CoRecursive: Coding Stories - Tech Talk: The Business Of Developer Tools With Lee Edwards

Episode Date: December 18, 2019

How do you build a business around tools for software engineers? Adam talks to Lee Edwards, a VC who spends a lot of time thinking about this question. "When I think about is this a good business, I think about is there value Accruing. The question is just how much. The question about is it a venture-backed business? The very, very oversimplified answer is do you believe you can get $100 million in revenue within 10 years? And those numbers are kind of fudgy. But if you can do that, you can IPO a company and it's kind of amazing that PagerDuty and Twillio each do one thing well and they're multibillion-dollar companies. " "Another interesting thing that venture capitalists talk about behind closed doors and probably never tweet about or say publicly because it makes them look bad. But you do often wonder if the founder of a dev tool company, a lot of times they're really altruistic and you know, I feel this way too, right? But venture capitalists are like, wait, don't give your stuff away for free. And it can sometimes be kind of like a conflict. I think when you're looking for an open-source founder, you need to look for someone as a VC that actually does want to make everyone money." Show Notes: Root VC The Business Value of Developer Relations - Mary Thengvall Code Climate Particle.io FlexPort

But I do think that if you build a really cool product and you want the most people to use it, you do need to be able to support it, and you need the money to do that, and you want to pay the people that are developing it. Welcome to Code Recursive, where we bring you discussions with thought leaders in the world of software development. I am Adam, your host. Hey, how's it going? Hey, not too bad. Oh wait, my audio is coming through the wrong thing here. Today I talked to Lee Edwards. Lee is a venture capitalist. He is a partner at Root Ventures. One area Lee is interested in is what I'm going to call developer tools.

So what is a developer tool? Well, it could be a IDE. It could be a API like Twilio or Stripe. It could be a database, a code linter. It could be a program verification tool or even a programming language. So how do you build a business around a developer tool? Well, if you listen, I will pitch Lee a random bad idea for a developer tool, and he's nice enough to play along and explain how he thinks about this space.

And along the way, we talk about making a business out of an open source project, why he's excited about Rust and WebAssembly, and a little bit about Star Trek versus Black Mirror. I have this idea. So here's my dev tool, right? I built something, I'm going to call it the pacifier bot.

So it finds like merge conflicts on GitHub and then it resolves them. It says like, hey, on the pacifier bot, you have this merge conflict. I made a new PR, your conflict is resolved. So this is my open source project. How do I turn that into a business? Yeah. So there's two kind of questions, right? About do you make some, like making

something a business? Sure. And then out here in the Valley, I think a lot of times people conflate that with make that a venture backed business. And there are two very, very different questions. Yeah. Start with the business. I mean, I think it seems obvious that if you built something that would save people time and kind of let them iterate faster, that people would pay money for it. And I think the question is, how much money would they pay for it? Sometimes that depends on time. I mean, sometimes I talked to VPs, directors, managers and engineering, and they'll do the math. Not everyone does this. But a lot of people will literally do the math and just say, Hey, you're saving every engineer one hour per day. I'm calculating this in their salary. This product is worth a lot.

So yeah, it feels like something like that if there's literal time saved in fixing the merge conflicts. But then I think the second order thing that non-developers, investors that are non-developers miss is there's kind of a value to the intangible speed of iteration. I think the reason that things like CI, CD have been valuable, the companies are willing to pay a lot of money for it, is just the difference in the world. Like I'm old enough to remember when most software was printed on a Golden Master and then shipped out in CDs and put in a box to sit on a shelf at Radio Shack and eventually Office Depot and Staples. So the costs of shipping bugs, the costs of shipping security vulnerabilities, and just the speed of iteration that developers can get feedback

on how their software is working, on how users are working on the software. Something like what you described, if my pull requests can get merged faster, I can answer those questions faster. And that has a flywheel effect on productivity. So when I think about, is this a good business? I think about, is there value accruing? And it feels like the answer is yes. The question is just how much. So I feel like you could probably build this, bootstrap it, maybe to distribute it. Distribution is always hard. Tweet about it, build a community.

