Anil is a tech entrepreneur (former CEO at our sister company Fog Creek Software) and writer. You can find him at his blog anildash.com and on Linkedin.
Check out the last time Anil was on the pod in 2020 to talk all things Glitch and Glimmer.
Shoutout to user pgrad for winning a Lifejacket badge on their answer to Using type hint Any in Django - NameError: name 'Any' is not defined.
TRANSCRIPT
Ryan Donovan: Tired of database limitations and architectures that break when you scale? Think outside rows and columns. MongoDB is built for developers, by developers. It's asset-compliant, enterprise-ready, and fluent in AI. Start building faster at mongodb.com/build.
[Intro Music]
Ryan Donovan: Hello, everyone, and welcome to the Stack Overflow podcast, a place to talk all things software and technology. I am your host, Ryan Donovan, and today, as we often do, we are talking about AI, but we're talking about it as a normal technology, a regular thing that is a technology in software engineering. And my guest today is Anil Dash, former Stack Overflow board member, and writer, and technologist. So welcome to the show, Anil, or welcome back, I should say.
Anil Dash: Thanks so much for having me. It's good to be back.
Ryan Donovan: Yeah. So, you wrote this blog post that I saw that the majority view of AI technology is as a normal technology. Can you talk about what that means?
Anil Dash: Yeah, I think there's a couple parts. At first, I actually wanna start by crediting a great researcher-writer-academic, Arvind Satyanarayan, who really coined that phrase ‘normal technology.’ But the broader concept, I think, is one that's been kicking around for a long time for folks who have been in the field of AI and related areas, or sort of just in the broader software development discipline, which is the idea of machine learning and adaptive systems has been around for half a century. A lot of folks know this. I have friends who were, you know, teaching in these disciplines and creating software, and Bayesian systems, and other adaptive systems for years and decades. And so, a thing that folks in the Stack Overflow communities will know is that it ain't nothing new under the sun, right? This is something that has been around for many years, and that certainly large language models and the sort of related tech represent a breakthrough and an evolution, and there's lots of cool things that they can do. But this is not something that comes outta nowhere, or that there wasn't any prior art, or that coders particularly are saying, ‘oh my gosh, we couldn't imagine that this was coming.’ It is maybe a step change, and it's really important to say that and. And the most important part of that is evaluating things in that context, because a lot of the claims that are being made are being made by one, people that are pretending that's not the case, or that are, and I think, upsettingly, flatly not true. It doesn't do that. You know, it can't do that. Or even sometimes, software can't do that. And so that's something where when we get into that realm, and I'm old enough, judging by the gray in my beard, to have been around a couple times when folks start to make the claims that are not what the code can do, not what software can do, not what technology can do. We're in dangerous territory. And so, yeah, I wanted to call it out because also, a couple parts. One is, you know, it's bad for the economy, it's bad for culture, it is bad because you have opportunists coming in. But also, as somebody who loves to make stuff and loves to see the brilliant coders I know make stuff, you miss out. Making stuff with code is awesome and it's this great creative endeavor, and we don't need the BS. Like, we don't need to lie about it. We don't need to make up that it, you know– we don't need to exaggerate because the part that is real is cool. And so, I just end up very, and you know, not to be emotional about it, but I am. I end up very resentful about these people who are misrepresenting it, saying, ‘oh, it, it, it does these exaggerated, magical lies,’ because the part that is real is cool. And we could evaluate it more honestly these ways, and also prepare against the harms, right? We could talk about what is broken, or risky, or dangerous more honestly if we just told the truth. So, that's all to say, you know, so many coders, engineers, product designers, all the people in these technical fields I talked to have kind of all said this almost uniformly behind the scenes, that I just wanted to capture it and share it.
Ryan Donovan: Yeah. I think, especially for non-technical folks, any sufficiently advanced technology looks like magic. And this definitely is that scale for most people. But when I talk about it with folks, I sort of come to the point where it's like, this is another layer of abstraction, and it's another layer of automation, two things that computer science has been trying to do since the field that was invented.
