# Developer Platforms with Kevin Moore

by [kirupa](https://www.kirupa.com/me/index.htm) | filed under [Interviews with Creative People](https://www.kirupa.com/podcast/index.htm)

[Kevin Moore](https://twitter.com/kevmoo) makes the case that developer platforms are product work, not just infrastructure work. He connects early web lessons, performance, open standards, modern runtimes, and AI to a single idea: platforms succeed when they reduce friction, respect developer judgment, and make powerful capabilities more accessible without pretending the tradeoffs have disappeared.

Watch the interview: https://www.youtube.com/watch?v=p9I2Ozvtpvc

A conversation with **Kevin Moore** | 1h 09m

## About this conversation

Kevin Moore comes across as someone who thinks about leverage from several angles at once. He cares about code, of course, but he is just as interested in the people using the system and the business reality around it. That perspective turns what could have been a narrow technical interview into a broader conversation about platform design. His early stories about programming and web development are fun on their own, but they also explain his bias toward concrete usefulness. Platform work, in his framing, is not an abstract exercise in elegance. It is about helping other builders move faster, make better decisions, and avoid unnecessary friction.

That sounds simple until the tradeoffs show up. Kevin spends a lot of time in the territory where performance, latency, technical debt, reliability, usability, and long-term maintainability all compete for attention. One of the strongest ideas here is that developers should be treated as customers. Once you accept that, documentation, onboarding, defaults, and operational stability stop feeling like secondary chores and start looking like core product decisions. The conversation also widens into bigger platform shifts, including missed transitions in older eras of computing and the cost of building in ways that ignore what users actually need. It is a sharp reminder that technical sophistication alone does not guarantee relevance.

The final stretch looks forward without losing discipline. Kevin talks about openness, the importance of standards, and why modern runtimes have made the browser far more capable than many people still assume. WebAssembly, richer APIs, and faster engines widen the kinds of experiences the web can support, especially for performance-heavy applications. When AI enters the discussion, the same sober tone holds. Natural-language interfaces and generated code can lower barriers, but they do not replace precision, intent, or craftsmanship. If anything, they make those qualities more important. In that sense, the best developer platforms do not erase expertise. They help more people access it and help skilled builders go further with less waste.

## What you'll hear about

- Developer platforms work best when developers are treated like real customers.
- Performance, usability, and maintainability are product choices, not side concerns.
- Open standards and open source widen trust, reach, and longevity.
- Modern web runtimes make far more ambitious browser experiences possible.
- AI may lower barriers, but clear intent and craftsmanship still matter.

## Jump to a topic

- [0:00](https://www.youtube.com/watch?v=p9I2Ozvtpvc&t=0s) **Why developer success matters**: Kevin opens with his background and the throughline of helping developers do better work.
- [7:00](https://www.youtube.com/watch?v=p9I2Ozvtpvc&t=420s) **Learning by shipping real things**: He revisits early programming and web projects that taught him through direct contact with users.
- [15:00](https://www.youtube.com/watch?v=p9I2Ozvtpvc&t=900s) **Platforms are product work**: The conversation examines what developer platforms need beyond pure technical sophistication.
- [24:00](https://www.youtube.com/watch?v=p9I2Ozvtpvc&t=1440s) **Performance is a feature**: Kevin explains why latency, responsiveness, and overall feel are foundational platform concerns.
- [32:00](https://www.youtube.com/watch?v=p9I2Ozvtpvc&t=1920s) **Big bets and missed turns**: He reflects on platform transitions, user needs, and the cost of getting a strategic shift wrong.
- [40:00](https://www.youtube.com/watch?v=p9I2Ozvtpvc&t=2400s) **Why openness changes everything**: This section covers open source, ecosystem expectations, and the credibility that openness can create.
- [48:00](https://www.youtube.com/watch?v=p9I2Ozvtpvc&t=2880s) **The browser as a serious runtime**: Kevin explores the web’s expanding capabilities and what modern runtime layers now enable.
- [58:00](https://www.youtube.com/watch?v=p9I2Ozvtpvc&t=3480s) **AI without the hype**: He closes on natural-language programming, generated code, and the continued importance of engineering judgment.

## To learn more

- [Kevin's Twitter](https://twitter.com/kevmoo)
- [Kevin's Blog](https://j832.com/)
- [Kevin's LinkedIn](https://www.linkedin.com/in/kevmoo/)

## Transcript

The automatically generated captions have been organized by speaker, lightly edited for clarity, and broken into paragraphs for readability. Names and wording may still contain errors.

### [0:00](https://www.youtube.com/watch?v=p9I2Ozvtpvc&t=0s) - Kirupa

Hi everybody. Today we meet with Kevin Moore, someone with the rare ability to balance technical debt with a deep understanding of customers and business needs. I've known Kevin by reputation since my Microsoft days, and more recently in real life at Google, where our paths cross pretty frequently. He has spent a lot of time thinking about how to help developers be successful. In this chat we get a glimpse into the insights and energy he brings to that challenging and fun problem space. Kevin, it's great to chat with you.

You're one of those people I've known for a large part of my career through reputation, but only got to meet maybe two or three years ago. So it's great to chat with you. For the rest of the audience, can you please introduce yourself?

### [0:50](https://www.youtube.com/watch?v=p9I2Ozvtpvc&t=50s) - Kevin Moore

Sure. My name is Kevin Moore. I like my Twitter bio because it reminds me what's important in life, since we always introduce ourselves by what we work on. I'm a very lucky husband to an amazing wife, a proud father of three kids, and, oh, by the way, I'm a computer guy at Google. It'll be 10 years in December, which is kind of wild to think about. Time flies so fast. The whole time I've worked on Dart, and then Dart and Flutter. Flutter kind of thundered along after we did the Dart thing, and we're still doing the Dart thing.

Work-wise, my focus right now on Dart and Flutter is the web stack. I proudly say I'm the only one on the team who owns a vertical. There are people who own the Android story, and Michael owns the Dart low-level platform story, but I own the web story, which includes both low-level and high-level pieces. We have two JavaScript compilers. We have APIs for doing interop with JavaScript. We now have a WebAssembly compiler, which we still kind of classify in the web space, although there's lots of interest on the server there and maybe we'll talk about that later. Then there's how we present and deploy Flutter apps on the web.

So it goes all the way down to the compilers and thinking about interop with low-level JavaScript, all the way up to how we deal with WebGL and URLs and all those fun things.

### [2:12](https://www.youtube.com/watch?v=p9I2Ozvtpvc&t=132s) - Kirupa

And your official title is product manager. Do you see yourself more as a product manager or as a developer? Where do you spend your time?

### [2:22](https://www.youtube.com/watch?v=p9I2Ozvtpvc&t=142s) - Kevin Moore

Actually, lately I've been pretty good about sticking with product management. I was notorious for a while—probably the right term—where, despite my role being product management, I did a lot of engineering, and I still do quite a bit. I love to code. I love solving problems. My joke for a long time has been that I'm an engineer who's disproportionately good at BS. So people say, you talk too much. So maybe we should make you a product manager, because I do love talking to customers and thinking strategically.

My flippant way of saying it is that I like being invited to all the meetings, but the meetings are where important things get discussed and figured out. Thinking at the strategic level is really fun. But I also like going deep, looking at commits into the Flutter engine or the Dart SDK, and landing patches myself. Yeah, I love that.

### [3:26](https://www.youtube.com/watch?v=p9I2Ozvtpvc&t=206s) - Kirupa

Doing that. All right, and we'll touch on all these topics. But let's go back in time. How did you get started with computers? What made you take the path of being a developer and then a product manager and so on? What's the first thing you remember that made you think, this is what I want to do?

