Latest / Oxide and Friends / Working on Wicked Problems

Transcript

Working on Wicked Problems · Oxide and Friends

• 0:00
  Adam Leventhal: I took that thirty seconds for my thirty second bathroom break before the show.
• 0:05
  Bryan Cantrill: You did the bathroom in thirty seconds?
• 0:07
  Noah Harlan: That was impressive.
• 0:08
  Adam Leventhal: Well, let's not go into the details.
• 0:11
  Bryan Cantrill: Does it involve a catheter? Because I... I've been thinking a lot about that.
• 0:14
  Noah Harlan: Feels like a jar situation.
• 0:17
  Bryan Cantrill: It does. You know
• 0:17
  Adam Leventhal: It's like the Amazon employees, like, with the secret... Dirty secret of Oxide employees. Like, I had to empty the Gatorade bottle.
• 0:24
  Bryan Cantrill: Yeah. I know. I know. You you... I saw your request for a...
• 0:27
  Your formal request for a ninety second bathroom break, and I think you saw that I denied it. And I... Checking the the convenient go pee in a bottle reply, which Augustus automated for us. It's very... Actually very useful.
• 0:39
  Adam Leventhal: Was Big time, Dave.
• 0:40
  Bryan Cantrill: Was just typing it out all the time, you know, just like.
• 0:44
  Adam Leventhal: Yeah. Hey, I mean, not that I want to trim any of this from the recording obviously, but do we want to do like a real hard reset right now?
• 0:52
  Bryan Cantrill: No, no, no, no. Why would we... Oh no. Nope. Sorry.
• 0:55
  That's just something you think.
• 0:56
  Adam Leventhal: Ship has sailed. Okay.
• 0:57
  Bryan Cantrill: That's fine. Has sailed and peed in a bottle while at sea.
• 1:01
  Adam Leventhal: Yeah. That's fair.
• 1:03
  Bryan Cantrill: No. You know, you... We've got all... We all have... I guess, no.
• 1:05
  Your kids are still slightly preteen, but we've got Adam and I both have teenage children. I feel like every road trip
• 1:11
  Noah Harlan: The older one's teenager. She's
• 1:13
  Bryan Cantrill: Oh, teenager. Okay. So I don't know if you have this, but I would say that every road trip with a teenage boy, you have this... And maybe this is just my teenage boys or teenage boys and former teenage boys, Adam, but it feels like we had the discussion at least once in every road trip of just like, I could just pee in a bottle. And me being like, absolutely not.
• 1:31
  Like, you know, it's fine. I'll just... Actually, I'll just do... I'll... I...
• 1:33
  You know, I can actually help myself on this one. And me having to intervene to prevent urination well at 70 miles an hour. Am I am I... Yeah. You're your own.
• 1:43
  No. You're your Thank you. Alright.
• 1:45
  Adam Leventhal: Never mind. And...
• 1:46
  Noah Harlan: I have girls. So...
• 1:47
  Adam Leventhal: Not... Oh, there you go.
• 1:48
  Bryan Cantrill: Oh. Oh, god. Yeah. There's
• 1:50
  Noah Harlan: there's civilized creatures.
• 1:51
  Adam Leventhal: And I have no teenagers right now. So for...
• 1:54
  Bryan Cantrill: You have no teenagers. That's right. Yeah. You're in the gap that you've got. That's right.
• 1:58
  My big
• 1:58
  Noah Harlan: my gap year. Yeah.
• 2:00
  Bryan Cantrill: You're in your gap years. Yeah. Yeah. I mean, it's a... It is kinda...
• 2:03
  It's kinda the eye of the hurricane, isn't it? Like, you got... Like, you know, you you you don't even get comfortable because,
• 2:08
  Adam Leventhal: If this is the eye, like, just
• 2:10
  Bryan Cantrill: you got the other side of the system is about to blow over. So just don't... Yeah. Well, no. It is great to have you here.
• 2:20
  We we we had taken some time off. We had OXCON, which was a lot of fun. We had the the the escape room in here in the podcast studio, Adam, which was a lot of fun. We... Our...
• 2:33
  Several of our colleagues made an escape room, which is And then I was out at Oktoberfest. And, Noah, you and I, just in terms of origins of this conversation... Well, I I should say this, Adam, were you... I I know you have tried to schedule so called vacation while I write... After I write a doomerist piece...
• 2:52
  An anti doomerist piece, hoping that you can avoid a podcast on it. I gotta... I, you know, I see what you're doing there. I see you.
• 2:57
  Adam Leventhal: It's little... Do I do like my It's
• 2:59
  Bryan Cantrill: little embossing.
• 2:59
  Adam Leventhal: International trips. Yeah.
• 3:00
  Bryan Cantrill: Mean Yeah. Exactly. Those surprise international trips. I mean, don't really add up, honestly. You know?
• 3:05
  I mean, you're going to countries that have visa requirements that you have. Mean, it's like... I'm just like, make it a little more plausible. That's fair. And so I know you were hoping to shelter in place and not have doom...
• 3:15
  Anti doomarism talk, but, the anti doomerist talk will continue until the doomerism, abates, I would say. And, Noah, you and I were at Monctoberfest together in lovely Portland, Maine, and, talking a little bit about this MTS live appearance that I had. And, Adam, did you end up watching that whole thing?
• 3:37
  Adam Leventhal: I watched the whole thing. No. But I watched much more of it than I wanted to. I bet you halfway.
• 3:45
  Bryan Cantrill: Yeah. Meet me halfway. I mean, I feel... I'm a I'm a little hurt because you read all of Eliezer Yudkowsky's book. I mean, you did do that.
• 3:53
  That you... That you do the whole
• 3:55
  Adam Leventhal: thing up,
• 3:56
  Bryan Cantrill: apparently. Much bigger time frame of it. Much worse, I feel.
• 3:59
  Adam Leventhal: I also read many... I I read more of the YouTube comments than I thought I was going to.
• 4:05
  Bryan Cantrill: The YouTube comments are... It it... You know what? It harkens back to YouTube comments of... I mean, I'm definitely in there with the kids.
• 4:14
  I got... Well, did you have any favorites that you'd like to...
• 4:17
  Adam Leventhal: You'd a class? I I was just scrolling back to to try to find mine. I mean, do you... I I guess I could always cut it. This Cantrill guy comes across as terribly shallow, pseudo thinker douche who happens to be in the right place at the right time and now sees the whole world through their government contract.
• 4:36
  I mean, that
• 4:36
  Bryan Cantrill: was that
• 4:37
  Adam Leventhal: was like a pinned pinned quote.
• 4:40
  Bryan Cantrill: It's a pinned quote. Oh, wait a minute. That was a DM that you sent me. That's
• 4:43
  Noah Harlan: not actually...
• 4:43
  Bryan Cantrill: Oh, wait a minute. Sorry. Am I reading from the wrong screen again? Jesus.
• 4:49
  Noah Harlan: I... My my overall feeling about it was that it reminded me a lot of... I grew up on a farm and we had these retrievers and they would go and pull groundhogs like out of their holes. And then they would spend hours just kind of playing with them in this masochistic thing that would leave, like, them bloody and eventually the the groundhog dead. And I sort of got that look in your eye as you sort of lost in on this guy, And it reminded me a lot of my old dog.
• 5:21
  Yeah. Yeah. Yeah. It's a very flattering thing.
• 5:25
  Adam Leventhal: The... I think the moment for me that captured that is I think Bryan's quote was, oh, Jesus Christ. Then launched in from there.
• 5:34
  Bryan Cantrill: Yeah. That was one that I kind... That was an expletive that I let loose before I actually realized that. This is because this was the what if we fully... We 're gonna fully automate the economy.
• 5:45
  Also, with the the line that he had... This guy's, like, not a doomerist, but he is a, I would say, doomerist sympathetic. And he had a line when I was decrying one of the things he said is science fiction. He said science fiction can sometimes be real. Well I'm like, is that the intellectual underpinning of humorism?
• 6:04
  Because I think was like we could feature that one above the fold.
• 6:07
  Adam Leventhal: Yeah. Like, science fiction sometimes comes true to which you're like
• 6:10
  Bryan Cantrill: Science fiction. Yes.
• 6:11
  Adam Leventhal: And and you're like, well,
• 6:13
  Bryan Cantrill: okay.
• 6:13
  Adam Leventhal: I mean, granted. And there there actually is... Was it an Asimov line or who who said it? Who... Which was like, you know, nobody ever invented something that nobody had imagined first.
• 6:23
  So it's like, yeah. I mean, sort of sort of that that is true. But you... You're... I mean, you have a very good reply, which was, you know, a lot of science fiction doesn't come true.
• 6:33
  Bryan Cantrill: Also, doesn't come true. Also, I'm wondering, is this something that he applies to, like, elsewhere in his life? You're like, what the hell are you doing? You're like, sometimes romance novels come true. It's It's like, okay.
• 6:42
  That's not actually
• 6:43
  Noah Harlan: And I think that it's actually I'm disinterested. To the kind of like the the Yudkowskian view of like, well, you know, like, there's a probability. There's some probability that like in the future there will be a billion shrimp that I need to save. And thus like, this is all worth it. And so like they are playing these weird like probabilistic games that are sort of making Douglas Adams, you know, bury his head in the ground, you know, and scream.