Anil Dash: Yeah, and it is a breakthrough there. I don't want to diminish that. It is a massively better version of natural language interface. It's a massively better version of–compared to like, we were trying to script something in a batch file, right? It does make that much more accessible to millions more people, and I'm not diminishing that in any way. When you make something an order of magnitude easier, you bring in two orders of magnitude more people who can do it. And that's a powerful thing. And so, I don't reduce or diminish how meaningful that is. But I think, the part that I think a lot of, especially very technically fluent people that I talk to, are sort of frustrated by– is they look at the hype around it and they're saying those folks are talking about it in these sort of magical terms, and that's not true. And then the other thing that suffers is the criticism, right? Because to push back on that, you end up having to be 10 times as critical, right? You have to be like, ‘these guys are full of it, and they're kind of destroying the world.’ And so, you sort of bring up the temperature on both sides, and I think that that ends up with the whole conversation get[ting] dumber.
Ryan Donovan: Well, it's kind of a greenfield market and the greenfield technology, so everybody's rushing in, right? I think I've talked to some folks talking about how this was like the case when the automobile was invented – that everybody sort of rushed in. There were, you know, hundreds and thousands of companies. But I think the difference is that nobody really– I mean, obviously, people understand what AI does, but there's a lot of people who don't. And there's a lot of taking advantage of that, right?
Anil Dash: Yeah. I mean, I think there's a good analogy there in some ways, but I think there's also a part where we know that a large language model is software.
Ryan Donovan: Mm-hmm.
Anil Dash: And yet one of the hardest parts is the interface is so anthropomorphic. And so, it's very hard not to see it as having sort of a degree of personhood. So, we can say intellectually it's, it's extrapolating from the data and reflecting it back to us in a way that looks very person-ish.
Ryan Donovan: Right. Right.
Anil Dash: And so, whenever we report bugs on it or talk about it, we're like, ‘it's thinking this, it's doing this.’ And so, when it confabulates or makes a mistake, we can call it a hallucination, and then it seems even more like a person in the way that we make our bug reports, right? And the challenge with that is then a non-expert person is putting the bugs in the wrong category. But as a technical person, we can kind of translate that. We know why it's doing that, but the effect of that is the non-expert people are being exploited, because they're sort of attributing a certain degree of magic to it that some of the people building these systems are taking advantage of. And I just also think of– that's very theoretical, really practically. One of the magical things about software is it's a deterministic, like zeros and ones is a great thing, right? A falsifiable assertion is a great thing. A test that passes is a great thing. We all know that feeling of, ‘I got this build to complete. I got the single light op, I got this HTML page to render,’ like whatever it is where you're like, ‘my code runs,’ is an incredible feeling and part of the reason why is because it's deterministic behavior, because of the nature of zeros and ones. And the nature of LLMs is not that.
Ryan Donovan: Right.
Anil Dash: And now, the problem is we are trying to apply non-deterministic systems to a lot of scenarios where you should have deterministic code. LLMs are bad at that thing, and so why are we trying to use them as the hammer on all these things that ain’t nails? That part is really– and the reason why is well, they've got all this investment in these tools that they gotta find all these uses for. And we have decades and decades of work where we're like, ‘no code's really good at this.’ Like, deterministic code is really good at this. Don't put a fuzzy tool on a non-fuzzy problem. And so, I think that's part of why I have such a passionate feeling about it. And I think so many of us are having that experience at work where—I'm lucky, I'm in a different stage of my career, but I talk to people who are earlier in their career, and they're saying like, ‘we have a really good scripting system at work that does this task, and now our boss is saying like, throw some LLM on it.’ Why? For what? It's automated. It works great. It's super reliable. Why are we adding a non-deterministic, more fuzzy, less predictable behavior to it?
Ryan Donovan: More expensive, too.
Anil Dash: More expensive, more compute-expensive. All those things, right? We know all the sort of negative traits of that system, just to say, ‘we have AI,’ right? When what we have is more reliable. And it's like, well, because we need 10 pounds of AI here. Right? And that's the thing that they don't feel like they can push back on. And that's a lot for me, where I'm like, I am fortunate enough, privileged enough, where I got to be on things like the boardroom for Stack Overflow, which is a highlight of my career, of my life, right? I'm blessed to get to do that, but most people don't have that power, so they can't say, ‘look, boss, that's not the right tool for the job. The humble bash script we have that has been running for six years is fine.’
Ryan Donovan: Yeah.