### [3:46](https://www.youtube.com/watch?v=p9I2Ozvtpvc&t=226s) - Kevin Moore

All right. My wife would tell you I've never been someone who wondered what I was going to do when I got older. It's always generally been about what I liked doing. There are so many places to start there. I remember this visceral moment in my junior year of college, when I had friends who were switching majors and effectively going from being juniors to freshmen again in terms of requirements. They had all this angst and uncertainty about what they wanted to do with their lives. I remember almost having a memory about a memory.

It's hard to describe, but I had this almost guilty feeling—my small-town Midwest comes out here—which was that I had never had that uncertainty. I had always known I was going to do computers. Just always. My dad bought a computer when I think I was in third grade. It was a Zeos computer with one meg of RAM and an EGA monitor. My mom actually found the invoice for it and sent me a link, and I think I tweeted it.

It's funny to think about, because with inflation it would probably be like $3,500 now, basically the cost of a Vision Pro, for an EGA monitor, one meg of RAM, I think a 30-meg hard drive, a five-and-a-quarter-inch floppy drive, and GW-BASIC. I learned to program in GW-BASIC and wrote code to calculate prime numbers and print them out. I'd set the program running and I didn't even understand square roots at the time. I look back now and think I could have written that much more efficiently in GW-BASIC, but that was the start.

Then in high school I got a job at a company where I literally started by mopping floors, sweeping, and breaking down boxes. It was the only company in my town that did anything with technology or electronics. Then they said they were going to start doing web stuff, so I got into it. I started out with Perl, then moved to—I'm really going to date myself here—ColdFusion. And then Active Server Pages, version 1. It was VBScript or JavaScript, but mostly VBScript. That was the first time I got paid for writing software, doing a tiny bit of Perl, enough to know I hated it.

Most of my time in high school I was writing VBScript and building things like real-estate websites. For a long time the backend was an Access database. There was one computer upstairs hosting the website. It was all local, not dozens of hits a day, but for a long time it was just ODBC talking to an Access database. At some point we thought maybe we should use SQL Server, maybe we should use a real database and not just a file. I remember opening the database on my desktop while the live site was also hitting it through the file system. It was just insanity.

That's just how we used to do things, well before the cloud. I remember when the table tag was new. I remember when it came out and thinking, oh my gosh, look what I can do with a table tag. I am that old. I remember the table tag.

### [7:15](https://www.youtube.com/watch?v=p9I2Ozvtpvc&t=435s) - Kirupa

I mean, you're not that old, Captain. I'm also that old, because everything you're saying resonates with me very closely. We were doing Perl for CGI-style apps, right? Yeah, exactly. When we saw a website with something like CGI-BIN in the URL, that gave it extra credibility. It was like, okay, these people know what they're doing. An ASP extension at the end also gave it a little more credibility than your typical GeoCities or Lycos web host. I would actually take my regular HTML files and just change the extension to. Asp to give myself extra street cred. This is dynamic, exactly.

And the only thing dynamic was that I started off with the web editor that came with the browser—FrontPage editor and the Netscape editor—because I didn't know how to write a table. I'd click the little table button, enter the grid and rows, manually like an animal, hit Enter, and that's it. You had this limited set of things you could do, and that's what really got me interested in the whole world. The funny thing you mentioned also: did you ever do anything with Apache or server-side scripting for templates? Because server-side includes were basically the only dynamic thing back then. Did you go deep into that, or more on the ASP side?

### [8:33](https://www.youtube.com/watch?v=p9I2Ozvtpvc&t=513s) - Kevin Moore

Purely on the ASP side. It's always interesting to look back at the little things you start with and how you stick with them. We were a Windows shop. It was a micro shop, the company was called R. D. Industries. I think rdi1. Com is still around in Milford, Iowa. Don, the guy I worked with, is still running it. We were an ASP shop. One thing I remember is that I wrote a little reusable login system. Even back then I was thinking about frameworks and reusability. There was no notion of encapsulation.

If you wanted something to happen at the top of the page and something else at the bottom—like omit content if the user wasn't logged in and show it if they were—you were literally just inserting code. You had to include one file at the top of the page and a separate file at the bottom, and they had to be in the right hierarchy in the HTML. They worked together, and all the variables just flowed through. There was no encapsulation. You could stomp on globals, but it worked. So I made this little authentication system so the people at the real-estate website could log in and update their own listings.

It was funny thinking about that, because they used to call us or send us an FTP upload with the images and descriptions for the properties. I did all this work to let them manage their own properties, and they were like, this is so weird. Why are you charging us so we can do the work ourselves? And I was like, because we're charging you 50 bucks an hour—which now would probably be like 300 or something—so we're actually saving you money. You don't have to email us and wait three days for us to change your images. You can do it yourself. It was a feature.

I mean, this was also back when you could paste HTML table bombs into a guestbook and destroy someone's website. I of course never did that, but there was always this sense that I was probably the first kid on my block to understand the importance of HTML escaping and SQL injection. It was a fun time because no one really knew what they were doing, there weren't that many resources, and there were no best practices. Was it secure? Maybe. Maybe not.

So it was like, yeah, I'll just cobble together whatever works and hope it works in the latest version of Netscape and Internet Explorer, assuming Internet Explorer was in a useful state at the time. IE4 was just...

### [11:18](https://www.youtube.com/watch?v=p9I2Ozvtpvc&t=678s) - Kirupa

Yeah, I remember the days of Netscape and then thinking, oh, this IE thing. I remember seeing Windows 95 for the first time and all that stuff. One of the things I worked on very early in high school was for an automotive dealership where they needed someone to update the cars on their website. Exact same story. They'd literally email me pictures of cars, and I'd take them into Microsoft Image Composer—it came free with Office. I'd resize the images, add some weird clip art like a little star, put a price inside it. It was tacky as ever. To be fair, car sales are still tacky today.

The flyers you get in the mail are still like that. They haven't evolved much. But they were sending me a couple dozen photos a week and I'd do the exact same thing every time. Eventually I thought, you know what, you should be doing this yourself. So when you mentioned you were doing a CMS, in my case I pretty much optimized myself out of a job because they were like, oh, we're going to do this ourselves now. Whereas I think you had better luck. They still paid you; you just did less of the rote work.

### [12:27](https://www.youtube.com/watch?v=p9I2Ozvtpvc&t=747s) - Kevin Moore

Clever on that side. Yeah, it was also fun because there wasn’t this front-end and back-end design split. I did it all. Oh my gosh, I have flashbacks. I recently looked in the Wayback Machine at a company called Hogemeyer Hybrids. They were a corn company. This is Iowa, right? I had made the background of the website a tiled corn kernel, a big one, maybe 60 by 60 pixels. It felt like a large image for that era, and the entire page repeated it. Imagine the most gratuitous yellow-and-orange tiled background. Then I had an animated GIF of their little old-school truck that hauled the corn looping across the top of the web page.

They loved it. I look at it now, and it would probably be retro-cool, but it’s completely, to quote my 12-year-old, so cringe. I got seven bucks an hour making that silliness for their corn company.

### [13:27](https://www.youtube.com/watch?v=p9I2Ozvtpvc&t=807s) - Kirupa

It was fun. I don't know if I had Simpsons much, but there was a great scene where Homer's first web page was basically that: a cyan background, Times New Roman text, random things moving around for no purpose, a hit counter at the bottom. Oh my gosh. I think it was the Mr. X episode, actually. It went from being wow, that's terrible, to exactly what you said: now it's retro-cool. Everyone's trying to get back to that. So I'm going to quiz you now. What's the darkest gray that's not black in the HTML-safe color palette? What is the hex?