• 7:11
  Bryan Cantrill: And so we... So, Noah, you and I were kind of talking about this because you would you would you would caught this. And as kind of a segue into your background, because, you know, I feel that you... And a lot of the work you've done recently, especially, and kind of the reason that that I wanted to have you on here. And I guess you and I were having a great discussion.
• 7:30
  I'm like, this should be a Munktoberfest talk. And I'm like, wait a minute. Munktoberfest is like a year from now. Like, it could not be... Munktoberfest could not be further away from now than it is right now here at Munktoberfest.
• 7:40
  And I'm like, actually, we should just have you on the podcast. But what I found so compelling is that you had you had a career as a serial entrepreneur. We talk about some... A bunch of things you've done there. And then decided to like, okay, I'm like, I I...
• 7:56
  I'm I'm done. I've kind of like, I've handed this thing off, this company off. And now I'm gonna go pick up this really thorny problem. Can you talk about like that moment and problem that you went to?
• 8:09
  Noah Harlan: Yeah. So it wasn't as clean as... Because the problem emerged from work that we were doing at the company. But just to rewind it back a little bit, I had this company for a number of years that was a software studio. Effectively, we helped other companies sort of think about and then implement pretty complex digital products of various sorts.
• 8:34
  And along the way, we had been hired on a project. We had a lot of we'd done a lot of work in open source and we'd done some public interest and, you know, kind of gov tech adjacent stuff and got hired on a project to help individuals who are going from temporary shelter to permanent shelter in the cities of New York and Baltimore try and reduce the administrative burdens, and we can unpack what that means. But, basically, try and improve that process, for those individuals. Was And was it cool? Agency that
• 9:13
  Bryan Cantrill: project. That engaged you? Was this, like, was this a an agency under those two cities or a state agency or who had engaged you on that project?
• 9:19
  Noah Harlan: So it was a think tank based out of DC called New America that had a couple different divisions that were looking at different types of public interest technology. And in actual fact, had looked at us a year earlier to take on this project. The city of New York rejected us and went with a different company. In the meantime, they hired us to build a jobs platform. This was like summer of twenty nineteen for the state of New Jersey.
• 9:50
  That was all like using behavioral science from this group called the Heldrick Center for Workforce Development at Rutgers. And it was really cool because I was thinking about like, if you are newly or long term unemployed, how do you get back into the workforce? And it's like all these little behavioral nudges that kind of try to help you not give up. And we built that platform and it turns out that in the spring of twenty twenty, serious unemployment suddenly became a really big concern as you may recall. And so that platform launched in the late spring of twenty twenty and they turned to us and they said, hey, remember that data locker thing from a year ago trying to help individuals experiencing homelessness?
• 10:38
  And we said, yeah. They said, well, the company that New York City made us hire actually completely shanked it and they haven't produced anything. And now the city of Baltimore is involved and we have to deliver a version of this thing and open source and we need to get that done by January 1 and it was like August. Could you... Would you mind helping us out?
• 11:00
  And we're
• 11:01
  Bryan Cantrill: like This is August of of twenty twenty. So this is like, I mean Yeah. Height of the pandemic. Yeah. I'm just like, wow.
• 11:07
  Noah Harlan: Yeah. Height of the pandemic. But we had we had this team that was come rolling off in this other project and they were really jazzed to do it. And so we dove in and we built it, we delivered it. We open sourced it.
• 11:23
  And we also, we didn't really get a lot of chance to step back and think through the product. We kind of had to deal with the requirements as they were delivered. And it was eminently apparent that there were a whole bunch of structural problems with the approach to it, that it was really gonna kind of replicate a lot of the administrative burdens that people were dealing with in paper, but just like replicate them in digital. But we had a sense of like, oh, here's how we could potentially do this different. At the same
• 11:53
  Bryan Cantrill: Can time you speak to that a little bit of like the details of why it was replicating it? I suppose like what you were kind of discovering as this thing was being deployed?
• 12:01
  Noah Harlan: Yeah. So for a family, and there's a difference between individuals and families when you talk about homelessness, but in this case, for a family that was entering a shelter in New York City, their journey to getting into some sort of permanent supportive housing, which is kind of the term for getting stable, was on average about eighteen months. And it was like this Kafka esque insane thing. You had to physically go to the start center in The Bronx. You had to bring your entire family there, which meant like if your kids were in school, you had to pull them out of school.
• 12:37
  You all had to be there on the day. You had to have all your documentation. Then they send you to some random shelter that will be your intake shelter, which might be in another borough and like none of your stuff transfers and then you have to like figure out how to deal with like getting set up there and getting your kid enrolled in a school that they can get to from that shelter and all of these other things. So the journey, that eighteen month journey, like half of it, not in sheer linear terms, but like half of it is just spent dealing with administrative burden. And administrative burden are all of the kind of paperwork and time taxes that are put on us.
• 13:12
  So the journey to The Bronx, filling out all this documentation, presenting your birth certificates and your social security cards and all these things that are required. And and there's kind of reasons for it, and we can go into that a little bit. But... And then you're asking people who are kind of in the worst moment in their life, and they're in places where they've been entirely disrupted, and the reason for them being in a shelter might be because their house burned down, you know, and they don't have any of that stuff or whatever. And you're asking them to kind of gather all this.
• 13:47
  We looked at that and we were like, okay. Well, you know, a digital data locker sounds great. But the way they did it was great. This particular agency at New York will stand up the data locker. Now, New York City has an enormous technology department that is god awful.
• 14:10
  And I will stand up here and say that, they are very bad at building things. They are very bad at running things despite the fact that we are like a nation state. We are so effing big. And there's a lot of reasons behind that, but we were like, okay. So we're gonna build this open source thing.
• 14:26
  We're just gonna hand it off to you and you're going to run this. But then when they go to the next agency, this data locker won't, like, easily transfer and that next agency will also have to stand up a version of this. So like you're dealing with like city, county, state, banks, you know, Maybe their kid goes to parochial school. Wait, all of them are going to have their own data lockers. You're gonna need one password.
• 14:55
  Guess what these folks are not doing? Password managers and all this stuff. So now all of a sudden they have like 30 different data lockers with their data strewn across it and they don't know where anything is. And so it it was flawed by design, both in terms of how you would run it, which, know, I have some strongly held views about public interest technology and open source, which is I think it's a fool's errand for the most part, and we can talk about that. And then it was also flawed from a from a user experience perspective.
• 15:31
  And our vision was like, okay, if you're going to try and do this, you need to do it either with one organization running it for everyone or as many people as you can in a community, or you need some very federated model. But I spent a lot of time working in open source at the Linux Foundation and, you know, setting up and running projects there. And I know how hard it is to build open source communities and doing it in an area where nobody is particularly incentivized is even harder. And so that was the kind of kernel of the idea that led us to start Keep as we were selling off my company Tubols. And I was like, I wanna go do that.
• 16:18
  So we spun it out and set it up as a startup, and that was the origin of it.
• 16:26
  Bryan Cantrill: Yeah. And... Alright. So the... And maybe this is a good kind of opportunity to...
• 16:30
  Because, I mean, what you're taking on in terms of this this document document management for folks that are between agencies that are requiring indigent care or are unhoused. I mean, this is all different facets of a wicked problem of kind of a classically wicked problem. And do you wanna describe that a little bit? I I... And I review that this paper that was written in 1973 that that kind of coined the term wicked problem.
• 16:58
  This is dilemmas in a general theory of planning by Horace Riddle and Melvin Weber. And I guess that they they had come up with the idea of a wicked problem some number of years earlier and had finally been one to write it down in 1973. Adam, did you have a chance to look at this, by the way, this
• 17:14
  Adam Leventhal: this paper? Yeah. I I read more of it than I than I watched the video, but... And I had not heard that term before. So when you were describing as a wicked problem, I I did not have the context.
• 17:24
  But, yeah, it was a... It's a great paper and a... And an interesting concept.
• 17:28
  Bryan Cantrill: It... And it kinda reads modern, I felt, for something that is 50 years old. It felt like this is like... Yeah. This feels pretty germane today.
• 17:37
  Does not feel like it's, like it's trapped in in amber in the seventies. It feels like a lot of these problems have remained, wicked. I I I don't know. Did you get a chance to... I don't know if you even, like, were thinking about it as if that, like, taxonomy is even, like, used by the people that are actually taking a swing at these problems?
• 17:57
  Noah Harlan: So I had not seen this article before. I was thrilled when you shared it. I got a chance to kind of like quickly skim it at a surface level. It reminded me a lot of some other work that I actually am very influenced by, which is like the work of Sydney Decker and in particular like his examination of synoptic legibility and what creates like chaos and systems. And also a guy named Dave Snowden at IBM, his Kinevan framework and like the relationship between chaos and simple systems and how they like simple systems when they break, they sort of fail into chaos.
• 18:48
  And I think that there were echoes of both of those in what I was reading, but I wanna sit down and, like, really parse it through. Because also the the author is, like, really obsessed with, like, you know, like, what what what did he open it with? It's, like, these very, like, florid... Like, George Bernard Shaw, he opens it with, which I was like, oh, okay. We're we're we're we're starting in a in a very lyrical place here.