Anil Dash: Please don't throw an LLM on that. You are going to break it. Or, even if it does work, we don't know when it will stop working, or somebody won't be able to debug it because we don't have a whole tool chain around this thing, and all these other things. And so, I think that's the part where I have so many people I talked to, like I said, that are earlier in their career, or in middle management, or worried about their promotion, or in these companies that are doing big layoffs, or concerned they're gonna be put on a PIP or whatever it is that are like, ‘I guess I have to suck it up and pretend like an LLM is the right tool for the job.’ And for them, that's sort of why I was like, if I can write a thing that they can pass on and say, ‘well, this guy who looks moderately respectable, his LinkedIn looks like he's not a total chump.’
Ryan Donovan: That's right. You've got a good CV.
Anil Dash: Yeah, yeah. I mean, and that's a thing where, like I said, I feel very lucky to get to do that, and hopefully I don't look like a total idiot, but it's a sad state of affairs when I'm not one of those, like, ‘oh, back in my day, the good old days,’ you know, there's no good old days. But there was a time when we evaluated the tech on, could it do the job? Could we put in zeros and ones? Could we put in a falsifiable assertion and validate it? And that's how we evaluated whether the tech was right for the job. As opposed to, ‘boss said, we have to have 10 pounds of this tech.’ Now, there are times in hype cycles where that's not true, and this time it's LLMs. 30 years ago, it was, ‘do you have Java?’ Right? Somebody's boss was getting on a plane–
Ryan Donovan: Just put an applet down there. Yeah.
Anil Dash: Exactly. Somebody's boss would get on a plane and read like, ‘we need to use Java language.’ And it's like, ‘well, is that the right language for the job?’ And it's like, ‘I don't care, use Java.’ And so, there are hype cycles. But to that point, Java was a programming language, and it was turning complete, so it could do the task. It might not have been the most joyful language to use. It might not have been the best syntax to use, but it was not the case that it couldn't do the job. And that is not the truth this time. It is not the case that an LLM is the right tool for the job that can produce reliably the output, and that is a hard thing for a non-expert to understand. And we have to get back to the decision being made by somebody that has that technical fluency.
Ryan Donovan: Yeah. And I think, you know, one of the things that happens in these sort of fresh hype cycles is that, especially with LLMs, people don't know the extent of what the tool can do, what the boundaries are. Because I think as LLM started growing, it's like, ‘oh, we discovered new emergent capabilities.’ People are doing agents and being like, ‘what else could it do?’ You know, they're still in the dream phase.
Anil Dash: Yeah. And they're growing and changing every day, and they're getting more capable, and they're an amazing tool. Yeah. I mean, I am the person who, you know, I haven't been a working coder in many years, and even when I was, I was not a great coder. I couldn't have been hired at Stack Overflow, you know? So, I was always so appreciative and in awe of the guys I knew, you know, ‘guys’ inclusive of all genders, that could do this really incredible skill. And so, when the new wave of LLM-assisted dev tools came out, and all of a sudden I was, ‘I'm back baby. I'm in the game again. I can make this stuff.’ And the truth was, I could always do it, but I only had so many free hours on the weekend. And so, the scope of what I could build had to fit within, let's say, four hours. And so within four hours, I could be like, ‘oh, there's this much that's happened in CSS infrastructure in the couple months since the last time I tried to do this, and therefore, what could I compress into four hours?’ And now, what it did was it accelerated what I could do in my available learning [and] building time.
Ryan Donovan: Right. But you already understood a little bit about computer science, right?
Anil Dash: Yeah. I mean, the fundamentals were always there, right? Like, what CSS is, you know, I'm again old enough where I was around when CSS was created. So, I had always had a foundation. So, I'd had the incremental building, and I always checked in every couple months of like, ‘here's the latest and what's happening.’ And I knew the fundamentals, but this was a kind of the perfect level of collaboration where I wasn't blindly vibe-coding, where I don't know what this code does. I could read it back and also know if it had done something inappropriate. And also, front-end is a perfect domain, 'cause it wasn't gonna introduce a security vulnerability or something like that. But that was a joyful return to a kind of creating with code that reminds me of why I started coding in the first place when I was like five years old.
Ryan Donovan: Yeah.