### [14:08](https://www.youtube.com/watch?v=p9I2Ozvtpvc&t=848s) - Kevin Moore

Who I want to say triple one. So my memory- and this is it'd be interesting to see in the doobly- do the comments below remember like it was, it was multiples of 51.. So it's 5150, 151, and then 102, 102, 102.. Profile for colors. So if you're gonna pick a color, you want to pick because, like, this is back when people still had 256 color screens, yes, and so it's like, oh, like, we'll stay in this safe palette so that you won't- I forget what it's called- when, like you know, when a 16-bit color image or you know, would get down, sampled and look bad.

So it's like, oh, no, we'll stay in this like web safe subset of colors, and all the units of the hex value were multiples of 51.. And so, like, oh gosh, there's FFF. And then the next one was like, I think, DDD or CCC, cccc, like, and you can use any combination, yeah?

### [15:06](https://www.youtube.com/watch?v=p9I2Ozvtpvc&t=906s) - Kirupa

And so on. Yeah, and again, it'd be interesting to see if this is you know, we're either gonna completely bore people or, like you know, we're totally there, it's not gonna, you know. No, no, no, people know what to expect when they're listening to this guy. Okay, because you know, I don't remember this, but I do know that when it exports some images in fireworks- speaking of tools that you know no longer around- are, you know, optimized for the web?

There was always a drop down for web, 216, optimize, like 216 colors, and I always think backwards and we're having other 50 colors, like, why is the other 50 not there? It may have something to do what you kind of mentioned here, like for website, certain colors probably just don't really work really well. Yeah, it was dealing with dithering and creating. That's right, it is dithering, that's the right word. And you had to create gifts and Gifts were there was, it was low colored gifts were low.

So if you want to do animated gifs, you needed to pick in this safe palette because, or otherwise, you get ugly dithering when you yeah, just yeah, that's right. Who's these days don't understand how? And dial up networks and spinning hard drives. 5400. Yeah, like this is. This is you know? I mean I always tell people like for me the most magical moment was like when my 28.8 US robotics modem got a firmer upgrade for the v90 standard and once I installed it, I could now connect at a whopping.

You know, 48, 40 Point, whatever the connection Beyond it, because there were two competing standards, right like one was something that three com with maintaining I think they can use their bodies was not owned by 3com at the time. They know from that Lucent was maintaining as well, and so, depending on which I, which ISP you chose, base was based on which of the modems that you had installed. It wasn't until that firmware upgrade that it all worked out. In my mind I'm like blown.

I'm like, wait what I installed, this weird piece of software- and of course I don't know about bios screens of, like these days I'm gonna even on a Mac forever now, yeah, like always, imagine it's like. It's like really just gigantic, like you know, blocky, black or blue screen with like some taxes hit, enter, warning, stopping in the middle might cause the world to collapse, and then a few minutes later everything turned off and it turned back on and now you're in a whole new world. I mean, oh gosh, you get into irqs and dmas, like that's right.

### [17:13](https://www.youtube.com/watch?v=p9I2Ozvtpvc&t=1033s) - Kevin Moore

I mean, I'm a company guy. Did I joke with you about this? I unironically owned a Zune when I worked at Microsoft. I still have a snowboard with a Zune sticker on it, and now it looks ironic. I'm like, no, at the time I put it on unironically. I was loyal. I was one of the first people to use Stadia, and I won't get into it because smart people worked on it and software is hard, services are hard, but I used Stadia. And now, if I'm playing my AAA games, I use the Nvidia GeForce streaming thing.

Every once in a while I'll be running around medieval England in Assassin's Creed and think, I used to wait for images to load. Now I'm on Wi-Fi streaming a AAA game in 4K. It's one of those history moments where you realize we've basically added three zeros to bandwidth and latency. It is amazing how far it's come.

### [18:13](https://www.youtube.com/watch?v=p9I2Ozvtpvc&t=1093s) - Kirupa

In our lifetime. Yeah, it's mind-boggling. Even our compression technologies have gotten so much better because our CPUs are so much more capable. The technology and algorithms often already existed; the question was how much CPU you could spend per frame. That was impossible 20 or 30 years ago. Now it's trivial. It's like, sure, mine Bitcoin on the side while playing a AAA game.

### [18:34](https://www.youtube.com/watch?v=p9I2Ozvtpvc&t=1114s) - Kevin Moore

Right. My wife is upstairs doing telepsychology with her clients, and it works. It’s kind of amazing. The world has changed. It’s one of the things I always talk about when I speak with newer people and newer grads. My daughter is five, and to her everything is a touchscreen. She comes to my Mac and touches the screen. I have to tell her, no, Apple’s current laptop lineup doesn’t support touchscreens, for example. Whenever she sees something brand new, for a brief moment I go back in time and think, wait a minute, I remember A, B, and C, and all these things that came before.

I feel like I waste at least 50 or 100 milliseconds going back in time to figure it all out. I’ve already expended energy remembering that history to the point where I need a break. New people today don’t have a concept of what the world was like back then. They see something new, and their energy jumps to the future very quickly. I’m still stuck in the past, thinking it’s mind-boggling where it came from. That historical comparison happens automatically before I can evaluate the new thing. Right now, we’re talking about going from 28.8 to what we have today. Someone newer who didn’t experience those things might say, “Wait, this is crap.

I see a little screen tearing here. How do you make it infinitely better?” I would never even get to that point. I’d still be amazed at how good the same technology is right now. In a transparent attempt not to bore the people listening, because I could do the old-man thing for a long time, even if you think about product management, there’s that same tension.

### [20:04](https://www.youtube.com/watch?v=p9I2Ozvtpvc&t=1204s) - Kirupa

I think that's genius. I don't want to oversell it, but the ability to hold two things in your head at once and not go crazy feels like intelligence to me. I aspire to do that. So as a product person, how do you maintain excitement and awe about what you've built and what's possible, while also keeping not a cynical eye but a critical eye? There's always room for improvement, room for polish. You need both. You need to simultaneously find excitement in what you've built and what's available now.

### [20:37](https://www.youtube.com/watch?v=p9I2Ozvtpvc&t=1237s) - Kevin Moore

You have to recognize what you’ve delivered and what’s available now, while also recognizing that there’s so much more work to do. Doing both is tricky. Is that being a pessimist, an optimist, or an opportunist in terms of the opportunity we have, what we could build, and how much better we could make things? Yeah, it’s a tension because you never know whether your own inherent experience is biasing you against seeing the next big thing. Or is that experience actually a skill? Because you’ve seen these things before, maybe you can spot which of the many next big things are viable and which are noise in search of a solution.

There’s also the flip side. You see the bug reports and issue reports. You’re deep in both our communities, probably more so in mine because of what I work on. You see people shaking their fists on Twitter and social media. God bless them for being vocal and trying to be constructive, but it’s easy to beat yourself down about the state of things you’ve delivered. Then you see survey results or talk to random people at conferences and realize there are many people who are delighted, having a lot of fun, and building really exciting things. Again, you have to find that careful balance. It’s tricky.

There’s that old phrase that, if you ask people what they want, they’ll ask for a faster horse. They’ll never ask for an automobile, never a plane, or anything like that. How do you balance those things?

### [22:08](https://www.youtube.com/watch?v=p9I2Ozvtpvc&t=1328s) - Kirupa

A version of that is: if you ask a monkey how to get to the moon, it'll say it needs to find a taller tree. So how do you switch your perspective?

### [22:18](https://www.youtube.com/watch?v=p9I2Ozvtpvc&t=1338s) - Kevin Moore

That is tricky too. I deal with that all the time: how much do you spend spreading peanut butter and making incremental improvements versus taking time to invest in something truly new? We could always be optimizing Flutter web rendering. There are endless opportunities there. But then there's this WebAssembly thing, and how do we balance our allocations there, even when for a while no browser supported it? I don't even know if there's a right recipe. It's just a lot of experience and, I guess, a little...