Open source is one good way to build a community. Making things that are... Writing content is another big one, right? I don't know how much content you can build around this tool in particular, but a lot of things. You can write a great blog post, distribute it on Hacker News. And then certainly putting things in the GitHub marketplace, AWS marketplace, getting things on Product Hunt.

I mean, there's a whole... All of the stuff around developer relations, developer community building, it's a whole discipline. I mean, at this point, it is a very real thing. It's like sales and marketing. There are experts on this. There are now books about developer go-to-market, developer relations. Mary Thangball has a really great book called The Business of Developer Advocacy, which I think may be the first book in space. And I think the question about, is it a venture-backed business?

The very, very oversimplified answer is, do you believe you can get $100 million in revenue within 10 years? And those numbers are kind of fudgy, but if you can do that, you can IPO a company. And it's surprising to me. I'm now kind of really grokking and integrating this information, but it's kind of amazing that PagerDuty and Twilio each do one thing well,

and they're multi-billion dollar companies. Yeah. There's a lot of things like that out there now. Slack, I wouldn't call strictly a developer tool, but it's definitely fair to say that developers adopted it first. And probably a lot of the accrued value actually comes from product development, engineering design teams. You can think about things like HashiCorp and Pivotal. They do multiple things, but within the space of deployment and cloud services. CircleCI is not a billion-dollar company yet.

I could be wrong about that. They did raise a bunch of money, but certainly not public yet. So you can imagine, I think, doing something really, really well. This idea seems a little hard to imagine. If you back up the envelope, could you make $100 million in revenue? But you could have... Maybe there's something else there, right? But a suggestion I would give to someone if they sort they were pitching me about, should I start this business? Or even just

discussing with me over coffee, should I start this business? Are there other verticals in the space of doing pull requests, removing friction? So I think we all know so much time, the larger the organization is, the amount of time that you spend discussing with other people, your pull requests and code styles. Like, you know, there are relatively smaller businesses and things like Code Climate and Code Cove. But maybe if you roll this up and you can kind of put it under a brand.

And again, like sort of, you accrue a lot of value if you can basically provably reduce the amount of time a pull request is open. Yeah, yeah. I think that's great feedback. The space for actually improving people's code via a pull request process. In the most general case, it's like, hey, your Python 2 library, here's a new pull request. Or even machine translation. I mean, now we can go really crazy on this. And I think most venture-backed ideas are really crazy. But machine translation, if you're familiar with this idea, works really well with natural

language. It's getting better and better. I think it's one area of AI that's improving insanely fast. But people are also using machine translation for weird things like, can I translate text into images as though the images were a language and you could maybe imagine someone doing that with code or with the ast that the code parses into there could be something interesting there what's an example is it like hey ruby's not cool anymore this is now in rust yeah maybe python 2 to python 3 is the easiest version but yeah like that would be crazy i, even if you were to move from a interpreted garbage collected language to a direct memory managed language like Rust, that sounds insanely hard.

I don't know, might be valuable. I mean, my original use case of the merge conflicts, there's very easy ones that can be done there. That's probably the bulk of merge conflicts. But in the hardest case, it's impossible. Like it's just like you need to understand what the developer's intent was and whatever. Oh, totally.