• 19:12
  Adam Leventhal: Yeah. No. I I had the... It took me a minute to sort of orient, like, where are we? Like, what what discipline are we in?
• 19:20
  Like, what are we talking about? Yeah.
• 19:23
  Bryan Cantrill: I I I just love, like, every every profession is a conspiracy against the laity. I feel like that just feels like a very modern complaint by the laity. I feel like Yeah. I mean, this is... That's like anti vaxxing in a in a clause in in, you know, by George Bernard Shaw.
• 19:40
  Right? I mean, is the... I guess that's what I... Yeah. You know what?
• 19:44
  I guess that that that the floridness didn't bother me as much, but, you know, maybe it takes one to know one on that regard. I I... You know, I'm I'm I'm very
• 19:52
  Noah Harlan: angry my spouse. It just was delightful. It struck you. Was... Yeah.
• 19:56
  Right. Yeah. Yeah. Yeah.
• 19:58
  Adam Leventhal: Mean, as the laity. I did appreciate it.
• 20:00
  Bryan Cantrill: As the laity. Exactly.
• 20:02
  Noah Harlan: As the laity.
• 20:03
  Bryan Cantrill: Well, and and and you should know that that that... And and I... I'm not... I I like Sidney Decker. Adam loves Sidney Decker.
• 20:10
  Is that fair?
• 20:12
  Adam Leventhal: Not really because you don't like Sidney Decker. I don't know that I love it, but I but I love... I I did love when Noah brought up Sydney Decker. Like, I couldn't... Like, there was a moment that went through Bryan's mind just just now, Noah, with, did Adam have time to bribe Noah?
• 20:33
  Like, is that a thing that could have happened to get us to Sydney Becker?
• 20:37
  Noah Harlan: I heard a snort. I just didn't know why.
• 20:39
  Bryan Cantrill: Exactly. Yeah. That that... That's a very strange and specific theory that you've developed, Adam. I mean, that's like...
• 20:44
  That's... Wow. Okay. I'd wait. Yeah.
• 20:45
  Maybe we should be asking some questions. We, the lady, should be asking some questions here.
• 20:50
  Adam Leventhal: The lady.
• 20:51
  Bryan Cantrill: The lady. There goes the lady again. Yeah. No. I I I think
• 20:55
  Noah Harlan: my Venmo to private so, Adam's payment doesn't come through on this.
• 20:59
  Bryan Cantrill: Exactly. That's right. Well, it's little too late for that. I've already got the screenshots. And I I know you're a cheap date.
• 21:05
  I was a little bit surprised, honestly. Next time, you know, the the... But I I thought that the... I mean, I I liked the... And I would actually be really interested in the in the city...
• 21:15
  My only... Oh, here's my only complaint with the city decker just to get it it a... Get it out there. The book Drift into Failure picks on commercial aviation a little bit too much. Mhmm.
• 21:25
  And Mhmm. Commercial a v commercial aviation to me is like the the most canonical example we have of a system that is learning from every failure and not repeating them. So it's like I I I have a... I just get a little bit defensive about... Maybe too defensive about commercial aviation in particular.
• 21:41
  Noah Harlan: I actually found drift into failure a a less... I thought it was sort of like an applied version of safety anarchist, and safety anarchist is the one that I've actually read through a couple times. Like, safety anarchist to me is like the the sort of fundamental intellectual underpinnings of what he's saying. And then he kinda go... And I think this is correct from a sequencing perspective.
• 22:01
  He then goes and like applies a bunch of those ideas and drift into failure. But like safety anarchist introduces like these really interesting ideas about like the sort of the kind of punishment of synoptic legibility and like what... And it's Goodhart's law before Goodhart's law in a lot of ways, which is like, you know, you measure the things that you care about, but like the process of trying to impose a system on that is going to intrinsically create these boundary lines where there's chaos. And I love the use of Paris as a description of that and like Oceane coming through and organizing the city, but then you wind up with the outside the city. And we're seeing that actually in the last couple of weeks, was just talking to someone about it last week where it's like, and then the the residents of that will come fighting into the organized part of the city.
• 22:56
  And those boundary lines are the same as like the bottom of the Kenevan framework. Like a lot of people overlook that little swirly thing that's between clear and chaotic, But the swirly thing is actually like a piece of Snowden's model, which is to say that when something in the... They now call it clear because they thought that calling it simple was offensive. But the simple domain, when something breaks there, everything is so tightly constrained that it drifts into chaos. That's the kind of the drift into failure that can happen, which is you think that you've organized a system and you put it together so well that it couldn't have a problem.
• 23:36
  And then when a problem emerges, you have no way to handle it because you are no longer applying expertise, which is like what the complicated and complex realms are all about.
• 23:47
  Bryan Cantrill: Yeah. This looks really interesting. And I... The... Okay.
• 23:49
  And I also feel that the safety anarchist... This is exactly what I needed because I actually felt with Sydney Decker. I'm like, I kinda... I like these ideas. It's the the concrete example of commercial aviation that is just rubbing me the wrong way.
• 24:00
  And it's kinda like when you're, you know, Adam, when you're reading LLM generated stuff that actually is like, I think there's like a good prompt that was used for this LLM authored article. And I would like to just read the prompt instead because I can't. My brain is, like, seizing on every sentence, and I'm... It's not allowing me to get to kind of the underlying ideas. And I I kinda feel...
• 24:20
  This is kinda what I need. I... From City Decker. So this is this is... Safety Anarchist.
• 24:25
  I'm actually... I'm gonna put it on the list. I I I Have
• 24:27
  Noah Harlan: you have you not read it?
• 24:29
  Bryan Cantrill: I've not read it. No. I've only read Drift Failure, and I haven't read all of Drift Failure. I read about half of it, and then it got thrown across the room. And which is extremely unusual for me.
• 24:38
  I... I'm I'm a finisher. Usually, I, it has a often a corrosive effect on my marriage that I'm a finisher or I'm actually forced to finish books outside where I can't complain about them loudly, because I will be reading it and complaining about it. And it's like, do... Like, you can't do both of those.
• 24:54
  The... So no, I did not. I didn't finish it. And I would I would actually like to read it. So this is gonna be good.
• 25:01
  And so how... And but so that's interesting that these kind of books on systems thinking were... Was that informing you as you were diving into this stuff? Or or or was this like... How were these the these kind of frameworks or were they?
• 25:18
  Maybe they weren't actually very helpful as you're going into like the really thorny underbugs. I mean, the wicked problems are typified by the fact that they are, they're hard to even define. I thought I thought it was one of the interesting observations was that the... In trying to define the problem implies a certain solution to it. And and then using canonical examples like poverty and homelessness as wicked problems because they're so multifaceted and it's...
• 25:44
  We have a hard time even agreeing what the problem is, let alone what the solutions are.
• 25:49
  Noah Harlan: One of the things that was interesting... So I I I did not come from the social service ecosystem. I didn't come from government work. You know, I had... My background is as...
• 25:58
  You know, I studied CS. I worked in film for a while, and then I spent most of my career working in technology in various places. And like my, I overshare this mantra, but like, I used to tell people like, oh, well, like, our goal is to find things that suck and make them suck less. Like, if you do that, you have built a really good thing. It's like the classic, like, vitamins and painkillers.
• 26:27
  Like, vitamins are nice and, like, yeah, if you got extra money, you'll pay... You'll buy vitamins. But like when you're in pain, you pay for a painkiller. And what we were seeing in this space is that it's just like there's so much pain, you know, literal, metaphoric, emotional, every every direction. And it's a really...
• 26:49
  I mean, you're talking about systems thinking. It's almost impossible to apply systems thinking because it is so labyrinthian. And that that was what was really resonating with the the article you shared, which is that, you know, the inputs and the outputs to any particular part of the system are incredibly hard to isolate. There's all of these confounding factors, and there's even just like fundamental misunderstandings that go from the micro to the macro. So at a macro level, and let's stick with homelessness for the moment, a lot of people think of it as a state problem.
• 27:26
  Every year The US counts its homeless population, and we do it on one day. It's called the pit or the point in time count. And people go out all around the country, and they go out and they go, you know, into parks and under bridges and to shelters and they add it all up and that becomes the official record of life. There are this many homeless people in America. But that is how many homeless people there are in America on that night.
• 27:54
  The problem is that homelessness isn't a state problem. It's a flow problem. You have new people entering homelessness every day. You've got people exiting homelessness every day, and you've got people oftentimes returning, so cycling. And in a world where it's constant...
• 28:11
  So the number of people that experience homelessness in a year, and homelessness is also quite broadly defined, but is vastly greater than the pit count because many people experience homelessness for a few months. Homelessness might be, I have no place to stay. I'm going to crash on friends' couches. Homelessness is not always... I think for many people, if I just say to you, imagine in your mind a homeless person, and I'm not gonna use the person experiencing homelessness.
• 28:42
  Imagine a homeless person. You picture someone on the street. Street homelessness, unsheltered homelessness is a tiny portion of the people who are experiencing some form of housing insecurity and the subset of those who can be categorized as homeless. And so what you quickly realize is that it's this really complicated landscape of people in wildly divergent situations. Like, you know, take L.