Anil Dash: You know, it really brought back the joy of coding. So, that's the perfect case of LLMs, and that's also a part that gets lost when we distort what they do. That's the other reason to talk about them as normal technology, is one of the wonderful things about code and tech on its own is it has a heart and a soul. If we would let ourselves talk about AI and LLMs as being part of that continuum of, now here's another empowering tool that is accessible and understandable, and part of that continuum instead of an ineffable magic unknowable thing that was delivered by a bunch of elves in the middle of the night. You know what I mean? Then it would seem like a thing that you could do there, as opposed to controlled by this secret cabal of people that you can't know about, and it's hidden behind this wall. And then also, maybe people look more at the open source, open weight tools and some of those other things. And so, I think that that's all of a piece where we need to shift that conversation into empowering people. And so, it does have me lit up about it because the spirit of Stack Overflow is the open sharing of information with the community, and a fundamental intellectual principle of that, when it was founded by Joel Polsky, by Jeff Atwood, and that the reason I was inspired by them and the community they were part of—not just them, right? This is not like, ‘oh, these two guys.’ It is a larger community they were part of. It was an ethos about democratizing access to these ideas, to these technologies, and what you think about is billions of dollars of economic opportunity, of knowledge, being able to be put in people's hands. And what that premise is, is this is not specialized knowledge that you hold. This is not something that you make a noble, privileged, esoteric– you don't hide this behind a myth. And so, when I see people making a myth, you know, it is a betrayal of that. And it's where we were before Stack Overflow. And in particular, the reason that every single one of the big AI platforms and Big LLMs is good at helping people code is because they were trained on Stack Overflow. Period.
Ryan Donovan: Yeah.
Anil Dash: Right? And it's the generosity of this community in sharing information openly that made it possible. And so, I feel a social obligation to be reciprocal, to share that spirit back. You know, and so to not return the favor and be generous in spirit, and say, ‘no, no, no, this is this rare thing that only this handful of people, handful of companies can do,’ and not listen to their expertise, or the community who are saying, ‘treat this like all these other technologies.’ And there are some exceptions. I'm sure there are some people that are like, ‘no, no, this is unlike anything else before. And you have to treat it as a magical special thing.’ But overwhelmingly, 99.9% of coders I talk to, of product people I talk to, designers I talk to are all like, ‘can we just be normal about this?’
Ryan Donovan: Right. To touch on something you said in there about democratizing the access to the esoteric priesthood, right? People can create without having the skills and knowledge to create now, which is interesting, but also I think a little bit of what when people see I created stuff, they're like, ‘ugh, you don't know what you're doing. You're just making stuff. This book of spells just fell in your lap, and now you're just reading stuff, and you've unleashed a demon,’ or whatever it is. So, I wonder – it's a very interesting technology. How do we push back against it without being as hostile as you feel like you've needed to be?
Anil Dash: Yeah. There's sort of two views of that, right? I think there's the—and I think this is true in a lot of domains—so, certainly artists have had this, I used to be in the music business and you can see a huge pushback there around, you know, you're making slop music and certainly visual artists having that around the sort of visual slop art that's taking over in photography and illustration. But staying within the domain of code, I think there's a lot of people that are worried around, ‘okay, you're gonna vibe code something into existence.’ Let's say, take the extreme example: I know nothing of code. I've never seen a line of code in my life. I use a completely prompt-based thing. I'm like, ‘make me a photo sharing app,’ and it puts it up online, or makes me a mobile app, and I put it out in the world. And we would presume it's probably gonna be pretty insecure. It's probably gonna be pretty non-performant. It's probably gonna have all kinds of bugs, display bugs, and usability bugs, and whatever, right? And that's the current state of the art. And you know, the last company that I was CEO of, we shared co-founders with Stack Overflow.
Ryan Donovan: Sure.
Anil Dash: It was called Glitch, and it was a community for building apps, and we had enabled millions of developers to build tens of millions of apps. And it was designed to be democratizing for people to build apps. It was a web-based IDE, but it was coding, right? And so, you had to know how to write a line of code, but the creation path there was, you would take an existing running app, so let's say you had a little simple React app, and you would click ‘Remix’ and get a copy of it, and then modify it. And so, there were people that would create on there that didn't know much about coding, and maybe they would take my first React app and take your one that had a blue button and change it to make it green, right? So, they might not have had a ton of knowledge, and maybe they would add bugs to it that would make it less secure. Right? So, there were definitely levels of knowledge to that that people bristled at. We had people say, ‘oh, those so-and-so's on Glitch don't know how to make secure apps,’ or...