I'm just thinking about there's areas where it's like a seemingly intractable problem, but there's like specific cases of it that are like totally able to be done. Yeah, and I think another problem that I've seen with a lot of in the past, a lot of this code generation type stuff, the complaint is, hey, this code is not very readable. My intent isn't really written here. It doesn't conform to the style guides of my company. But in this part, your readers that know more about or your listeners that know more about this than I do might just shake their heads and be like, this VC has no idea. This technology is not as advanced yet. But I do think that a lot of times the idea of like style transfer, and I think a lot of the machine learning algorithms around translation, this is one thing they do really

well, at least in some of the demos I've seen. And certainly it's moving in that direction. So you could imagine a machine translation of one language to another, where style is part of what it's translating and understanding. So it's like those image apps where it's, I made you look like Picasso, but instead you're like, hey, I made the code look like it was DHH or somebody. Yeah. I wonder if a little sci-fi here, but certainly in a language like Ruby or in like JavaScript, the style is just like pretty wildly different. Unlike Go where it's sort of, you will use the style or this won't compile, you know, or even like Python's tabs and spaces. But yeah, I wonder if you could

sort of train on a corpus of a particular style of Ruby or Python and then the code you generate is actually in a similar style. That's probably the case. The interesting thing is like there is a large corpus out there with just all the open source software, right? If you were able to somehow learn on it. I wonder if that's part of why Microsoft views GitHub as so valuable. I think for a long time, people have talked about, hey, GitHub has all the code in the world. And people think about proprietary data that only your company has access to, that that sort of makes your company valuable. People have talked about that with GitHub forever.

It's unclear to me that GitHub has ever taken advantage of that data. Part of it might be privacy. But yeah, it feels like there is opportunity there that maybe Microsoft saw when they put that huge price tag on GitHub. Yeah. So for the conflict thing, it's interesting because all these Git repositories, they have their full history and you can actually look through and find places where there was a conflict that somebody resolved, right? So in fact, there is like, you could build a giant corpus of people resolving conflicts in

code and use that to... Yeah. The other thing that would be funny is I think definitely human conflict, emotional conflict often comes into play with merge conflicts. Imagine you go over to the person's desk and you're like, when you have to figure out like, okay, well, what was your intent here? And then sometimes you realize like, wait, I'm trying to move the code in a different direction than you are. Let's argue about this. Another thing that I would say to someone that wanted to try to turn this into a venture-backed business as opposed to kind of a lifestyle cash business is the very biggest version of this idea would be to just sort of fully vertically integrate this stuff. And what is, if you reimagine source control and code collaboration

with this as a core concept, if that's possible, I don't know if that's possible, but could you essentially build a source code repository like GitHub, GitLab, Competitor? And that sounds insane. But if your specific idea, and maybe it's this idea or something else, if your specific idea is so powerful

that you actually create a much better experience much higher quality product for the thing that you're integrating with phrase i'll use this like sort of how do you capture the most value how do you build the biggest version of this business and an extremely ambitious founder might be able to do that certainly that's the case with github right i mean what was the advantage of github over just hosting git on a server part of it is hosting a git server which is annoying but part of it was the pull request and the collaboration. That idea was in the community. That idea was so powerful that they verticalized and made this huge, huge business. It's a simple idea if you think about it. Just the idea of reducing the feedback loop between,

hey, we're both pushing code, but why don't we put the communication in one area that we both go to every day? So they figured out how to completely verticalize that product. They could have made the pull request like a command line plugin for Git or something, right? See, my ideas just aren't big enough. That's the problem here. Well, a lot of times a smaller idea can be made bigger. That was maybe a little bit of my point. But the other thing is, I think it's kind of silly. Sometimes people try to force their idea into a venture-backable business model rather than just say, Hey, can I provide this product on the internet and make some money? Like Patrick McKenzie is kind of the master of this. He, you

know, he's blogged a lot about it. I think it was like the crossword puzzle thing was his first one. Bingo. It's bingo. Bingo is bingo. Yeah. I mean, bingo cards, I think. That's awesome. Yeah. And 37signal is a similar thing, right? They're always kind of flaunting that, you know, they didn't raise venture capital and they made a ton of money. So can an open source product become a business?

Yeah. Like, so the code for my merging logic is like sitting in Git and like people add things, right? They're like, oh, I added a merge strategy for YAML documents or whatever. Yeah. Do you mean just in the sense that the product is open source and you need to now tack on a monetization model to it? Yeah.