### [22:52](https://www.youtube.com/watch?v=p9I2Ozvtpvc&t=1372s) - Kirupa

A bit of luck as well. Intuition, luck, and all these things play a role. Another phrase I remember came from economics classes I took in college. I mentioned computer science, but I didn’t like any moment of it. My way of learning was different from how it was presented, with walls and lots of text. I did the bare minimum needed to graduate with a CS degree and took a bunch of other classes instead. I really enjoyed economics. One thing that always stuck with me is that, if you’re a company making a profit, at some point that profit gets driven down toward zero, either by competition or by obstacles in the market.

It’s always in your best interest to keep finding something new before you get to zero. Competition, market obstacles, and too much attention can all erase the advantage. I see that applying so much in our world. Peloton is my favorite example. Peloton completely saturated its market during the pandemic. Once everyone who could and wanted to buy a Peloton had bought one, that was it. You always talk about moats and sustainable advantages. In developer tools, where you’ve worked on developer platforms for a very long time, we see the same thing all the time. Just consider all the technologies we’ve discussed in the last 20 minutes and what eventually happened to them.

They’re all still around in some form, but the players who built businesses around them or monetized them are no longer necessarily the same ones. ColdFusion isn’t something people talk about much anymore. ASP became ASP. NET. The idea of scripting in the browser or on the server is still very much alive across many companies. Things like server-side includes or Apache-era programs aren’t what people reach for now. They go to Rails or to more modern apps like Next. Js. You have dynamic front ends and back ends in a monorepo, where one JavaScript codebase handles everything.

It’s always funny to wonder how much of this will be seen as irrelevant five years from now, and what much better thing will have come out by then.

### [24:54](https://www.youtube.com/watch?v=p9I2Ozvtpvc&t=1494s) - Kevin Moore

Oh wow, there's so much to pick apart in that. But maybe the answer is that some things will always matter. The importance of user experience, the importance of responsiveness, even the speed of light. Think about the ability to do a streaming game. You and I are having this conversation right now, and as good as our networks are, I'm assuming the software is recording a local copy that it's going to upload later.

### [25:34](https://www.youtube.com/watch?v=p9I2Ozvtpvc&t=1534s) - Kirupa

Things are coordinated exactly, which is why I use this particular setup. It does exactly that.

### [25:39](https://www.youtube.com/watch?v=p9I2Ozvtpvc&t=1539s) - Kevin Moore

And I think you and I are across a lake, you know, I mean, 10 years ago we would have found a reason to get together, which may have been good to be in the same room with you. It's like, oh no, it's like I can just walk to the conference room and, like, you know, and I also always tell you last week, so you know, it only takes so much, so yeah, some things will always matter.

And then I guess, maybe that's, maybe that's the trick, right, which is understanding- and I think this comes with some experience- which is there's a set of things that will always matter, like fundamentals, whether it's human nature, or fundamentals of complexity and whatever that is, like the complexity of your data, the complexity of your software, the complexity of the user experience you're building, realities of humans. And then there's things that are parochial, that you just don't know, like if you only ever did you ASP, we're like, clearly, this is the way you do things.

Or if you've only ever programmed in one programming language, if you only ever used one operating system, you know, like, oh, like, there's actually many ways to do this and even being around a while, like there's certainly benefits, I can't stand for four in the morning coding anymore. But just seeing, like, how often things repeat- because I mean we were talking earlier. I worked on, you know, WPF, what we called Avalon at Microsoft- you and I both did xaml, XML, like all the stuff. Like you know, at the time, fresh out of college, was like, oh, XML is everything.

Of course, everything will be XML and soap, and it's like you've been around a while. We're like there's many ways to encode data and like this is overly verbose. You know, oh, I had another example and it totally left me: but like these kinds of things, right, you're like, or even just like, clearly we need to make a proprietary stack and use the operating system to build all this stuff.

And of course now my experience with, with working on flood and other things is like you can make super high performance, High Fidelity things that are completely open and run great everywhere, right, like, and there's benefits to owning the stack and there's trade-offs there. Like do you want to build something proprietary that people have to buy into or do you want to build a platform that anyone can run anywhere? And there's inherent trade-offs in those things, and so, yeah, figuring out like what's flavor of the month versus like oh, this is the fundamental truth about how things should work.

### [27:47](https://www.youtube.com/watch?v=p9I2Ozvtpvc&t=1667s) - Kirupa

It’s tricky. It very much is. You mentioned cycles, and we’ve both been through a few where we see the same story play out with a slightly different cast and setting. We’ve both worked on platforms that were very successful and on some that were horrifically unsuccessful. The interesting thing is that, if you and I were asked to build a new platform today, we’d bring all our learnings and optimize for what would lead to the most success for the product and the developers. You and I probably graduated college around the same time, in the early to mid-2000s. Was it ’02? Iowa State, proud grad, yeah.

When we started working on these things, such as Apple in your case or tooling on the WPF side for me, there were people around us who had already seen the cycle before. It might have been our first active cycle, but others had already been through a few. They had watched similar platforms rise, hit constraints, and either adapt or fail. In the back of my mind, I always think that if I’ve been through three cycles since then, the people we worked with should have known these things too. They’d probably already been through three cycles of their own and had their own lessons about how these things should be done.

Yet despite that, things didn’t go the way they could have gone. I always go back to why they didn’t. Was it deliberate? In a large organization, you can optimize for many things, and they don’t always point in the same direction. One is making customers happy. On developer platforms, that especially means making developers happy. Another is making the company’s core platform successful. At Microsoft, how do you make Windows very successful? That might run counter to making developers happy, which might mean being open and easy to use. The product, platform, and developer audience can each reward a different choice. Then there are purely personal gains as well.

Inventing a shiny new object can be better for my career. Growing my team abnormally can be better for my career. I might optimize for those angles. Whenever I look back, I can see the agenda items people had and align them with why they did what they did. Those incentives help explain choices that otherwise look irrational from a purely developer-focused point of view. It gives the whole thing a different color. I’m curious whether you’ve had the same experience.

### [30:01](https://www.youtube.com/watch?v=p9I2Ozvtpvc&t=1801s) - Kevin Moore

Oh, interesting. I guess it’s been pushing 20 years now, so I can remember being in those meetings. I worked on Longhorn at Microsoft, and frankly, there should probably be Harvard Business Review case studies about Longhorn. I’ve never talked about this in a public forum, so hopefully I won’t get roasted. It’ll be interesting to see whether other people I worked with have similar recollections because I’m still in contact with a few of them. Microsoft was so high after XP. It ran the world, and that success shaped what seemed possible inside the company at that moment. They thought, why not shoot the moon?

They were going to redo the UI stack in managed code and make it hardware-accelerated. They were going to redo the whole file system and make it a database. WinFS was this huge thing. Then there was identity, which they were going to redo as well. I was on the team when they said, “We’re done with IE.” They scrapped the IE team because the browser was done. Think about that 20 years ago, then think about how far the web has come in 20 years. You and I are doing high-quality HD video conferencing and recording on the web. I clicked a link. There was no 20-minute installation with three service packs.

I launched a URL. Yet 20 years ago, someone in Redmond decided the web was done and didn’t need a dedicated browser team. That’s hubris, right? It’s the hubris of thinking there’s clearly nothing else to do here and that we can do no wrong. Let’s shoot the moon. That confidence touched the browser, file system, identity, and UI stack at the same time. You lose track of what users really want. As far as I know, no user was asking for a database in the file system.