Ryan Donovan: Mm-hmm.
Anil Dash: And we had kids that were on the site that did make stuff that would, you know, hammer somebody's site when they're trying to scrape content, or do whatever, and then we'd have to reign them in and, you know, whatever. So, when you have millions and millions of devs on the site, you're gonna have some bugs. So, we had a version of that. It was not as extreme as, let's say, the worst-case slop app, but I got a glimpse of the future.
Ryan Donovan: Yeah. Right.
Anil Dash: And then nowadays the AI generate app – I think of one of the popular ones out there is called Lovable.
Ryan Donovan: Mm-hmm.
Anil Dash: I felt it's an extraordinary tool, and I guarantee that people have used it to make insecure apps.
Ryan Donovan: Oh, for sure.
Anil Dash: 'Cause there are people on there creating. So, I think we have a couple questions that are packed in there, one of which is: do we want to give Prometheus fire? Do we want to give people the power to build, without the necessity of, ‘you gotta go and study–‘
Ryan Donovan: Pandora’s box is open, right?
Anil Dash: Right, right. Yeah. Do you have to study data structures for four years, and compilers, and whatever? For me, I'm like, no, absolutely not. We can't have the hurdle be that high. Again, Stack Overflow's ethos is about giving people the access. And so, I don't wanna be a gatekeeper, and I want to bring down the barriers, and so I want the principle of access to be there. And I think that's a good thing. And I think the fact that they're lowering that bar is great. So, the principle? Great. The execution is a range from great to terrifying, right? And the category argument would be—and the refrain that every AI product in the world says, ‘well, this is the worst it'll ever be.’ Right? And I'm like, ‘eh, maybe.’ And you know, their argument is always, ‘AI is always getting better.’ And I'm like, ‘have you guys ever heard of [BEEP]ification?’ Things don't always get better. You know what I mean? But also, part of the answer that could fix a lot of this stuff is using deterministic code. We have security scanners, we have profilers, we have software tests. These are things that exist. Part of the answer is just don't use LLMs for everything. I don't mean to be radical, but what if we use security tools?
Ryan Donovan: I mean, easy there, tiger.
Anil Dash: Yeah. And that's part of what is broken-brained about everybody in this conversation, is the thing that is missing about the– we've polarized this sort of extreme conversation is: the only approaches are either all LLMs all the time, it's only ever gonna get better. The whole bug is that we're not putting enough LLM on it. Or shut it all down. Stop anybody from having accessible tools that empower them. And I'm like, if we treated these things as normal software, use the things for what they're good at, a little bit of LLM to help create, and a little bit of traditional tools that help improve security. Figure out how to use the damn things together, 'cause that's what we've always done. I feel like a crazy person that nobody's doing that.
Ryan Donovan: And I sort of feel like there's gonna be a time where the LLM and AI coding tools figure out caching. Right? They start caching components–
Anil Dash: God forbid.
Ryan Donovan: And I've talked to a company that has–
Anil Dash: The hardest problem in computer science. Here we come.
Ryan Donovan: Yeah. They sort of vectorize it at the component level, so when you want to do the coding, you bring up a secure, time-tested component.
Anil Dash: It would be very, I don't wanna say trivial 'cause this is one of those ‘I could write that in a weekend’ things, but it would be a known approach to constrain and evaluate the code and say, ‘did you reuse from this known set of resources?’
Ryan Donovan: It is a solved problem, right?
Anil Dash: And to train the model on these things. Those are constraints that we can apply when creating code, if we were just thinking about these things rationally. And we're not, we're being religious and dogmatic about them. And that's a terrible sign. And I think that's the kind of thing where if we were starting by describing a problem to be solved in a particular domain, instead of talking about a wish that we had, that a certain technology would be the be-all, end-all solution, we would approach things very differently. And so, the META thing I would say, the bigger thing I would say, is LLMs broadly being applied to specific problems, trained with specific data in specific domains, I think, are interesting all day long. I think a lot of the folks that are probably listening to this are like, like I know I am, are playing around with– I have a local model on my machine. I've written, I don't know, two or three million words over the years on my blog and columns I've written, or whatever. And so, I have my little local model, it's trained on that stuff. And the transcript of podcasts that I've done, and my old blogs, and tweets, and whatever. And so, that stuff is really interesting to me, and I can use that to help improve my writing now, and do whatever. And so, all day long, that's useful to me. The things that a public LLM thinks about me are absurd. Right? I was showing my kid, I have a teenager, when I was talking about beginning of the sort of rise of the public LLMs, like, don't trust these things. I was like, ‘let me show you what it thinks I am.’ In the beginning of it it was like, ‘entrepreneur, writer, technology,’ like, cool. And then it hallucinated multiple children that I don't have.