Yeah. I definitely think that's possible. There's a couple ways to do it. One is the concept of open core, which is relatively controversial, but there may be part of your code base that's closed source. And there was kind of a Twitter kerfuffle a month or two ago about a conference I actually went to that was called OpenCore Conference. And the open source purists were sort of like, this is not open source,

because part of it is closed source. That is literally true. But there's a lot of power in open source and open source communities. But I think that sometimes, in order to really make a project big, you do need kind of money to run the project. I mean, like NPM has money behind it. And then you look at a lot of the things that are sponsored and run internally by big companies like React, React Native, TensorFlow, Keras. If those were not inside of large companies,

they would probably have to be venture-backed businesses. And yeah, certainly, I think it's very hard to imagine what venture-backed React Native is. You can't force someone to pay for it. But there are other things like if you think of InfluxDB has a model where their influx is open source, there are people forking and using influx for different kinds of things. But if you want them to host it, you pay money. And then the other thing that I think is interesting is if people fork your open source project and try to build a business around it, that's what a lot of investors are worried about and a lot of people starting these kinds of businesses. You have to wonder, can the people that fork your project actually build a bigger community than you?

Because I think communities are very sticky and in a lot of sense, they're what investors would call a moat or a competitive advantage. It's hard to build a loyal community, but once you have, a lot of times they're not really interested in playing with fork. That's not always true, right? I mean, I think in the case of Linux, there's lots of different communities that have forked off of a lot of the original Linux code like Ubuntu and CoreOS. AWS is what I think of, right? Like InfluxDB. I don't even, I'm not aware of what InfluxDB is. But like if I build a database, like Amazon's just gonna offer hosting for it. And people will use that just because they're already on AWS, right? Yes. So I think a lot of times, I have definitely heard people being upset about this.

The AWS will just grab this stuff. That's a real risk. And I think that's one thing that OpenCore tries to do. There are different kinds of open source licenses that I think try to protect you from some of this stuff. But when you try to cut off the cloud hosting companies from being able to do this, you sometimes inadvertently cut off your open source developers from being able to do it for free. So it's a difficult needle thread, but certainly Hortonworks and Elastic Search. And I really like Algolia as well. I'm not a public company yet, but I had to predict it's certainly going public in the next couple of years. AWS can and does host stuff like that. But for whatever reason, people are picking the first party hosting. Same with WordPress, right? Automatic is a very large, valuable company that...

Actually, I'm sure that if you interviewed Matt Mullenweg and asked him what things would he have done differently, he probably would say that he wished he'd done paid hosting earlier and things like WP Engine and these folks never even got a chance to come off the ground. But now that WordPress.com does this, people are picking first party because it just feels better supported. And it probably is better supported. The people that wrote the software, certainly the ones that are going to

make sure that all the new features are getting updated in the cloud hosted platform that are probably looking at how users are using the product and trying to integrate that into the hosted version. I mean, I have to assume, yeah, like, obviously, they haven't captured all that value. Like WordPress runs so much stuff, right? Yeah, that is definitely one way to look at it, right? I mean, I think it I'm pretty sure it's hundreds of millions of revenue, but, you know, it ought to be a trillion dollar company company probably, I guess. That is a big risk a lot of these companies run is that the free version of your product might be so good that no one wants to really pay for the paid one. Yeah. I assume it's intentional on his basis.