I’d go into meetings where people were talking about building business apps, and I’d ask, “Doesn’t this make more sense on the web?” Their response was rooted in needing to sell Windows, which is detached from asking what experience we want users to have and how easy it could be. Google talks a lot about the importance of starting with the user scenario. Whether Google does it perfectly is another question because it’s a huge company. In general, though, asking how to make the best possible user experience is different from asking how to keep selling Windows licenses, subscriptions, or next year’s device.

It becomes pretty clear when selling version plus one, the next device, or another subscription is the real concern instead of what’s best for the user. When you’re high on your own success and making huge amounts of money, it’s very easy to lose track of that. I don’t know whether there’s a solution other than finding perspective and modesty. It’s tricky because the people who can convince others to follow along and convince management that something is worth pursuing aren’t always the same people who have modesty and perspective. The overlap in that Venn diagram is pretty small.

### [33:30](https://www.youtube.com/watch?v=p9I2Ozvtpvc&t=2010s) - Kirupa

That overlap is tiny. Yeah, I think this is where product management really comes in. One of the big tensions is how much you optimize for the user experience and what the user wants versus how much you optimize for the business. Back then, Microsoft made almost all its money from selling shrink-wrap software. Even today, except for maybe Adobe, Autodesk, and a handful of other companies, tools are usually enablers of something else you’re trying to sell. You’re selling the shovels and pickaxes, and what people use them for varies from company to company.

From that point of view, I can see why Microsoft decided the internet wasn’t something to invest in too heavily beyond a certain point. They weren’t going to monetize it directly. If they made the browser much better, it might have been to their detriment by weakening the operating system’s value. Around that time, OpenOffice was starting to do big things. You also saw the rise of Zoho and other companies that became successful over the years by showing that you could avoid the operating system, avoid device lock-in, and do everything in the browser itself. That made the browser more than a document viewer. The browser becoming more successful posed a threat.

If my memory of history is correct, the internet became a major focus at Microsoft when Netscape started talking more loudly about running all the world’s software in the browser and no longer needing anything else. That got Microsoft interested because it would cannibalize its own product lines. The browser’s success could make Windows itself less central. If you look at it today, though, the amount of code in our browsers is probably greater than everything that was in XP or those earlier systems. We shifted the operating system upward rather than removing it.

You’re still running an operating system, but when you click Chrome, Safari, Firefox, Brave, or another browser, you’re basically loading another operating system. It has a URL bar rather than a Start menu and taskbar, but we’ve moved the complexity and functionality from one layer up to another. We replaced one kind of shell and interface with another. We’ve made significant progress in the last 10 years. The web is viable, and it’s a great platform for building apps. We’re doing this call right now without installing anything.

That’s because the installation already happened when you installed the browser: the codecs, many years of work with the W3C on standards for image and audio transmission, webcam and microphone access, and all the other mechanics of the internet. Years of platform work are hidden behind clicking one URL.

### [36:05](https://www.youtube.com/watch?v=p9I2Ozvtpvc&t=2165s) - Kevin Moore

If you think back even 15 years, it was all install questions: what version of Flash do you have installed? Or Silverlight, I mean?

### [36:11](https://www.youtube.com/watch?v=p9I2Ozvtpvc&t=2171s) - Kirupa

That was our world, exactly. I didn’t know a world beyond the plug-in. The plug-in gave me more capability than the browser could. It could bypass security restrictions in the browser and write to the file system. Whether that was a good idea goes back to how no one really understood the security implications at the time. The same was true of cross-site scripting or cookie harvesting, where a Flash file on one site could take a domain cookie from one domain to another. Those capabilities escaped restrictions the browser was trying to enforce. No one really knew all these technical and security details, but we did it.

Then we learned the hard way that they were bad. What’s interesting is thinking about all those little asterisks and then thinking about the iPhone. What was Apple’s reasoning? You could argue that Apple didn’t support Flash because it was a vector that bypassed the App Store and its revenue. I’m certainly speculating. I don’t do this for a living. There were also questions about security, privacy, and managing memory. All those experiences meant that the web had to get much better because this hugely powerful, important platform didn’t support Flash. I remember when YouTube required Flash. Then, for a while, you could prototype and play with YouTube without a Flash plug-in.

I’m sure most people don’t even remember this. Now, of course, YouTube doesn’t need a plug-in, and most of the web doesn’t need one. I also remember the plug-in that let you cache email. Was it Gears? Gears.

### [37:42](https://www.youtube.com/watch?v=p9I2Ozvtpvc&t=2262s) - Kevin Moore

So yeah, before that, Gears was such a cool little toy. That was before the browser supported any real notion of offline storage or anything else. That's part of why I'm proud of being at Google. Dart was a bet from the Chrome team, and Flutter also kind of came out of Chrome. There were flops and silly experiments that didn't pan out, but the Chrome team kept pushing on what the web platform could do. And it's cool to be part of that legacy, because now it really is a great platform.

### [38:19](https://www.youtube.com/watch?v=p9I2Ozvtpvc&t=2299s) - Kirupa

Great platform, yep. We often see the same pattern. Microsoft was investing heavily in the desktop. Today, we see similar tensions with app stores. The web could be better, but there are reasons certain vendors might prefer an app-store-first approach. History repeats itself with different players and a different venue. What always strikes me as fascinating is that we’ve moved into a world that is generally more open. You aren’t locked into any particular tool. Some companies made that possible, including Microsoft with VS Code. You can build on any platform without being locked into a particular language. You can write something competitive with Microsoft’s products. You can even write Swift in VS Code.

When we first started working at Microsoft, I never would have imagined tooling going that far. We have a truly agnostic platform for this, where the editor doesn’t dictate the operating system, language, or vendor you must target. I vividly remember being in Las Vegas for a conference. We always went to Las Vegas for conferences, although I’d much rather go to San Francisco. A director or GM at Microsoft, a couple of levels above me, was scoffing about an intern on the C# team. The intern had asked, “If I work on this, does this mean I can’t work on Mono in my free time?

Can’t I go back and continue contributing to Mono?” The executives were aghast, annoyed, and disgusted. They treated it like an absurd question from a silly kid who wanted to do open source. Then one day it was, “Oh yeah,. NET is all open source now.” You can run it on your Mac, in Docker, and everywhere else. That reversal would have been hard to imagine in the room in Las Vegas. From the outside, since I don’t work for Microsoft anymore, it has been super cool to see Microsoft recognize the value of openness and see the work it has done with VS Code and other things.

Even on the Wasm and engine side, there has been a little trolling on Twitter recently from some marketing people. In terms of Microsoft’s contributions to Chromium and related technologies, though, the company now contributes to shared infrastructure that its competitors also use. It’s great for the ecosystem.

### [40:35](https://www.youtube.com/watch?v=p9I2Ozvtpvc&t=2435s) - Kevin Moore

It’s exciting, and it makes me happy. Developers benefit, and there’s always a business angle to why these things are done. That’s fine because the pendulum has swung to a point where the business direction behind the decisions also benefits us greatly as developers and engineers. That wasn’t always the case. It wasn’t always true that what was good for the business and what was good for the user were on the same side of the coin. That’s where the culture of the company and what you ultimately wanted from your career often played a big role.

Do you want to sleep well at night knowing that you did right by the developer audience you care about? Or do you want to continue being seen as a key part of a team and its internal priorities? Docker was such a mind-blowing technology. The company is working hard to find a business model, and it’s tricky, but it gave this huge gift to the world. There was a realization that we can’t do another proprietary thing because someone else will just make an alternative. The same thing applies to databases at Microsoft or Google. I grew up thinking the choices were Oracle or SQL Server. Those were your options.

