 at the time, 8 CPUs was a lot. Right? You couldn't get that on an 80 x86 box.
• 1:34:04
  You know, you're in fact, the big thing though was that Sun could scale vertically, but they never really figured out the horizontal scalability. Yeah. And and I think that if if they had done that, if they had had a better horizontal scalability system, they could have, you know, done a totally different kind of architecture to continue to support as systems are scaling up. And I think, you know, it's it's unspecified here, but one of the nails in the coffin was NGINX. You know, once you started having these architecture systems that supported horizontal scalability, so cheap commodity hardware could do in mass what you could do with these vertically scaled sun boxes.
• 1:34:47
  And I I think that's really the inflection point for sun.
• 1:34:50
  Speaker 1: Okay. This is actually a really interesting point, actually, because I I think you're you're absolutely onto something. Obviously, I mean, you're very very much in our current thesis, so, obviously, very strongly agree. But because you're actually getting to the larger impact of open source, namely, we would not have been able to build we humanity would not have been able to build these scale out systems without open source. If NGINX had been strictly proprietary, you couldn't do it.
• 1:35:14
  Economically, you couldn't do it. And Google couldn't have built what Google built without open source. Amazon could not have built Amazon built with with, without open source. They needed actually open source to make that commercially viable from a software perspective. And you're right that Sun never really and there are a bunch of reasons for it that I think have to go do with go to market and the way people are incentivized and compensated and a bunch of other things.
• 1:35:39
  But, Sun definitely did not really embrace those scale out systems.
• 1:35:45
  Speaker 10: Yeah. But also, I wanna give credit. A lot of people don't realize, you know, again, some people said that they're too young to remember Sun. I remember them well. And I remember going to this one conference, and this is before 9:11.
• 1:35:58
  And there's this one trading firm on Wall Street. And I was asking, like, so why do you use like, every single trader they had had a SunSpark station on their desktop. Like, not just a PC, a SunSpark station. And and it was like, well, why do you have that hardware on everybody's desktop? She says, because if we can calculate derivatives microseconds or milliseconds faster, we win the bid, you know.
• 1:36:22
  And it was that level of hardware, that silicon. I mean, that's still available today in, like, high frequency training. And now there's a 6. But at the time, like, that was what you were trying to do is that, you know, you would make 1,000,000,000 on whether you could just calculate faster. And, you know, so the the you know, so for those kind of use cases, Sun was just unparalleled.
• 1:36:43
  But obviously, you know, the reason why x861 is because 90% of use cases don't rely upon that kind of millisecond scale difference. Like the most users, blink of an eye performance is fine enough.
• 1:36:55
  Speaker 1: Well, I meant x86 was just faster. So it's like if you
• 1:36:59
  Speaker 8: Well, it it took
• 1:37:00
  Speaker 9: a long time for x86 to get to that point.
• 1:37:02
  Speaker 1: It was it was true by I mean, it's interesting that Tom says he saw the writing on the wall by 94. Certainly, for me personally, the inflection point was at 9:98. So in 98, the, with the with the the Pentium Pro, it is very quick. And then certainly by by 99, 2000. It's like it's and we were trying to kind of express today.
• 1:37:25
  At the same time Adam, how many of those Cheetah plus meetings did you get roped into?
• 1:37:29
  Speaker 2: Too many. Oh,
• 1:37:31
  Speaker 1: god. I'm so sorry. I wrote the unit of those, didn't I?
• 1:37:34
  Speaker 2: Yeah. You you absolutely. Thank you for that. But but yeah. I mean, like, you're So I get
• 1:37:37
  Speaker 1: have you been waiting for me to say that for, like, 20 years? Yeah. Kind of. Alright. Kind of.
• 1:37:41
  Speaker 2: I've been waiting for an apology. I think it was overdue. But but, yeah, I've noticed that your, like, your, your desktop back in your dorm room is a lot faster than the sun box in the computer lab, for sure.
• 1:37:52
  Speaker 1: Right. So, yeah, you exactly. You had an even more visceral experience where you got, like, this supposedly fancy sun lab at an undergraduate institution. And you're, like, actually, like, your sun lab is not as fancy as my dorm room computer, by the way. Yeah.
• 1:38:06
  We've got, it's got a we've been trying to keep you to an hour. Clearly, this is a this is an evergreen topic, I think in part I think because, like, Sun was lovable. I think the the reason that this is an evergreen topic is people want to reason about this because there were there were aspects of sun that were, technologists appreciated. And we'd like to know, like, was the asteroid fated to collide with with the Sun Microsystems planet? In conclusion, yes.
• 1:38:36
  No. I don't know. On that note, thank you very much, everybody. I don't I think we maybe, I actually did a little bit of an exciting news, in that we at Oxide are, are bringing up our first server boards this week. Extremely exciting.
• 1:38:51
  So we've got, we we go to power on, next week. I'm about to fly out, so to do part engage in software bring up. So I don't know that we're gonna be here next week, but for all of the right exciting reasons. And as one of our colleagues at bring up always goes, much better than plan or much worse than plan. There's no or much worse than plan.
• 1:39:10
  There's no in the middle. So, So we'll see you in 1 week?
• 1:39:12
  Speaker 12: Or 17 weeks. That's exactly right. So on that Bring up can bring
• 1:39:15
  Speaker 1: up can really bring you down. That's right.
• 1:39:18
  Speaker 9: Hey. Good. Up can really bring you down.
• 1:39:20
  Speaker 1: That's right. Hey, with the knock on wood. Alright, everyone. Take care. Thank you.
• 1:39:24
  Speaker 3: Thanks, everyone.