He wants it to be what it is. Yeah, I think that's right. I mean, that's another interesting thing that venture capitalists talk about behind closed doors and probably never tweet about or say publicly because it makes them look bad. But you do often wonder if a founder of a DevTool company, a lot of times they're really altruistic. And I feel this way too, right? It's like how much has been given to me for free? I had a great job where I was writing Ruby code

and Rails code. And both of those are open source and free. People gave that to me. And I had an awesome job that I loved. And so you can't help but feel grateful about that and want to contribute back to the system. But venture capitalists are like, wait, don't give your stuff away for free. And it can sometimes be kind of like a conflict. I think when you're looking for an open source founder, you need to look for someone as a VC that actually does want to make everyone money. And so you can kind of view that as greedy and capitalist, and maybe it is. But I do think that if you build a really cool product, and you want the most people

to use it, you do need to be able to support it. And you need the money to do that. And you want to pay the people that are developing it. I think a lot of times the totally pure open source models, there's lots of discussion in the community about, hey, these people are spending their time, shouldn't we figure out a way to pay them? And so you see things like Patreon and I can't remember the names of the other products, but there are Bug Bounty and stuff like that where you want to pay people for their time. Another good way to do that,

to pay them meaningful money, is to venture back your company and pay your employees health insurance and benefits. Yeah, I had Heather Miller on the show and she was talking about open source infrastructure, she called it, right? And I, I guess kind of the, you know, the open SSL story where like, the guy was making minimal money to keep this project going that everybody depended on. And it led to a major problem.

I remember that one. There must be something between, there needs to be some model to keep these open source projects that we all depend on able to support themselves or something i don't know yeah i think that's right and i think that you can crank the lever too far there's been tons of backlash against sun for them you know really really trying to pull java into proprietary there is a kind of thing with open source where it's um it's the word that they use for this it's kind of like crossing the rubicon or what's the unringing a bell right once something is source, it's really difficult to then make it

closed source. So I think it is important when you do this strategy to think about what's a fair and acceptable way that your community is not going to get backlash against you for monetizing. Part of that's just being a good actor in the system and being the adulterous. So the other thing I guess I would say is it's not any different in open source than it is with other businesses. I think that for the most part, there are certainly exceptions. But for the most part,

people that start businesses, I think want to see their thing in the world. And I think that they believe it's a good thing and they believe everyone should do it. And they believe they can make money while doing good. I really, you may call me naive, but I really, I mean, I talk with a lot of founders and I think that that's the case. I tend to be much more of an optimist. I'm much more into Star Trek than Black Mirror. But Black Mirror is so good. I don't know.

To be honest, I don't actually really watch it and can't watch it. I think that is kind of a tangent, but I think that building dystopian sci-fi these days is almost too easy. Nobody's building Gene Roddenberry Star Trek anymore. And I think that it takesstopian sci-fi these days is almost too easy. Nobody's building Gene Roddenberry Star Trek anymore. And I think that it takes more imagination and can be more difficult to imagine how does technology make our world better in the future. I think that was just like one of the most brilliant, beautiful things about Star Trek originally that may have been lost. Also, they had no money, right? Well, that's very inconsistent. They say in Star Trek First Contact that they don't have money.

But then there's the Ferengi that have latinum. And there's definitely lots of episodes where Captain Sisko trades gold press latinum for different goods or whatever. And so, yeah, it's very weird. Maybe the Federation is more like China, where they're kind of like a little bit communist inside and a little bit capitalist outside. So you mentioned that, I forget what you said, hard things are easy now. Would you back somebody who is building like a Rails CRM or something? Our interest,

my particular interest and Root Venture's interest is in, do you think there's always going to be a top of that waterline? There's going to always be things in software that are hard. Software is getting more capable as hardware is getting more capable. Algorithms that used to be on academic papers are now getting used in

production. Things like neural networks, right, which we tend to think of as new, but that stuff was written on paper like 40 years ago. So it's certainly, I mean, who knows about this like Google quantum supremacy thing, right? I think there will always be hard software to be written. And that's what I think we're mainly interested in investing in. Which is just, you know, it's not to say that I'm bearish on everything else. It's just to say that I think we believe we have an advantage in assessing that stuff and knowing people in our network that are always going to be chasing that really challenging technical problem.