Now you look at AlloyDB, and it’s a crazy scale-out, highly resilient, redundant system that’s also Postgres-compatible. That’s the best of both worlds. Google has some spectacular special sauce there, and it’s good to have that differentiation. I’ve talked about Microsoft too much. My inner brain still says “we” when I think about Microsoft, and that never quite goes away. With Google and AlloyDB, though, lock-in can’t be the starting point. People always joked about making things sticky so customers had to stay. Your stickiness can be the user experience. It can be how many nines of reliability you offer.

The idea that users should code to your proprietary, one-off standard and be stuck with you is something the ecosystem has outgrown. We aren’t going to play that game anymore. I need to be able to move somewhere else tomorrow. A service is only one API call away from another cloud provider. A minor object feature can let you adapt your existing logic to work on something else that needs to be done. Portability itself has become part of what users expect. The lock-in has shifted. I don’t know exactly where it has shifted to, but it’s definitely no longer in the layer of APIs and tools that we use.

Back then, portability simply wasn’t the norm.

### [43:07](https://www.youtube.com/watch?v=p9I2Ozvtpvc&t=2587s) - Kirupa

If you wrote in Java, or in something like ASP or VBScript—VBScript isn't supported anymore—there were real walls between these worlds. I remember back in the day you'd literally set the script type in HTML and choose JavaScript or VBScript. We had all these options. What fun times. Oh my gosh.

### [43:27](https://www.youtube.com/watch?v=p9I2Ozvtpvc&t=2607s) - Kevin Moore

And people got mad at us. Dart wasn’t the first team to try to put another language in the browser. That’s actually a great object lesson. There were very aspirational people who founded the Dart team. They’re no longer at Google, but they’re still doing interesting things. It’s good to be audacious and say, “Let’s add a new language to the browser.” There was a lot of friction, even within the Chrome team. I think that’s to the Chrome team’s credit. From a pragmatic perspective, supporting two separate heaps and garbage collection systems is very difficult.

From an ecosystem perspective, can you imagine a world where Mozilla or Apple would do the work to support a different language? That didn’t work out, frankly. I quote this a few times, and it’s where my Iowa background comes out. I think it’s a Garth Brooks song: “Thank God for unanswered prayers.” Thank God we didn’t get into the browser the way we wanted. It would have been a pain in the rear end to manage release cycles and everything else. Web teams have to deal with versions and compatibility. There was also the Native Client experiment in Chrome, and later Portable Native Client, or PNaCl.

The idea was to take Intel assembly, and later ARM assembly, and see whether we could get assembly language running in the browser. There were cool things. I think Google Maps used it. A few products used NaCl and PNaCl, so these weren’t only theoretical experiments. Clearly, though, it was going to be a one-off and weird technology. Then Mozilla created asm. Js, and now we have WebAssembly. That sequence gradually moved the idea toward a shared standard. It’s delightful to be on two teams, Dart and Flutter, that came from the Chrome team and are now back in the browser.

We waited for WebGL to get good in the browser, and we waited for garbage collection to be supported well. Now we have crazy-fast runtime performance. We’re making it work in the browser today, and it’s not a one-off, Google-specific hack or a plug-in. It uses standard web technology that will soon be ubiquitous. Wasm and Wasm GC are progressing. If you want a crazy, high-performance language model and high-performance, cutting-edge graphics, some of what you can do in the browser today is within rounding error of AAA fidelity. The performance, frame rates, and everything else are spectacular, with no plug-ins or one-offs.

About three weeks ago, we finally flipped all the bits so it runs on Firefox. That’s spectacular. From an ecosystem perspective, for a while it was Chrome-only, coming from a Google open source project, so people rolled their eyes. Now it works in Firefox, and the discussion has changed. Instead of being a Chrome-only, Google-only thing, the question is, “Where’s Safari?” It’s amazing how a little bit of work shifts that conversation and changes how people interpret the project. This gets back to our old discussions about browser support and ecosystems. Not having to deal with IE anymore makes our lives much easier, but I hate monocultures too.

Having to think about Firefox and Safari matters, although I’ll bite my tongue on Safari if I can’t say something nice. They’re certainly doing a lot of work. Standards work is slow and tedious, but it creates a better web. We’ve had to think harder about things. That comes with legacy constraints, workarounds, and less-than-ideal results, but the web is stronger for it. I’ve always remembered the saying, which I think Al Gore quoted in one of his movies: “If you want to go fast, go alone. If you want to go far, go together.” It seems like more people are going together on things. The results are spectacular.

It’s the best time ever to be in computing. It’s super exciting.

### [47:36](https://www.youtube.com/watch?v=p9I2Ozvtpvc&t=2856s) - Kirupa

No, I completely agree. And when you go beyond that into AI and related work, the bar is so much lower now for someone who wants to build an app or build technology. Just go do it. I don't know much about WebAssembly myself, honestly, but when I found out Figma was built using that kind of technology, it made me stop and think. Something I use all the time for graphic design was only made possible because of these platform-level advances. I saw the founders' postmortem after the acquisition, where they talked about the early experiments and how it all began.

I was like, wow, that's where it started, and here's how it ended, all because the technology unlocked a whole class of applications they couldn't have built performantly otherwise. That makes me think about what's possible next, and about the things being built right now that no one fully understands yet, but that could be game changers in their own way.

### [48:34](https://www.youtube.com/watch?v=p9I2Ozvtpvc&t=2914s) - Kevin Moore

Yeah. I know you've spent some time on the AI stuff. I haven't, at least not on the machine-learning side. I've played with Bard a bit and some of the other tools. I think it'll almost certainly be a big deal. I'm still curious to see what happens on the programming side as we scale from one page of code to full libraries, frameworks, and applications. I'm glad there are many players in the space, and I'm glad there are so many options in the open realm. Monocultures are bad, so it's good to have choice and variety. It's cool to see it evolving so...

### [49:14](https://www.youtube.com/watch?v=p9I2Ozvtpvc&t=2954s) - Kirupa

... Quickly as well. I've had to speculate that maybe prompting becomes a kind of programming language. It's a little like Terraform, where you have a YAML file and say, I want infrastructure A, B, and C, I want it to talk to A, B, and C over here, send it to some magical machine, and it comes back with a URL endpoint you can access. I may not understand everything happening under the hood, but I know the syntax well enough to structure the request so the system can understand it. In many ways prompt engineering could become its own programming language. BASIC made English-like expressions more normal.

Python took that to another level by making power accessible quickly without too much syntax. Now AI might be the next step, where you go to ChatGPT or Bard and type in pure English. Then it does things, and the quality of how you specified what needed to be done determines what you get. You can be extremely detailed and specify conditions: do this, don't do this, and so on. That becomes its own language. I think that's where things start to get really interesting.

### [50:28](https://www.youtube.com/watch?v=p9I2Ozvtpvc&t=3028s) - Kevin Moore

I don't want my old-fogey side to show, but I've seen the demos. I'm compelled by the demos, and the example is people doing real work. If you're doing like write this paragraph for me, or. The joke—I've seen it a few times now, and I think it's hilarious—is the people who say, okay, take these three bullets and turn this into three paragraphs. So it seems like I have some gravitas. Then on the receiving end, it's like: okay, take these three paragraphs, just give me the three bullets.

It's like what an insane waste of CPU cycles, megawatts, or whatever you want to call it, like, is this what we're going to do now? It's certainly compelling, but also, like English can be vague, and when you think about an algorithm, it's like no, I want the conditions and preconditions to be very explicit. And so I have a strict notion of what I want to run and when I want it to run and how I want it to run, and all the exceptions and everything else.