• 29:07
  A. You know, on the night of the fires in L. A. Where like Pacific Palisades burned and, you know, there was another community and forgive me for forgetting the name. There was a historically black community.
• 29:22
  Out to The US, that also burned. All of those people were homeless. Now many of them had the resources where they could just go to a hotel or, you know, they could find a place. But, like, some portion of those people didn't or couldn't. And all of a sudden LA's homeless population dynamic completely changed.
• 29:44
  And it wasn't just people who are on the street. And so when you start unpicking that, you're like, oh, okay, well where do you even start with this? And in our case we were kind of granted this limited thing where it's like, hey this is a common problem that occurs in homelessness, but when you start to expand your aperture, you realize that it's like everywhere. I have two kids in public school in New York City, and when one of them switched schools, I had to find a birth certificate. I had to find a proof of residency.
• 30:17
  So I needed utility bills to show that I lived at the address I did. So we were zoned for the school that we're zoned in and blah blah blah. I can go down a whole rabbit hole on education in New York City, but that's that's another podcast for another day. Anyhow. And, you know, and I'm I'm pretty organized.
• 30:35
  Like, I I I literally have, like, one laptop sitting on top of another laptop as I'm talking to you. Like, I, you know, like, I have resources. I have, you know, Dropbox and, you know, Google Drive and, you know, blah blah blah blah. And it was a mess. When I tried to get a real ID, I had to apply for a new social security card because somehow I couldn't figure out where mine was.
• 30:58
  And I could do that really quickly because I had figured some of those things out. But the problem is that, like, as you move down the socioeconomic ladder, the number of times we're asking people to do these things goes up, and the number of different things we're asking them to do goes up. Like, there was a program that we were working on early on... Working with early on in LA, and they had a housing support program where they were providing certain types of housing support to some people in South LA. And it was a wonderful nonprofit called Fathers and Mothers Who Care.
• 31:35
  And they had a particular program that for some insane reason required people to requalify every two weeks. Like, every two weeks, they had to submit new documentation to prove that they hadn't found proper housing and needed and were at a certain resource level. And the thing that became also infuriating... Sorry. And then I'll sort of step off the the soapbox for a second.
• 32:02
  But one of the things that became really infuriating, like, you know that meme of, like, it's Rocky and Apollo Creed's hands and it's like the, you know, party a, party b, and then like the hands clasping is like the thing they agree on.
• 32:19
  Bryan Cantrill: What
• 32:20
  Noah Harlan: became a... Yeah. So what became apparent, I've actually made this meme many times over, so I need to just keep it on my desktop and maybe later I'll share it in the chat if I have a moment. Like the left and the right, and I use those terms very, very broadly here. The one thing they agree on is administrative burden.
• 32:38
  They both love it. And they love it for very different reasons, but they both love it. Maybe maybe it's from creditor, but it's it's it's it's it's, two Yeah. Two giant Into Like like... Yeah.
• 32:52
  Two big arms, one white, one black, holding hands.
• 32:56
  Bryan Cantrill: Right. Exactly. And so the the it... Would they... Well, in addition to...
• 33:00
  Also, I think, being against data centers, think, is the other thing that this, like, at the moment is transcending Yeah. Yeah. All all political boundaries. Interesting. Right.
• 33:09
  The one... I mean, because it's... I mean, who wants administrative burden? I mean, the... Just like the way it's even, like the the way it's branded.
• 33:16
  Yeah. It sounds pretty clear. You don't want administrative overhead. So presumably, that's helpful to have that kind of focus at least, political focus.
• 33:27
  Noah Harlan: No. No. No. No. I'm saying the opposite.
• 33:29
  No. I'm saying that they both
• 33:30
  Bryan Cantrill: want it. They both want administrative burden. I'm sorry.
• 33:34
  Noah Harlan: They both want it. They're both proadministrative reasons. Yes.
• 33:38
  Bryan Cantrill: Oh. Oh. On one side Oh, no. I was sad for us.
• 33:43
  Noah Harlan: Yeah. Yeah. Sorry. Yeah. I led you down the garden path and then I I
• 33:47
  Bryan Cantrill: You really did lead me
• 33:48
  Noah Harlan: down the
• 33:48
  Bryan Cantrill: garden path. The...
• 33:52
  Noah Harlan: On some level, like the making a policy... Making a particular thing hard to get unless you really, really deserve it is this weird thing that they both agree on, but for slightly different reasons. Like one group
• 34:08
  Bryan Cantrill: I was gonna says get dark enough. Yeah.
• 34:10
  Noah Harlan: Yeah. One group says, look, you need to... We wanna make sure that this program is really narrowly tailored. So unless you meet these like 32 criteria, you're not in this program, you're in a different program because we want to help just this group of people and then we're going to help this next one and then we're going help this next one. And then on the other side, you're like, we want everyone not to be in these programs unless they really, really are the right people to be in them.
• 34:36
  And so, yeah, you wind up in like Texas with a 57 page snap application. And there've been people who've been working against this, you know, and writing books and so forth. I think that we're making some progress on it, but by and large, those systems are just head desk when you start to encounter them. And you're like, why? It doesn't need to be this way, but here we are.
• 35:03
  Bryan Cantrill: That must be very frustrating because it really should be the one thing that we agree on. These program... We wanna... These programs to be... That not have administrative burden.
• 35:11
  We kind of would... We ideally would like the opposite, which I'm sure was your feeling as a... Someone who's kind of coming in as a... As an outsider, relatively speaking, to these kinda... To these...
• 35:22
  It must be very, very frustrating.
• 35:24
  Noah Harlan: Yes. I can only imagine. Yes. I
• 35:27
  Bryan Cantrill: Okay. So It is. So how do you... Given all of that, like, okay. So then how do you, like, find the the the the the kind of the lane to be able to make some progress on it?
• 35:40
  Because I mean, the flip side of this is when you can knock some of this stuff down, you can have like a real and immediate effect on a on a person's like real life. I mean, this is this is the stuff that is like much more relevant than many of the things that we develop in in technology.
• 35:58
  Noah Harlan: Yeah. So a lot of it involves stepping back and thinking about the user. And I'll actually shout out, and I think I I saw him in in the room, Phil Robinson, who I've had the tremendous good fortune to work with for the last, oh gosh, I think seven years now, six, seven years. And he was head of UX for us and we're still working together. We've been through three companies together, which is great.
• 36:32
  And it starts with really being able to think about what is the experience of these individuals? And that was the fundamental problem of the original open source data locker, which is like, you were thinking about a solution, you weren't thinking about the experience of your user. And your user is not going... Both the administrator side, which is like, I'm sorry, but let's say you could stand this up in New York City. Well, guess what?
• 37:00
  The other city we were working with was Baltimore. Baltimore doesn't have a very large IT department. You know, they don't have a $100,000,000,000 city budget like New York does. So what are they gonna do to stand up a bunch of IT infrastructure and run open source software and keep it patched and provide support and all of that? Like, they're not going to.
• 37:24
  And that's kind of my, the gist of my rant on public interest technology where they're like, source, I'll save the world. I love open source. I have spent years of my life dedicated to building open source communities and and seeing them thrive, but it is the wrong tool. And every time I see another, like, you know, book come out from a think tank about, like, the future of, you know, government is public interest technology and open source communities working together to build application layer software. I'm like, point to the application layer software that actually exists, that is open source, that people enjoy using with the exception of, like, signal and developer tools.
• 38:04
  And it's pretty rare.
• 38:05
  Bryan Cantrill: Interesting. So so what you mean by that is not that, like... I mean, look. We love open source. Open source is great.
• 38:10
  But the idea that, like, a technology being open source is gonna facilitate us being able to use it for these public interest functions. It it leaves... There's a there's a lot that's being left unsaid there in terms of of the way this is operated, way it's actually productized. I mean, because open source projects are great as kind of like these infrastructural components, but they are the things that which you're gonna build an actual end user facing application out of. They're not...
• 38:35
  Are... They themselves are not a replacement for actually building that.
• 38:39
  Noah Harlan: No. And so, what ends up happening is like, they're not thinking about the users on both sides, like the operators of the platforms, the organizations, and then the individuals. So then, when we started looking at this, we wound up partnering with an amazing organization in Los Angeles, a different one called HopEx. It's the largest homeless service provider in the country as a private CBO. Just to give you a sense of their scale, as an individual nonprofit, they help more individuals experiencing homelessness every year than any other municipality anywhere in the country with the exception of LA County and New York City themselves.
• 39:20
  And they... So they cover all of Los... South Los Angeles. And one of the things that became apparent really early on is they they loved the idea that this really centered their clients and was empowering their clients to have access to their own data. And they were like, look, we wanna use this.
• 39:38
  But what became really apparent early on was that unless we could build a bunch of tools that were going to be useful to the organizations, they weren't going to use it because they've got too much other stuff they're working with. The individuals are super untrusting of, you know, they're not just gonna download random software and that the best path to helping individuals organize this information was the people they already trusted and were working with. So these frontline case managers who in many cases were people with lived experience, like people who had actually been through getting housing stability through a place like HOPEX. And then we're now working at HOPEX to help the next group of people. And so we started focusing more on that side of the platform and eventually what Keap became was a case management system for these nonprofits across the social services ecosystem.