Ryan Donovan: Oh, okay.
Anil Dash: And I was like, ‘hey kiddo, this is what it thinks you are.’
Ryan Donovan: Right.
Anil Dash: And that's very clarifying when it lies about you, personally. And so, you're like, ‘why is that happening?’ It's because there's this, again, almost religious belief in AGI or whatever. I'm like, ‘that has no utility to me, that has nothing to do with my reality, that has nothing to do with the problem I'm solving.’ I write code to solve a problem. I write code 'cause I'm trying to make a thing. When we write code in that world, like I said, it's a joyous thing the first time you get the build to succeed, and first time you get the thing to light up, and it works. And when you're like, ‘I have a tech I've heard of, let me try and guess if I can make it work for every problem in the world,’ where do they do that at? I've never heard of that.
Ryan Donovan: That's, what is it, resume-based development. I've heard that just talked about Kubernetes in that way.
Anil Dash: Yeah. Yeah. Right. Like, I gotta apply it to everything I could know of.
Ryan Donovan: Yeah. You talked about the technology sometimes getting worse over time. I think what I worry about is the software developer, as a whole, getting worse over time because of this. Because you need less expertise to use it. You know, I like the idea of people getting their first glimpse of code and be like, ‘oh, here's what this does.’ But then not knowing how to go beyond that.
Anil Dash: I'm not too worried about that, and I'll tell you why. So, one, every level of abstraction that's ever happened in coding, people said, ‘well, the coders are gonna suck now,’ but I think coders are builders.
Ryan Donovan: Mm-hmm.
Anil Dash: And I think they're always gonna gravitate to whatever level of power tool lets them build the thing they want to build, and also nothing more. Right? We are lazy. And so, I think you find your level, right? I never went further down the stack than I had to to get to what I was trying to build. But-
Ryan Donovan: But assembly is still there. People still build an assembly when they need to.
Anil Dash: Right. So, I think there's like an inverse pyramid where you get fewer and fewer people further down the stack. So, there will be people that are largely prompt-based engineers, I think. And there'll probably be an order of magnitude more of them than there are scripters who are in order of magnitude more than there are, you know, declarative coders that are in order of magnitude more– all the way down, right? But there are still people writing assembly, right? And there are one order of magnitude less than there are people writing C, or something like that. But I also think of the most passionate coders that I can think of in my life are Rust coders.
Ryan Donovan: Yeah.
Anil Dash: And if you'd have said that cohort would be growing 10 years ago or 20 years ago, I don't think people would've imagined that that was a large and growing cohort. That level of systems language? No way. Right? And as you talk to Rust Coders, they all started with Python scripting and things that were considered toys. So, I think 100% there's gonna be somebody that starts as a prompt engineer who becomes a Rust coder. 100%.
Ryan Donovan: Somebody goes down the rabbit hole.
Anil Dash: Yeah, exactly. So, you have to have a gateway drug. So, I have high confidence in that. People will always be curious. People will always– well, the things will always break. The abstractions will always leak. That it's always gonna be an error message where you're like, ‘what the hell is that?’ And also, what are the lessons of Stack Overflow?
Ryan Donovan: Mm-hmm.
Anil Dash: Coding is profoundly social. People love hacking together. They love showing off. They love being proud of what they made. They love being curious together.
Ryan Donovan: Knowledge is profoundly social, I think.