So what's the... It's not PayPal now, I guess. What's the hard problems now? Ah, yes, totally. So I think that... I mean, it's a generic answer, but certainly AIML. These are all very challenging, cutting-edge technologies. And even when you see things like TensorFlow and Keras that are making building neural networks a lot easier, at least mechanically. And this is, again, an area where the devil's in the details. Where's your data coming from? How do you clean it? Can you run these things in production and collect the data quickly and retrain and deploy the models? So I think there's that.

And then I think there's a lot of algorithms that are really early that are promising. Like I mentioned, machine translation. I think that reinforcement learning is another area that technology is hard. So again, as a venture capitalist, at least with my thesis area, I'm kind of looking at the technology. And I look at the technology first. I expect that businesses will always take the business. We'll think of the business first.

I'm not interested in investing in science, but I'm interested in investing in businesses that can only happen because of these very recent and new and powerful technological innovation. So I don't know investing in businesses that can only happen because of these very recent and new and powerful technological innovation. So I don't know what those businesses are. If I did, I'd probably quit and start the next $10 billion company. But what I'm hoping is that I'll recognize it when I see it, get hit by the lucky truck, and get to work with an awesome founder that's figured out something that's both a great business using a great technology.

So also interested in software for embedded systems. I think if you follow the trend of hardware development, you can often predict where the software is going to go, I believe. I think you and I spoke one-on-one about Rust being an area I'm interested in. I think that that's one area where Rust will really excel. When you need direct memory management and compiled software, and Rust is... Again, many people might disagree with me, but let's say I feel more productive in Rust than I do in C++. And I think many people agree.

Yeah, Rust. So it's interesting. So Rust is trendy in a way that probably C++ never has been. I know that's true. I think C++ was all the hot rage in the 80s. I mean, it was just like very exciting overseas. Yeah, maybe in my tenure as a developer. But I was just thinking about, okay, developers were very, I mean, sometimes we follow trends more than we think we might admit to it, right?

So Rust becomes hot. And then all these people who were previously, I don't know what they were, like Java programmers or Ruby programmers. They just want to do Rust things, right? And then the problems that are well suited to Rust are probably a bit different. So if that means we'll have like, if there will be successful businesses created merely because of a trend where this language looks cool and then people go to it and then... Yeah, well, I have a couple of thoughts on that. I think one is, I mean, every programming language is turning complete.

So in theory, they could do everything. So I do think that some... I mean, Ruby got deployed doing things Ruby never should have done. And that is also the case with Ruby. But that isn't to say that it was always stupid. I think that if you were prioritizing productivity on your main web app, then your team is sort of proficient in Ruby. So you built some ETLs in Ruby, which for say large scale data is absolutely stupid. But for maybe smaller scale data is totally fine. And it has the advantage of everyone on your team knows

how to use them. So at Teespring, before we were at really massive scale, and we did have ETLs in Ruby, but the nice thing was that software developers on the product team could really understand if their data was making it into the data warehouse because they knew Ruby. So I don't think it's necessarily bad that sometimes languages get sort of stretched into areas where they're not the optimum language. And one reason I'm interested in Rust is I do think today it does something that

unlocks a lot of business potential. So I'm really interested in WebAssembly for a business reason. So I think it's another business model that a lot of developer tools, or even really just any enterprise service, a very common business model is the free tier to sort of get you

on board, getting you comfortable with the product, getting you excited about what it can do. And then there needs to be a way that, yeah, again, if the company is hosting your product, they need to start charging you money. So if you think of like Heroku, where the free tier, free dyno thing was like such a huge engine for their growth. But it's a loss leader. They're paying money for everybody on the free dyno. And it's probably a big cost driver for them. But if you think about it,

there are some products where maybe you can... Maybe some of the computation that you're doing can actually be local to the device. And then as soon as you need more compute, you would go in the cloud and then the company will start charging you. And it makes sense. I think no one would sort of object to the idea that,