• 40:39
  So not just homelessness, but really everything from like doulas to housing to nutrition to substance abuse, you name it. But what was amazing, and this is just such an incredible story about the power that you can have on one person's life, About six weeks after Hobbitx launched with it, we were on one of our regular calls with them. We're spending a ton of time with them trying to figure out like, okay, what's working, what's not, what could we build next, etcetera. And they said, hey, guess what? It worked.
• 41:14
  It freaking worked. And we're like, okay, what? And they said, had this guy in our shelter. And when he had come in, we did his onboarding with Keep. We stored his documents, and he was out and about, lost his paperwork, and he got stopped by the cops.
• 41:32
  They said he looked like someone they were looking for, you know, whatever that means. And they were going... And he's like, I'm not that guy. And they were like, well, prove it. He's like, well, I don't have any of my ID on me or anything.
• 41:45
  And they were gonna arrest him and take him in until they could figure out who he was. And he's like, wait a second, wait a second, I have on my phone. And most of these people do have phones and they just may not have a prepaid plan. And he was like, look, I got this stuff on my phone that they put there when I entered the shelter. And so he pulls it and they're like, all right, well show us what you got.
• 42:08
  He pulls out his phone. He's got a picture of his ID. He's got a picture of a copy of his intake form, which had the contact information for the case worker that he was working with. They call the shelter and speak to the case worker. The case worker is like, yep, this guy is who he says he is.
• 42:21
  We saw the ID when he came in. And they let him go. And he went back to the shelter. Now it was a Friday, and he was gonna spend the weekend and probably part of the next week sitting in jail in LA County. And LA has a three night rule so that if you give up your bed in a shelter for three nights, they release the bed and make it available to someone else.
• 42:45
  So he would have come out of jail having done nothing wrong and would have been back on the street. And it's hard to fathom how far back a fall that is because just the number of bad outcomes that happen at that point is is staggering. So he stayed in the shelter. About a year later, we're talking to Hopix. You know, there's a whole case study, Google it out there on this, that AWS published.
• 43:13
  About a year later, we're talking to them, and they're like, hey. Remember the guy, like, from right at the beginning when we first started working with you? The guy where, like, having his documents helped him in a situation that was not related to housing, the very premise of what we were trying to do? We're like, yeah. We just got him into long term stable housing.
• 43:30
  He's stably housed. And they were like, if he had gone back out on the street, that would not have happened. And so that to us was like, it was just, you know, it was just that pure hit of exhilaration of like, oh my God, I can't believe this actually worked. Now, does that on its own solve big problems? No.
• 43:53
  Not necessarily. But it was an indication that it was a different way of thinking about it, which is like, let's not build a system for government. Let's not build some... You know, let's not try and be Accenture or Tyler Technologies or whatever and try and like get the billion dollar contract. Let's actually work on the ground and try and solve some of these problems from the bottom up with a really specific focus on the user experience of the individual.
• 44:20
  Bryan Cantrill: Yeah. And I think that's so interesting because you think about like... This thing about like about, life document management and just having this cascade of problems when you don't have it. I'm just describing where it's like, now because you were in jail over the long weekend and you lose this, you lose that, and now it's like you... Things begin to to, to unspool.
• 44:40
  And, you know, how do you quantify what the actual effect is of life documentaries? Like, well, it's hard to quantify because not having this is so acute and and and can kinda spiral in so many different ways. I mean, I love that taking that from the the really... From the the grassroots and kinda the ground approach. I mean, one of the things you you kinda had mentioned to me is this some kind of confusion for those that are, on house or suffering from homelessness, or or housing insecurity, to what degree do they have a phone?
• 45:12
  And I... Because I think that, I mean, is something that that I think it's interesting to me that that this is like the kind of disconnects that exist when you're tackling this problem, because you would think that like... Do wanna describe that a little bit because
• 45:25
  Noah Harlan: that that was... Yeah.
• 45:25
  Bryan Cantrill: Yeah. Yeah. The disconnect was surprising to me.
• 45:29
  Noah Harlan: So the number of times that, like, I've encountered someone where I've just wanted to, like, scream in in the last five or so years working on this is hard to count. But one of them was... We were having a meeting with... It was a woman who worked, I think, for New York Health and Hospitals Corporation, which is the the city run hospitals in New York. And she was responsible for a bunch of their housing stuff.
• 45:59
  And she said, you know, we're talking to her about Keap and how it could be used, and there was a bunch of interest in the org in potentially using it for for some case management stuff. And she was like, yeah. But the problem is that, like, homeless people don't have phones. It's like, no, that's not true. It's not true at all.
• 46:21
  Actually, an astonishing number of them do. And we we sort of dug into it to try and figure out why she thought that. And eventually what we realized is that, like, if you speak to... Like, this this became, like, one of those great, like, shibboleths where, like, you know, know, the the scene in Inglorious Basterds with the three fingers where it's like, oh, I I see Where, like, I could tell if you worked in policy around homelessness or if you worked boots on the ground was whether or not you thought people had phones. And the difference was that, like, individuals experiencing homelessness have a pretty high rate of access to phones.
• 47:02
  What they don't have is postpaid cell phones like most of us have. So they get a SIM card and they use it for a while, or nowadays actually there's so much ubiquitous free WiFi that they actually don't even have phone numbers. They just use it via WiFi. So like when they rolled out Link NYC in New York, which were these, online kiosks, had free WiFi and you could tether your phone to charge it, they suddenly became these... At night, it was like moths to a flame.
• 47:34
  You would get these little gathered homeless encampments around each one of them, much to the chagrin of the Link NYC people. And what it was is they're like, hey. We can charge our phones. We can get free Wi Fi. We can do, you know, we can do whatever we wanna do, you know, for better or for worse.
• 47:51
  And the administrators, though, they're looking at the reports, and the reports show that nobody's got a phone number. So they're like, well, there's no phones. They're like, you know, 5% of the people who come into our shelters have a phone number. Therefore, 5% of the people have phones. And it's just those great disconnects where it's like, yeah, but you haven't spent the time actually understanding your user because frankly, on some level, you don't care about your user.
• 48:15
  You care in the abstract, but you haven't cared enough to actually go and spend some time figuring out, like, why are these people here? What is their motivation? What what do they want out of this? And and we had to learn a lot of this the hard way. Like, one of our very first projects was up in New Hampshire working with Easter Seals, and they were working with veterans who were experiencing chronic homelessness.
• 48:39
  So this means you're cycling in and out of homelessness over a long period of time. It didn't work. It just flat out didn't work. And the reason was that what we discovered is, first of all, very small population. And a lot of these guys were living in the woods and they were suffering from serious mental illness and they did not want a phone anywhere near them.
• 48:59
  And then they would come in every so often and they'd be like, Hey. They would cycle and they would say, Hey, I wanted some support. I want to go to the VA, but I lost my... I can't prove that I'm a veteran. I need a new DD214 or whatever.
• 49:13
  And they would help them for a while, and then they would kinda lose them for a year or two years, and then they would come back. And, like, that was just a population that this was not a solution for, and at least the original, like, data locker piece of it. So, you know, there's a lot of learnings.
• 49:29
  Bryan Cantrill: And this is, like, what... Also one of those aspects of Wicked Problems. Like, there's not gonna be a solution to this stuff. I mean, you gotta be content with it. Alright.
• 49:36
  So this... The the... We're we're gonna have to pick and choose a little bit. We're gonna have to have a solution. There are gonna be multiple facets to solutions as well.
• 49:44
  Yeah,
• 49:45
  Noah Harlan: and the inclusivity arguments are infuriating. They're like, Hey, yeah, it sounds like a good solution, but actually like 15% of people this won't work for, so we're not going to do it. And you're like, But you could be helping 85% of the people, and that would make a big difference. And then we could spend other resources that are now freed up on the 15% that need it.
• 50:06
  Bryan Cantrill: And you also... In terms of, like, picking and choosing a little bit too because you also... I'm sure you've got folks that are beginning to to teeter where their their problems are beginning to compound. And they're gonna become not only are they gonna have much more hardship, but they're gonna be much more expensive societally if we don't kind of provide the infrastructure with the kind of the right... Because there's a kind of a timing element to this.
• 50:30
  And it's like, I mean, can you speak to that a little bit?
• 50:33
  Noah Harlan: Yeah. So so actually, there's there's three populations, and it's so funny. Every time I say this, I think that... I said this many, many times in front of folks who are from these three communities that I'm about to describe. And each time I'm waiting for them to get really angry at me making this comparison, and every time they're like, yep, that's exactly right.
• 50:55
  So there's three groups of people that have remarkably similar dynamics that you would never associate with each other. Children coming out of foster care, individuals coming out of incarceration called reentry. And this is generally long term incarceration, not like you were in local jail for a couple weeks, but like people who were in prison, an individual is coming out of the military.
• 51:22
  Bryan Cantrill: How interesting.