Anil Dash: Yeah. And so, I think that that aspect is always gonna pull you further down the stack. So, I don't worry for a minute about. Oh, this is gonna attenuate people's curiosity down the stack at all. A couple things I'd say, real briefly, just to as we're wrapping: one of the reasons I wanted to come on, as you can probably tell, Stack Overflow is really near and dear to my heart. This community has always been, even before I was on the board, before the site had a name, Joel and Jeff were talking about the idea, and there were mailing lists and other communities that people used to answer these questions. But the ethos they had, and that was part of the larger community they were all part of, was that everybody should have access to this information, like I said. And you know, that is still something that I think we have conviction in. They both do. I know. And you know, Jeff recently announced a project he's been doing around, you know, the broader domain of universal basic income that he's been very successful with this company in supporting people who have the least. And that's dollars, but it is based on the same generosity of spirit of taking people who have the least access and give them the tools, the information, or, in that case, raw resources. You know, to get access like that is what this place is about. This community has always been about. I think that's something profound, and it's worth saying out loud, explicitly. And this community, people who are in this community, in this podcast, should still be championing that. It is not set out loud enough. It is something that people are not familiar with enough, you know? All these years later, people sort of take it for granted, but people come to this site every day because there is a coder kindness. And so I just, you know, it's like I said, it's the thing that I want to say explicitly 'cause it's understood, but if you don't say it out loud, somebody who's new to coding, a kid who's like coding for the first time, or somebody who's new to the industry, they might be like, ‘well, maybe that's true, but I don't know.’ And there are people who are, to be direct about it, exploiting that. Because I think there's also been some unfair exchange, right? Where there have been takers, there have been platforms that have taken from the community and not given back, and AIs that have taken from the community and not given back. And I think it's hopefully gonna get more fair. I think everybody's reckoning with, ‘how is everything on the internet gonna be fair game for every model, every LLM?’
Ryan Donovan: I mean, you still need the internet to continue to make more data as an AI, so you have a self-interest.
Anil Dash: That's right. The analogy I use is, you know, a farmer can't eat all of the corn. They need to have some to plant the seeds for next year, right? But I think there's, you know, a whole generation of kids that have grown up listening to ‘Taylor's Version’ for Taylor Swift owning her master recordings. And so, they know, even if they don't call it intellectual property, they're like, ‘oh, an artist should be able to own their work and have control over how it's exploited.’ I think everybody who's ever answered a question or asked a question on Stack Overflow has as much right to their work as Taylor Swift does. And so, you know, thinking thoughtfully about that as a community and still fighting for those ideas about being generous to each other, and having control over that stuff is really important. So, you know, I've been ranting the whole time, but I just, I wanted people to know that that's been true from day one, and to hear it from somebody who's been lucky enough to get to watch it and to be fanboy number one from Stack Overflow, for the beginning. AndI was there when Jeff came up with the name Stack Overflow. We were in a hallway at Microsoft. Right? So, I got to watch the genesis of this place, and to watch it take off, and to watch people have their greatest moments of creativity and curiosity, and to, you know, see it all along. And so, that's still there.
Ryan Donovan: Mm-hmm.
Anil Dash: And so, just to sort of encourage people to carry that forward and just kind of shout it from the rooftops, because we're in a moment where people are trying to figure out whether that's going to keep thriving.
Ryan Donovan: Right.
Anil Dash: And it will, but it requires sort of nurturing and fighting for it.
Ryan Donovan: Right. Community takes work.
Ryan Donovan: Well, it's that time of the show again where we shout out somebody who came onto Stack Overflow, dropped a little knowledge, shared some curiosity, and earned themselves a badge. Today, we're shouting out a life jacket badge winner – the junior lifeboat: somebody who came to a question that had a score of negative two, was sinking, and they dropped an answer that brought it up above the waterline. So congrats to ‘pgrad’ for answering, “Using type hint Any in Django - NameError: name 'Any' is not defined.” So, if you're curious about that, we'll have it in the show notes. I'm Ryan Donovan. I edit the blog and host the podcast here at Stack Overflow. If you want to reach out to me about topics, comments, concerns, et cetera, you can email me at podcast@stackoverflow.com. And if you want to reach out to me directly, you can find me on LinkedIn.
Anil Dash: I'm Anil Dash. I'm a proud alum of the Stack Overflow Board and a writer. You can find me at anildash.com, and from there, you can find me on all the different social networks that I'm on that are linked from there.
Ryan Donovan: Alright. Thank you for listening, everyone, and we'll talk to you next time.