hey, this company has to charge me money now because they're hosting my code. So WebAssembly is a pretty good way to do this. Because I think more and more, the browser is the place where people are running their day-to-day work and certainly running things in JavaScript. TensorFlow.js is very, very cool. Without WebAssembly, it's certainly not going to achieve the full utilization of your laptop CPU. We run around with these MacBook Pros that have multiple cores and we're hardly using them

if we're running TensorFlow.js in the browser. So Rust is great at compiling to WebAssembly. And I think more importantly, it's really tracking as the WebAssembly spec is moving very, very quickly. And Rust is tightly integrated. Certainly other languages will do this. If I'm right and this ends up being important, every language will do it. But today, certainly it needs to be a compiled

language. Actually, that may not be true. I don't know if you know the answer to that off the top of your head. Well, it would have to compile to WebAssembly, I guess. Right? Right. So I suppose you could take an interpreted language and compile it to WebAssembly, I guess, right? Right. Yeah. So I suppose you could take an interpreted language and compile it to WebAssembly, but that sounds like a harder compiler's programming language problem than taking something that already compiles and then just picking a new

compiled target. I think so. Yeah. But yeah, so I think that's one really great advantage. And then the other thing, like I mentioned, is in some kinds of devices, the memory management really matters. Yeah. Same robotics, right? You can't afford, you're running this OODA loop, right? This sort of observe, orient, decide, act loop. And you have a garbage collection event in the middle of that. And our robot, there used to be this joke that was like, yeah, if the robots ever take over, tell your kids to run during the garbage collection cycle. That's awesome. I mean, I guess that gets at what I was thinking was,

okay, so I'm interested in Rust. I think it's pretty cool. And then they're like, oh, Rust is for embedded. So then I just start working in embedded, which I never had an interest in before, just because I want to play around with Rust. So then all of a sudden you have more interesting, just embedded stuff happening.

I don't know. Totally. I think, like you said, programming languages are often sort of brand driven and trend driven. And to the extent that a language can get someone interested in a new domain, I think that's super cool. I was thinking about quadcopters. I guess my understanding is like the reason they took off was because when cell phones started having gyroscopes, all of a sudden gyroscopes became super, super cheap.

Yeah, I think what's the phrase some people use is the peace dividends of the smartphone wars. So the smartphone wars are more or less at peace now, let's say relative to what they were 10 years ago. And yeah, like the scale of these things, like the scale of these things has made the cost go way down. And also just the investment in innovation. And that has unlocked a ton of stuff, like definitely in robotics. So another area that we think is like future looking and kind of what's next on the frontier of technology is what are the peace dividends of the self-driving car wars? We're seeing LIDAR, which is a big component of self-driving cars. That stuff is getting lighter weight, lower power energy. Now, there are companies that are trying to get to the $100,

$50 LIDAR future, and we're nowhere near that right now. But if they're successful, can you imagine the number of applications of LIDAR in the world? I wonder about a data business. So if LIDAR is cheap, could I make a company where I just give it away? I pay people to put a LIDAR thing on their car and then I just try to sell all that data. Yeah, I mean, that fits a lot of the models and stuff that we see today. I think that when hardware can be really cheap and you can treat it as a loss leader,

certainly that's the case with Particle. I think six years ago when they started, their particular chip and cost they were able to bring it down to was awesome. It was a competitive advantage and it let them give it away as developer kits for free and it got people hacking like you're saying like basically the availability of cheap hardware also you know arduino certainly outside of our own book arduino did this as well in raspberry pi you give this stuff away for free and then people can build

a business where the hardware is essentially free and what you're paying for is the software. So Particle really moved in the direction, as I mentioned, of now that the hardware is commodity, we don't even think of it as a hardware business anymore. And could you imagine something like that happening with LiDAR? I mean, wow, that feels like very far in the future that LiDAR would be close to free. But it could happen.