• 51:24
  Noah Harlan: The commonality between those three is that all three were in kind of different forms of custodial environments where they were not, generally speaking, responsible for where were they gonna sleep at night, how are they gonna get food, etcetera. Like, those were largely addressed for them in some way, shape, or form. Not necessarily in a great way, but it was addressed. And then they come out, and if they don't have access to stable housing and employment pretty quickly, things start to slip. Now the dynamics are different for the three of them.
• 51:59
  So if you come out of prison, you go through reentry and you don't get stable within about forty five days, the likelihood that you wind up back either because you violated parole or committed some new offense is extremely high. So that's a lot of recidivism right there, which is we just didn't set people up for sex. And so if you work with people and say, look, let's make sure that before you walk out the door, you have access to all your documents. You've been plugged into programs. You've had social referrals made so that there's a support network out there for you.
• 52:32
  The outcomes are completely different. The same thing with foster care, but at a slightly slower rate. So the number of kids that mature out of foster care and along the way they've been in so many different places, they don't have their birth certificate, it was lost, you know, kind of three foster homes ago and they don't have a copy of their GED or their high school diploma and so on and so forth. We're just like, great, you're an adult now. Good luck.
• 52:58
  And there again, you see a similar dynamic. Now they often are in a slightly better trajectory, but their journey into, you know, drugs, crime, you know, and falling down, is very, very real and acute. And then for the for the veterans, the military has, like, what they call, like, a TAP program, the transition assistance program. And so they send you down to like Fort Hood in Texas or whatever. And I'm not really talking about like, you've spent your career 35, you're an NCO or something like that.
• 53:35
  I'm talking about the younger soldier who's been in for three years or whatever, they've done their job and now they're transitioning out. And so like you go through this week long program where they're like, we will teach you life skills in this one week. And then they go and they wind up like couch surfing at their buddy's place and they hang around Austin for a while. And their journey is usually like eighteen months to bad outcomes, where if they don't figure it out in about eighteen months, then you start seeing all the same dynamics occur that you see in the other two populations. There you have your problems, and it's gotten a little better in the past couple of years.
• 54:15
  The US has made huge strides in veteran homelessness. It's actually the one area where the statistics are wild. If you really do surround people with a lot of resources, you can fix this. Veteran homelessness over the last ten years in America has plummeted. But like if you need a new DD214, and that's this file that proves that you are a veteran, so like it kind of unlocks all the support services, you apply for it, but then they send it to your last place of residence, which is your mom in Minnesota that you haven't been to, and it takes two months to get there.
• 54:47
  And so in those intervening period, you're not getting services potentially. Again, it's gotten a little bit better. They've been doing more technology investment to make that process faster, but still these transitions are just the like, these kind of like threshold moments where things go awry. And if you can intercede in those moments and sort of provide some more infrastructure that can help people. That's a combination of like, do you have access to your information?
• 55:13
  Your core vital documents, do you have some place where those are? Because those fundamentally unlock everything. And if you... In addition to that, have you been plugged into other services, which is what Find Help, which ultimately acquired Keap does, which is these social referral networks where it's like, hey. We're gonna send you over to this organization that'll help you with this piece of it.
• 55:36
  This organization that's gonna help you with that piece of it and this organization, and we're gonna track those to make sure you get those services. But it's all being done kind of like if you wait for the government to figure out how to connect all the different systems at all the different levels, like local, county, state, federal, it's just... You're... It's never gonna happen. And there's two places in the world that have done it and have done it really well.
• 56:02
  And I wish there were more kind of taking of lessons from them.
• 56:05
  Bryan Cantrill: Two different Scandinavian countries?
• 56:07
  Noah Harlan: No. Neither one is Scandinavian, though one is close. Estonia, has done it spectacularly well. You wanna see something really fucking cool? Go check out what Estonia did when they left the Soviet Union and what e Estonia looks like.
• 56:25
  And the other place that's done it is India, which has done it more recently, but they rolled out a digital ID called card called the iAard card, and you can access virtually any government service anywhere in the country, and then a lot of private services like banking and so forth with that digital identity platform.
• 56:43
  Bryan Cantrill: Yeah, interesting. Okay. So in terms of the actual, so you developed this technology or, and developed, I mean, you're developing not just the technology. I think that's part of your point about the open source argument. It's like, you need to actually develop something that's actually gonna serve the user.
• 56:57
  So an entire application that the user can actually use that is gonna give them what they need. And then the customers for that end up being various agencies and nonprofits and NGOs. I mean, obviously, HOPEX ends up being effectively a customer of that or a partner for that.
• 57:17
  Noah Harlan: Yeah, that's exactly right. And really it was a lot of these community based organizations that were like, we need a tool. Like half of the CBOs in America are still managing their clients with paper or spreadsheets and don't actually have better tools to do it. And there's a variety of solutions, but you get something like Salesforce for nonprofits. You know, like...
• 57:47
  So, like, you know, six foot eight penny off comes marching through the door and says, I am bestowing upon you the glories that are Salesforce. And, we will even give you someone
• 57:57
  Bryan Cantrill: This does not feel like you're speaking metaphorically. I did... This is one of those where I can't tell if this is brilliant satire or not, but this feels this feels extremely literal where it's like, have like... I have now granted Salesforce upon you. And I...
• 58:10
  I'm leaving, by the way. So good luck with it all. But don't worry. There's this group of consultants that that have I introduced you to the consultants that that can consume your IT budget, getting you on Salesforce for nonprofits.
• 58:23
  Noah Harlan: And so there's a lot of these kind of everything boxes. Like Salesforce is amazing, but it's like an everything box. It's why you need someone to like, you need like teams of people to come in and implement it. Like if you're a really big, complicated, sophisticated organization, holy cow, it can do big, complicated, sophisticated things. Like...
• 58:40
  You know? But if you're a $200,000 total budget CBO made up of, like, four volunteers, none of whom ever went to college and just have it in your heart that you want to help your community, yeah, Salesforce isn't helping you one lick. There's just no way in which that's going to be useful. I'm not making this about Salesforce. It's true about a lot of the different that are out there.
• 59:07
  We're happy to integrate with those type of platforms, but they're not built of and for these types of users. Again, coming back to Hopix, when we went out and spent time with Hopix, what we realized was the frontline staff at Hopics are not social workers. They're not people who had master's degrees in social work. They didn't spend four years in college, two years getting a master's, know how to use complex systems. So they're super technology literate.
• 59:37
  They've used SAP or Oracle, or maybe they worked in a hospital, so they used Epic systems or something like this. They're not. They're people in a community that are like, Hey, this is a job where I can help some other people in my community. So we had to make sure that every aspect of what we were doing was simple enough that you could just pick up and use it, and that it was intuitive, and it was actually really, really simple, not really complicated. So you say, Oh, well, guys created this technology, which is very flattering.
• 1:00:14
  Thank you. Like, I... I'm hesitant to call it a technology. Like, there's nothing inside of Keap that is at all like, wow. We really broke some ground here.
• 1:00:27
  It is all about how you stitch those pieces together into something that is useful and meaningful for its users and that solves a problem. But if you looked under the hood, our infrastructure is pretty straightforward. When a new engineer comes on our team, they're committing PRs in a couple of days because it does not take a long time to figure out our stack.
• 1:00:51
  Bryan Cantrill: Right. So much the better. I think that the breakthrough is in that you're actually making something that's actually useful and can be operated.
• 1:01:00
  Noah Harlan: Yeah. And you just have to continuously remind yourself of that. Like, don't let yourself get lured into complexity.
• 1:01:08
  Bryan Cantrill: Right. Well, I was gonna ask... Okay. And just because, you know, and I I don't know, Adam, if we've got a specific time for it or not, but I just can't resist myself asking. How...
• 1:01:19
  The... One of the, utopian or optimistic views of of LLM adoption is that it will... This kind of Jevons paradox where we will have many, many more people authoring software, which have much more software, software that would, that was not kind of previously economically viable to write. And it does feel like there would be like, this is the kind of places where you would have that, where you would be able to have people put together systems that they might not be able to put together otherwise. Is that...
• 1:01:50
  Is there any truth to that? Mean, this is this is... Does it help you accelerate in terms of the systems you're developing? Or actually, should say, not accelerate, more like develop... Does it help you develop better systems?
• 1:02:01
  Noah Harlan: So look. I... I'm I'm a pragmatist about LLMs. Like, I think they're pretty amazing and can do some pretty amazing things, but I also, like, I think that the community around them is just such utter nonsense. I think that the problem here...
• 1:02:20
  So when we started Keap, I raised money. We raised money from impact oriented VCs, so we had some very cool investors. Our two biggest investors was an organization called Acumen, Acumen America, and Ali Corp. Both of which were incredible and incredibly valuable. Ali Corp in particular, Kevin Ryan, I think you and I were talking about him the other night.
• 1:02:49
  Like, talk about, like, the legend of legends, you know, just generational run, but also the nicest guy. And I think one of the things that they said is they're like, hey. You've got a pretty big team for an early stage startup. And I said, yeah. That's because we can't screw it up from day one.
• 1:03:13
  Like, we are... The very first person that we help on our platform is going to be someone who is in the most vulnerable place in their life, and we are securing their most important documents. Like, I can't let that information get out. We need to ensure that our security, our privacy, our focus on accessibility is all there from day one. And you can't just be like, hey, make sure that no s three buckets are, you know, open to the public.