And so, starting with vague language and getting there, like for UI, I'm super excited and I think it'll be interesting to see, certainly in our world with Flutter and UI. So it's like, okay, like, can I go in and say, like, I want the UI to be more playful. Again, this is me just speculating; I don't know anyone working on any of this stuff, but I think it's interesting to play with and think about. I want the UI to be more fun. I want it to be looser or tighter, I want the margins bigger. I want it to be more—I don't know.

I could imagine giving Bard a Windows 95 app. You can imagine this at least, and say, okay, make this look like a Star Trek UI, and figure that out. It's like, okay, but like, what's the artifact and code I generate, whether it's Flutter code or CSS and HTML, whatever? Like, what's the artifact you generate underneath so that works and accessibility still works and the hit testers are still set up? Like, I'm sure people are working on it, but like, in my mind I don't have that threaded through yet. Like, how are we going to make that actually work? I'm excited by it.

I'm ultimately more hopeful, until I saw people on Twitter saying like I have no program experience, but here I created a snake game. I was just like, wow, like for not having any idea of like what in the code does what, the AI system was able to help them and the quality of it was reasonably well done. Also, it wasn't like I was like: yeah, it's actually not bad. Like it takes account of the refresh rate- so it's not like faster for some people or some other people- takes into account, like various input modes.

I'm like that was actually well thought out by the AI system, which was, of course, trained probably on some very well made snake games. It pattern-matched from A to B, like what needs to be done here. But at a certain point, though, a lot of what we do is kind of like: go through the cycles again, right, it's just repeating what we've done in the past, just trying something slightly different, or maybe the same exact thing without any variation, because something was really successful 20 years ago, and you're like: yeah, I'm not changing that formula at all. I'm gonna repeat the exact same thing right now.

The context might be different, but the output and inputs are going to be exactly the same. It's exactly what an AI system does in many ways, really well. And I think the AI and UI part always excites me, and actually someone I talked to earlier, Chloe Barrow, where I talked exactly about this thing- she's an HCI computer researcher, and, like human computer interaction. The thought is: what if we get to the point where the UI that I'm interacting with is dynamically generated by the AI system?

I might prefer colors that are more in the blue color, like I like blue colors and like gradients, for example, you might prefer, like more like natural colors, like grays and greens and so on. And it doesn't matter, because when I'm viewing an app, it's going to be dynamically recreated visually for my particular taste. The logo might remain the same for branding purposes, but everything in the app is going to be different. Like I prefer vertical navigation, you might prefer horizontal navigation, or you might prefer more images, less text and things like that.

Each person who has their own personalized app experience, where the core experience will be whatever the optimized outcome is, like I want to sell more products, I want people to understand my portfolio, things like that. That goal will be the only constant there, but how it translates in success for each person is going to be completely hyper-personalized, and that's the part I'm actually very excited about. It makes me think of so many things. The way I've been thinking about the AI impact is as the difference between carpentry and architecture. We need a world of savvy carpenters, like craftspeople, right, which is and driving through suburbia or whatever else.

In Seattle, I'm happy to see local governments and things that are getting smarter and Savvy about density. Right, with the housing crisis and everything else. You build, and it ends up being a lot of cookie cutter, right, there's only so many ways to hang a door and do it correctly. I remember we built a backyard cottage, behind my house a few years ago before COVID. That was great, but these two craftsmen were back there cranking along by themselves. They built a whole mini-house like it was a little bit kind of a template-based thing.

It's meant to be fast, but they're just experts and we wanted the staircase to go off to one side instead of the front so we could pull a car in the back, whatever. And he's like, oh, okay, so you need to. You need a deck this wide. It's like, okay, is it 30 inches or 36 inches? And I listened to him, to the phone call. He's like, Okay, and like then I just watched him and he didn't have a design or like it wasn't IKEA- with like a list of parts. He's like, oh, I need to build a deck.

And so you think about, like what's the minimum width so the door can open, so that it's ADA compliant, and what's the run of the stairs that you can get up safely? All the rules and it's just all in his head. And he just sat there and built the deck and I'm sitting there blown away. I was like, how could you do that? And then what I realized was that you and I could do the same. Someone's Like, okay, make me a Chrome app to like take in first name, last name and date of birth, whatever else.

And we think, oh yeah, master-detail forms, and like it's the exact same thing. And so to bring it around to AI- sorry for my long analogy, but like things that we think are magical, right, it's all relative, and so maybe it's like if you're the very beginning of programming, the AI stuff is spectacular. Just like if you're at the very beginning of music, what's available to you today versus 30 years ago with YouTube and everything else- like you can get started so fast.

And then, if you're the best, if you're the top percentile of whether it's graphic design or audio design or writing fiction, whatever, you're probably safe, right, because someone has to build inputs to the models.

And then what I'm worried about is like I think there's a lot of folks and jobs in the middle there where you've gotten Savvy, right, like you spent your life making reports or you spent your life doing summaries of the quarter or whatever, or building master-detail forms over data stuff, and I think those jobs are going to be the most tricky because there's a lot of people making a lot of money and paying mortgages and sending kids to school and saving for retirement doing that type of work and I struggle to see how that's going to survive, You'll have to adapt.

### [57:45](https://www.youtube.com/watch?v=p9I2Ozvtpvc&t=3465s) - Kirupa

Yeah, it's going to be very tricky. We call it creative destruction, where you're supposed to move to higher-value work. It's one thing to say, okay, I used to do physical billboards and print work and now I use Photoshop. That's a learning curve, but there's still a next step. Here, I think we're seeing whole categories of work disappear entirely. Take summarizing a document. There's only so much you can do there. Past a certain point there isn't some post-summarization activity waiting on the other side that can't also be automated away—unless you're the person writing the AI that does the summarization.

### [58:36](https://www.youtube.com/watch?v=p9I2Ozvtpvc&t=3516s) - Kevin Moore

In my mind, it's Craigslist and classified ads, yeah like, and again I mean at the risk of being philosophical as we approach an hour, but maybe it's appropriate, right like there was so much good journalism that was supported, investigative journalism, just local awareness of news, even bringing together community, right, like for good or bad, like locally owned paper, like a sense of community and connection people around me, like it was all built on, like looking for used car 1980s or newer, four doors prefer AC. Like, and then Craigslist came along.

It was spectacular, like it changed the whole industry, but also like all the importance of Journalism and community and local identity. The stuff is just like, oh, it's built on, like it's like the joke about the whole internet being built on this tiny little thing maintained by someone in Nebraska. Like the idea of, like civility and freedom of information and awareness of non, oh, not overly polarized. News—oh yeah—was built on classified ads and those make no sense anymore because Craigslist and eBay, yeah, I mean, this goes back to what we talked about earlier on.

Like do you work on the next big thing or do you make the existing one a little bit better? And you look at why Craigslist succeeded. It totally makes sense because the user convenience was massive. You had this database. You could see, like people who were there. It's the chicken-and-egg problem with any social network: I mean classify something, people posting them. You need people searching for them and the timing of it was just right. So the Craigslist team did a good job on that angle. And then you go back and go like could the classified industry have foreseen this? Could they have seen that? The internet was getting really big.

People are using that as their primary source for, like basic information, news they go to search. Look things up in AltaVista at the time, or Lycos. Now everyone is like using Google and they try to go with like the most relevant thing there. How can we adapt our service to make that happen?

And oftentimes it goes back to we're so pigeonholed into like the next milestone, the next thing needs to be done and so on, that we never, we could never take- actually we don't have the bandwidth to take a step back and go like: wait, are we still solving a problem worth solving or has the world moved on to something completely new? And I think that's the question there. Well, and it's also like if you're a horseshoer or a livery stable company, like, are you in a better position to go build automobiles than just someone starting from scratch?