• 1:03:47
  Make no mistakes. Go. And so I couldn't just take a typical, like, person who's done, like, coding boot camp and be like, well, it's also... This was 2021, before kind of the real explosion that happened about a year later. But I think there are lots of really cool things you can do with LLMs that, you know, if if you are careful and thoughtful about it, that can accelerate and can make these systems easier to interact with for our customers, you know, for these communities, like their ability to interact in plain language and be like, show me a report of, like, x or y, and then doing that in a way that's safe and secure and private and all of those things.
• 1:04:36
  Yeah, that's pretty profound and I think will make But I'm as surprised by Jevons paradox that you were describing where it's like, oh, there'll be thousands and thousands of people all solving this problem in ways that are secure and private and scalable.
• 1:04:55
  Bryan Cantrill: Yeah. We we just... I mean, obviously, you're you're among friends in that regard to this whole, this this iconoclastic idea that LLMs are tools that that should be used as such as opposed to kind of like closing this agentic loop and then go wild. Because I do think that you've got... I mean, just your point about making sure that you keep the system simple.
• 1:05:13
  You don't wanna overburden it with technical complexity when the when the really the problem is the the way we kind of develop the system interact with the user. But I think you also mentioned the security. I mean, the... These are life documents you're talking about. So this is like big bull's eye painted on them from an identity perspective.
• 1:05:31
  And like, the security is actually very important for this. This is like right up there with your banking information, because it can be used to get your banking information. So it's like like the... This is not something that you wanna vibe code into existence. You really need to take...
• 1:05:47
  Because the stakes are really quite high. And the stakes are quite high and the folks that would be afflicted by this are are, especially when you're dealing with the populations that are unhoused and so on. These are like, these are vulnerable populations. Like the last thing you need is like, my god, my identity got stolen. And now my, you know, I'm kind of like beginning to to kind of claw my way back into housing and all.
• 1:06:08
  I've got you just like your your story about the person who was gonna be arrested. And, I mean, it's like, the security of this is not a theoretical problem.
• 1:06:16
  Noah Harlan: It's No. No.
• 1:06:17
  Bryan Cantrill: It's gotta be made secure.
• 1:06:18
  Noah Harlan: It's... Yeah. Exactly. It it it... It's just, you know, the the thing that is keeping the head above water is so thin for so many people.
• 1:06:33
  If you are going to become a load bearing structure for them, you have to be committed to bearing that load. You know? You can't play. You you just can't... Like, the move fast and break things doesn't apply.
• 1:06:49
  Like No. That's my moral framework on it. And so it's it's it's just a different model. Like, because, you know, I mean, Facebook screws up and like, oh, my videos don't look fast enough on Instagram. You know?
• 1:07:05
  Like, oh, hacked my Instagram and they posted embarrassing stuff. Like, those are bad for sure, and I don't want that to happen to anyone. And I and I know that they spend a lot of time fixing those things, but like the consequences and a lot of that's really, really low. But there's a reason why like JPMorgan Chase has like a massive security team that is like, you do not want people accessing these accounts. Because it's like load bearing structures for organizations and, you know, whole swaths of our economy.
• 1:07:36
  I'm not not to necessarily put them up on a high hill, but, like, they're just facing different scales of consequence.
• 1:07:46
  Bryan Cantrill: So, I mean, I think for kind of a younger technologist that wants to get involved on some of these wicked problems, because I do think that the... And one of the things I had posited online is that if wicked problems are gonna be the ones we've got left for sure. I I I don't see anyone talking about the... About, you know, AI solving homelessness, right? I mean, that are are solving...
• 1:08:09
  I mean, why am I saying this to myself? Of course, people are saying this. But I guess we should... We could chime for a future episode when I go to another unrelated risk. But I think it's obviously ridiculous.
• 1:08:22
  Right? And I think that like one of the frustrations that I have... We have with some of these kind of forecasting arguments is they don't spend enough time in the real world with real people that have real problems. And they kind of lose track of our fundamental responsibility as technologists, is to develop technology that makes a real difference in people's lives. And working on these wicked problems on the one hand, very multifaceted and thorny and can be frustrating in all the ways they're frustrating.
• 1:08:51
  But the other hand, gives you, you know, these these anecdotes of this person who doesn't even, you know, you presumably don't even have a... Like, they should and should be kind of unaware of the work that went on to give them what they needed. For folks that wanna get involved, what are some good, like, entrees into this? Like, how how are are there there companies, organizations, should people get involved if they want to get involved? They're like, this is, you know, I've done the FANG thing.
• 1:09:22
  I am suffering from on WEI. Really want to change my career radically. And this is really interesting to me. I... Where where does one go as kind of a next step?
• 1:09:34
  Noah Harlan: So there's a lot of different ways in. I do wanna call something out just in... That you sort of said in passing and and push back on it. Well, not push back on because you weren't making an assertion, but you mentioned it. And I think a lot of people think that this is the right way to frame this.
• 1:09:51
  Like, I'll say something that these problems don't get solved. These problems can be lessened. They can be ameliorated. You can apply the antifragile mindset to them and say, how do we make it so that when someone experiences homelessness, it is brief and it is... You know, it has the least impact possible.
• 1:10:17
  But there will always be another person falling into homelessness tomorrow. And when you... You know, I think it's it's very much a kind of Silicon Valley, you know, I'm going to set out to solve the world's problems, you know, like, we're going to solve, you know, food insecurity. Well, there will always be someone who is struggling with that for some reason or another. And so the question becomes, how do you improve the system so that you can limit it?
• 1:10:51
  And, you know, that flow that I talked about, like how do you titrate the flow? Not how do you shut it off? Because it will always be there and you'll never be in a situation where you're like, we have solved homelessness. Everybody in our society has all that they need. That's just not structurally how the world works.
• 1:11:10
  A hurricane will come through and wipe out a community, and a bunch of those people will have been just on the edge, and now they're plunged. And there was no predicting it, there was no stopping it, etcetera. But in terms of where can young people start, first of all, just a little bit of a hype, but we are hiring for some roles, you know, specifically in the keep team over at Findhelp. So if you're interested, feel free to reach out, ping me on Twitter or email me or, you know, find me somewhere. Probably not blue sky.
• 1:11:42
  I got yelled at too much and had to leave. But other places, I I I should reinvestigate it. Maybe it's maybe it's good. I think that also
• 1:11:54
  Adam Leventhal: You just
• 1:11:55
  Bryan Cantrill: have to, like... You have, like, getting yelled at has to be your kink. That's like me on the YouTube comments. You're like, no. I'm like, actually, I like this now.
• 1:12:01
  This is
• 1:12:03
  Noah Harlan: Someone called me a UX Eugenicist, and I was like, okay. Alright. I'm done here. Anyhow, I love it. The phrase has stuck with me, and I think I really want that to be my Twitter bio now maybe.
• 1:12:19
  There are interesting places like, for instance, the Robinhood Foundation in New York has a thing called Blue Ridge Labs that is thinking about these problems. Now, again, I will caution about public interest technology, but there are organizations, FindHelp is one of them, findhelp.org. We are a public benefit corporation, not a nonprofit, just to be clear. There are a variety of companies that are solving different parts of this. And then I would also say, I think that go into the community and talk to one of these organizations.
• 1:12:55
  Go help out with the pit count. Going on a pit count is a wild experience and they're looking for volunteers and that comes up every January, early February. I think, you know, think about the areas that you have expertise in and that you have an interest in and a passion in and can feel connected to. Because I think also on some level, can't forge some kind of emotional bond to the work that you're doing in this space, you will get burnt out because it is a classic Sisyphean task where you spend so long. I mean, to get to the point where we're at today took me three times as long as I thought it would.
• 1:13:44
  And everyone said it's gonna take you longer than you think. And I was, I'm just someone who's like, No, I can do this. I can do this. I can do this. And it is hard, hard work because everything in the system is just working against you from the the sort of sclerotic institutions themselves, their lack of capacity, the insecurity of finance.
• 1:14:07
  Like how many times have we seen this landscape shift in the last decade as we swung from the first Trump administration to the Biden administration to the Trump administration, you know, new legislation, the big beautiful bill, which introduced all these new work requirements has the whole industry scrambling to figure out how are we deploying benefits, reorganization of all of these financial streams and everything. So it's a very, very complicated landscape. And you have to have some degree of personal resilience to say, gonna work in some problems where I'm not going to see necessarily immediate benefit. But if I can find those little things like that story of that guy, I think probably kept us going for six months, just like
• 1:14:49
  Bryan Cantrill: we would just
• 1:14:49
  Noah Harlan: come back to, like, we helped him. We helped him.
• 1:14:52
  Bryan Cantrill: Right.
• 1:14:53
  Noah Harlan: You know? And that was something. Like, his life is better because of what we're doing, and how could we do that for five more people, 10 more people? And we were helping more people on the way. And now, you know, literally thousands and thousands of people in Los Angeles, whether they know it or not, have been helped in some way, shape, or form by the organizations that are using Keep out there, and that's amazing.
• 1:15:15
  But you need those things that really drive you like that.