You're probably in a worse state, like just because you have your existing infrastructure and existing customers. I mean, you look at all the politics around green energy and people like, oh, we'll push hydrogen. Or, in Texas, we're going to tax electric vehicles far more than the gas tax we get from. Like it's like, oh, what do you do? To like, maintain a hold, and just be utterly resistant to change. So I think there's economic and societal things we will need to reflect on. I have two brothers who are musicians, and so, thinking back you saw the ads in the early 1900s about the evil, like the evil robot.

It could be ML today. Like it's so funny. But like the scary ads of, like the machine recording. It's like if you wanted to enjoy music, you had to have someone perform it live for you, and so that means a lot more. There's a lot of mediocre musicians in the world, because it's like if you wanted music at your theater, you had to hire someone, and so it's this interesting dichotomy which is it's easier to get great at music today than it ever has been.

I can watch lessons from the best guitar player, the best trombone player, whatever, for hours and hours for free for the cost of a YouTube subscription or watching some ads. At the same time, the demand for that skill as a way to make money is dramatically lower.

### [1:02:28](https://www.youtube.com/watch?v=p9I2Ozvtpvc&t=3748s) - Kirupa

Exactly. I recently wrote about the power law and how it applies to AI. We already live in a world where a very small number of people capture most of the value and there's a long tail of everyone else along for the ride. AI narrows that even more. Take music. I might once have been willing to pay some amount for background music for a video. Now I can go to an AI tool, type in what I want, and download something. Is it as good as what a great human musician would create? Probably not. But I don't have the trained ear to say this is absolutely the wrong thing.

So from my point of view it meets my price point of free and the quality is good enough. And that comes at the direct expense of a qualified musician who might otherwise have done that work. It feels like the same pattern you saw with Fiverr and other services. Before, maybe you hired a logo agency. Now you pay ten dollars to a website and get eight variations and say, okay, good enough. It's the same story repeating itself over and over again.

### [1:03:32](https://www.youtube.com/watch?v=p9I2Ozvtpvc&t=3812s) - Kevin Moore

Yeah. From what I've heard, it isn't actually a Chinese proverb and is probably misattributed, but the phrase 'may you live in interesting times' still feels appropriate. We certainly live in interesting times.

### [1:03:47](https://www.youtube.com/watch?v=p9I2Ozvtpvc&t=3827s) - Kirupa

Very much so. You mentioned earlier that being really good at something may give you a better chance of succeeding, but what if the thing you're best at no longer matters? What if you're the best horseshoer in Washington State? Does it really matter if there aren't that many net-new horses on the road anymore? It's not a thing at scale the way it used to be. How do you figure out that you're exceptionally good at something that no longer matters? Because you've probably tied a lot of your identity and your history up in that skill. How do you ever realize that?

### [1:04:23](https://www.youtube.com/watch?v=p9I2Ozvtpvc&t=3863s) - Kevin Moore

Horseshoeing. So let's talk about horseshoeing. It's just as an example here, right? I think it's an interesting analogy which is like the difference between the best horseshoer and the average horseshoer. Does it practically matter? Yesterday for a, for a nine-year-old birthday party, we went to a stable and they rode in a circle on their horses. So I just I have a very visceral thing about horseshoeing. It's like the difference between the average horseshoer and the best horseshoer. Is it enough to really matter?

I mean, maybe if you're racing, maybe if you're at the Kentucky Derby, and then maybe if you are the best horseshoer—not in Washington State, but in Kentucky, right, you probably still have a job, right, because there's people who still bet on horse racing, right, and those things really matter. But I like this analogy, even with lifting weights, right, like the difference between the average strength of a 25 year old man versus the strongest 25 year old man in the world, maybe 3x or 4X, it's not 10x by any means- right, like is that I can bench 150 pounds versus 1500, like no one can bench 1500.

The difference between the average surgeon and the best surgeon is probably 10x or maybe 100x. The difference between the average musician the average basketball player. Like there are skill based, mind-based things. I guess horseshoeing in some respect is a skill right, but, like there's no 100 x or 1000 x thing there. And so the fact that we have technology now where, if that person's in Milford, Iowa or Seattle Washington or Mumbai or Kathmandu, like the fact that they, we have access to that person's brain and they have the tools to do things, that's spectacular.

Because, before I mean as much as I dislike Friedman from the Times for many reasons, he did say some good things. Right, which is like, oh, like you'd rather be a C student from the States than an A-plus student in many places in the developing world, and that's quickly changing and I think generally that's a good thing. Right, like people can be empowered in all these places. Like all you need is an internet connection and a browser, like a $200 Chromebook, for most intents and purposes, is as good as the $3500 Vision Pro, whatever, in terms of being able to get a lot of stuff done on the internet.

And that's a great thing, the democratization of those things is great, but there certainly will be a cost. I guess we'll see. Yeah, some will benefit, some will not, because it goes back to the thing I mentioned earlier about: at some point the gains or the benefit you have will be driven down to zero by competition or so on.

So any unique monopoly that we all may have had in like being able to hire the best talent or being able to do some of these things as a country, for example, are going to be made more competitive as quality of life increases worldwide, and people might decide to stay there and build what use their brain power to benefit their own localities, as opposed to coming here, which is a net gain for them. A loss for us and we still might be still on the positive end of the benefits.

It is gradually going to be decreasing to zero, if the equation of that whole is true, for, every other part of the world- and I think to try to wrap things up a little bit too, I need to go, I have other, I have web things to go worry about, which is exciting. So anyone who can bring about the stuff I'm worrying about is great, but thinking about our history too, like, there's a power of platforms, and so I know you're doing work, but I actually have no idea how much of what you're working on now is public. It's really cool. Folks pay attention. Yeah, not public at all.

Yeah, I can't say it. But, like building tools and empowering people, like we're in the same business. I think we both loved it. Clearly, we've come back around. I took a break to go work on other things and came back to platform stuff. So, building tools, building platforms—and I think broadly about platforms, right, whether it's a government or a culture, a tool. Like how to make people feel safe, make people feel productive, make people feel connected. Letting people contribute, helping people learn, feel safe in that space, feel free to make mistakes.

Paying attention to things that matter versus things that don't and you can think about in terms of, like the operating system you run, or you can think about your gender identity, right, like, should it matter, and then yeah, like, no, like, and that's good. So I'm excited to be working on the platform. I am here, and then you think more broadly about the communities we build, and the governments we support and the policies we support.

Like, it's all about platforms and connections and not thinking zero-sum, and that's, I think, also the power of platform thinking right, which is, if you think in terms of like, well, I want the biggest slice of the pie and I don't care how big the pie is. Like, it's very reductionist thinking and it's very backwards thinking. And if you're always thinking about growth and change is a good thing and making the pie bigger, there's something really good there: a way to have fun at work and also have a good life and have a world we want to live in.

So, I'm excited to be on that journey and road with you too, and hopefully we'll keep it up.

### [1:09:11](https://www.youtube.com/watch?v=p9I2Ozvtpvc&t=4151s) - Kirupa

Well, Kevin, there's no better way for me to conclude this than what you just said. That was fantastic.

### [1:09:16](https://www.youtube.com/watch?v=p9I2Ozvtpvc&t=4156s) - Kevin Moore

Okay, it's all recorded, right? Because there's no way I could come up with that again on command. I'm really excited you're doing this. Dimitri and some of the other people we both work with—it's so good to capture conversations like this. So keep having great conversations. Thanks for doing this.

### [1:09:34](https://www.youtube.com/watch?v=p9I2Ozvtpvc&t=4174s) - Kirupa

Yeah, anytime. And absolutely, next time around, hopefully I'll have you again. There's much more to talk about. I can totally sense it. We could go on for hours, I'm sure. Thanks, Kevin.


[Browse all Interviews with Creative People](https://www.kirupa.com/podcast/index.htm).