• 1:15:18
  Bryan Cantrill: What I think this is really, really important because I think, I mean, this is part of the definition of a wicked problem is that it won't be solved. So you have to accept that and you have to accept that you're gonna have, that the problem is gonna be or the problems are gonna be punching you in the face over and over and over again. But that if if you can actually make progress on them, that progress can be personally gratifying at a level that you're just not going to get elsewhere. I think we've quoted actually on the Oxide, when we describe our principles and values in terms of Piet Heinz problems worthy of attack prove their worth by pushing back. And you know, is a problem that pushes back a lot and this is true of hard problems and it's true of wicked problems that you're going to get pushed on your ass a lot.
• 1:16:14
  And so you want, you have got to have the right psychology going into it, that it's not going to be a quick hit. I love that guidance of get local and get actually like participate in the pit count. Like what, you know, get, if this is a problem that speaks to you, get into the actual, like guts of it and learn, you know, the way it is now. So you can help help make it better. And I think that that is, I think it's very good advice in every domain.
• 1:16:44
  I think that I I guess, we're ending with where we began, namely, me exacerbating generational warfare. Although, you know, actually, Adam, had my, my my 19 year old listened to the interview and it it was it was reassuring to get... Not all Zoomers is what I would say on the...
• 1:17:06
  Adam Leventhal: That's good.
• 1:17:07
  Bryan Cantrill: Yeah. Exactly. But just like the the... Because, like, look, I... Fine.
• 1:17:12
  I accused I accused the interviewer of not having calloused hands. I... It... By which I... It's like, you're just like need to...
• 1:17:26
  There's real value into kind of starting out at the bottom and kind of grinding a bit and learning how things work. And I think like jump starting yourself at... Has a real... Has some disadvantages where you kind of lose track of the way things actually work. So get those hands down is what I gotta say.
• 1:17:45
  Noah Harlan: I realize I didn't give great advice about where to find some interesting jobs in the space, but I would say go look at the portfolios of some of the the impact investors in
• 1:17:56
  Bryan Cantrill: the space. I thought that... Yeah. So, like, I can about Corp. Yeah.
• 1:18:02
  Yeah. Yeah. Yeah. Yeah. Sorry.
• 1:18:03
  Noah Harlan: Go ahead. Has a... Ali Corp has an impact line. They also have a bunch of stuff that is an impact. But there's a there's a bunch of...
• 1:18:10
  And you can Google and find list of impact oriented VCs. Like, Even in New York City, there's the partnership fund for New York City, which does venture investments and stuff. And you can look through those portfolio companies, you can see what are the types of problems that they're solving. Most of those companies are probably out there looking to hire smart, capable people. And so you can sort of find them, whether it's environmental causes, social causes, criminal justice, food insecurity, you name it.
• 1:18:39
  There are a fascinating panoply of companies that are trying to solve these things and In general, I have my own views about the broader VC community. We've talked about it a little bit in the past, but I think that a lot of those organizations... Because what's interesting is Acumen is a very large investor, but they're a nonprofit in Evergreen. So whenever they make money on successfully having an exit, it just goes back into the fund. There's no investors that get paid out.
• 1:19:10
  There's no limited partners. So their goal is just to keep that ball moving forward. There's another backer of ours was the Sorenson Center in Utah. Oh, it's that. Of like Jim Sorenson, and, you know, they do incredible work.
• 1:19:27
  And... Yeah. So I, you know, lots of folks, but you can hit me up and I can...
• 1:19:32
  Bryan Cantrill: Well, and walking their port coast is a really good idea. Mean, the other like, one of the goofy consequences of the world we live in, in which income inequality has become really acute. You also as a current that turns out can't actually take it with you. And when these people croak, not that I'm wishing the demise of any of any billionaire or trillionaire, but the, ultimately, like, you're gonna see a lot of of folks looking to put money to work in ways that are nontraditional. And whether that's pure philanthropy or this stuff.
• 1:20:07
  I mean, think it's really interesting that you said you were a PVC. Is that common for for Ali Corps for for their investments to be PVCs versus... So PVC is a public benefit corporation as opposed to... It is a still a for profit corporation, but a for profit corporation that has a public benefit as part of its charter.
• 1:20:30
  Noah Harlan: Yeah. People are interesting. They have this dual mandate. Again, Ali Corp had one fund that was specifically for impact investing. And so I think it was pretty common in that fund.
• 1:20:41
  You know, they invent... Like, they also... Like, they founded MongoDB, like, not a PPC.
• 1:20:47
  Bryan Cantrill: Not a PPC. Exactly.
• 1:20:48
  Noah Harlan: Yeah. I think when you look through the portfolios of these companies, you'll see a lot of PPCs, and they have a dual mandate. It's actually pretty cool. With a normal C corporation, like a regular company that most people are familiar with, you're required to maximize shareholder value. And if you don't, you can actually face certain types of legal ability.
• 1:21:16
  Like if you take all that money and you spend it by just giving it away to some charity, your investors or your shareholders can be like, Hey, wait a second. You hurt us by doing that, which is fair. It's a corporation, it's meant to make money. With a public benefit corporation, you get a dual mandate. So one of your mandates is, yes, you try and run it like a profitable company and make money and make money for investors, your shareholders, but you have a second mandate that is some sort of public benefit.
• 1:21:47
  It is articulated in your charter and you are allowed to prioritize that over making money. And so if you do something that advances the mission of your company, your shareholders, your investors, whomever can say, Hey, wait a second, you didn't maximize value for us. You invested a whole bunch in this that didn't make a lot of money. And you're like, Yes, but that was our mission. So it's a very interesting place.
• 1:22:12
  Bryan Cantrill: Yeah. Highly recommend. I think this will be the subject of a future episode at some point, but highly recommend Eric's recent book, Incorruptible, is very good in terms of describing how to make enduring companies, which is... It's, might be top of mind for us, Adam. So something to...
• 1:22:29
  Not to not to show our hand too much, but we're, very read up on PBCs here at Oxide. Yeah. Maybe maybe we did tip our hand too much. But, you know, you have to you have to stay tuned to future episodes. Well, has been...
• 1:22:46
  No. This has been terrific. I really appreciate you coming on and describing. I mean, I think this stuff is so important. And again, you...
• 1:22:54
  I I I love the way you phrased it of like just this this one anecdote of this real difference made in this person's life is like, that's like six months of fuel. Like we can go for a long time on that. And so on the one hand, these wicked problems push you down. On the other hand, these human stories can really, pick you back up and propel you. I assume in a way that is unique in your career, which is pretty exciting.
• 1:23:21
  Noah Harlan: Yeah, definitely. It's been an interesting epic in my career, which has kind of been a series of epics, epochs, not e p I c. Some of them were e p I c, but most of them were e p o c h. And, yeah, it's something I'm, I've learned so much. I came into this such a neophyte about so much of this, but surrounding yourself with really smart people, find the best people you can, get them all aligned and kind of rowing in the same direction, which is true of, you know, anything that you do.
• 1:24:03
  And it's pretty amazing. And I've been really, really lucky to have an incredible team. You know, a lot of the folks who are on my team actually started with me at two bowls, followed to keep, and then after we got acquired two and a half years ago, I've been at Find Help since and are still on the team. And I think that that speaks a lot to both the culture that we built, but also the mission.
• 1:24:26
  Bryan Cantrill: Yeah. Yeah. That's great. It's great. There's a lot to be said for being mission oriented in your career and don't let the noise suffocate you.
• 1:24:33
  There's there's a lot of folks that are mission oriented. Well, that is great. Thank you again for joining us. Really appreciated the conversation in Portland as well, but this is really, really terrific. And well, you know, we'll try not to phrase everything as anti dehumanism, but this one was, I resist on this one.
• 1:24:54
  Sorry about that.
• 1:24:56
  Noah Harlan: All right. Well, thanks so much for having me. I'm thrilled. Love the show. I've listened to it for a long time, and I'm thrilled to be on the other side this time and happy to to talk about these topics and also happy to go on big, big rants.
• 1:25:11
  Like, I... My voice was hoarse the next morning after you and I were chatting late in
• 1:25:17
  Bryan Cantrill: the lathered up. That was good. Yeah. No. That was good.
• 1:25:21
  The the the... There's nothing about, yeah, getting one another to lathered up on a good rant is is really singular. It's terrific.
• 1:25:27
  Noah Harlan: And and any opportunity to be at Munktoberfest, you know, is great. So, yeah, Steve's... Steve, I've known for thirty plus years. He's the best.
• 1:25:36
  Bryan Cantrill: Yeah. And, Steve, if you wanted this to be a Munktoberfest talk, like, you you know... And it's not too late. It could could be a Munktoberfest talk for for 2027. The agenda for 2027 is wide open.
• 1:25:44
  But for 2026, it is too late. We we we got it as a podcast instead of it. But I I feel like this would be a great much over fest talk at some point. And the... I'll drop a link to the interview.
• 1:25:55
  The the the apparently controversial interview that I did where I accused Zoomers of not having calisans in the chat. And yeah, we'll I think we we're going to be around for the next couple of weeks. Adam, if we've got any time off plan. We will will see you next time. Yep.
• 1:26:13
  Awesome. Thanks everybody. Thanks Noah.