And certainly there's other technologies that will move in that direction. So I had a guest on the show, James Couple. And so he had this interesting idea. Basically, he said like there is just like so much old code out there, right? What was his example? Like coal factories that we depend on for infrastructure and they're just running all these piles of old code and nobody knows how it works and it's a huge problem.

And he said the most valuable business that could exist would be something that could take something like the Y2K problem, but solve it en masse, right? Certainly, you look at Flexport, right? We're not an investor in Flexport. But Flexport is a very, very successful example of a company that looked at old industry and brought good software to it. The founder of Flexport had two import export businesses before that that's not to say i mean he wasn't operating ships or you know but he understood supply chain and he understood the importance of um palletization which is kind

of like one of the key things behind flexport oh cool it's like shipping containers is what we're talking about yeah yeah flexport basically makes it easy to ship things cheaply overseas like say for example from china to the u.s and like one way they i can't claim to understand Containers is what we're talking about? Yeah, yeah. Flexport basically makes it easy to ship things cheaply overseas, like say, for example, from China to the US. And one way they... I can't claim to understand their business deeply. But as far as I can tell, one way they get really great cost efficiency is solving the sphere packing problem for these pallets. Most of modern shipping is really built around containers and pallets. Sorry, I guess I should say it's actually containerization rather than palletization.

So you have pallets that fill containers and the containers themselves are modular and they're quick to load and unload. But if you have a small amount of stuff and you want to stick it in a container, it's relatively fixed cost. But if you're able to kind of spear pack it, then you can get a lot of economic efficiency.

So yeah, I mean, containerization was a thing in shipping before it was in software. You can see innovation going the other way. That's interesting, though. Yeah, it's like an algorithm, sort of an algorithmic perspective on freight, I guess. It's very smart. I mean, they're worth many billions of dollars in their last valuation. I think definitely an overlooked unicorn. I think when people are kind of upset

at looking at WeWork and what's happened there and sort of decry the sort of looking at the trying to predict the decline of Silicon Valley, I think there is actually a really amazing set of companies that are going to be coming out and IPOing in the next couple of years, you know, market willing, and that will hopefully reverse people's opinions about this. I thought maybe a good place to end might be like, what business doesn't exist, right? In your area of specialty, what's not there? Yeah. Again, with the caveat that if I knew something very specifically, maybe that would be what I'm doing. I guess one business I would love to see, but I think would be so hard, is how can you generate...

We're seeing sort of AI influencers. I'm trying to remember the name of her, but there's an Instagram influencer who's completely digital. Are there ways to make AI-generated sort of actors? And maybe they don't star in the film. Maybe that's a little too sci-fi, but are they extras? Can you generate special effects? These kinds of things.

Movies are so expensive, and it's such a huge industry. And maybe the only way to do this is to verticalize and make a studio that's doing this. Or maybe the way to do it is to build the next generation Maya or 3D Studio Max where animators are using insanely powerful AI and they don't even really realize

it. Like Photoshop. Photoshop actually does some really incredible stuff under the hood. They've always done really interesting algorithms and now they're getting into really cool computer vision image processing type stuff that you don't even think about. Even the iPhone photo app is stuff like this. So the business

you'd like to see is replace Hollywood. Just... Well, I guess what I would love to see is the artists and creators really being able to focus on writing great stories and great characters and scripts and content. And certainly, I think actors have sort of more or less like a theater minor. So I don't want to get rid of actors. But I think that there are a lot of ways that you can reduce.

You think about how much money it takes to make a blockbuster movie. A lot of it's marketing, but a lot of it is actually making the movie. And I believe there's more good ideas out there for scripts and movies and stories that are coming from all kinds of amazingly talented artists. There's way more of those than there are blockbuster movies. I do think that

the role of an investor isn't to be the ideas person that drives the company with puppet strings. I think that the model that I'm trying to pursue is to really have a network of awesome founders and hopefully be good at picking the ones that will be successful and the ones that I want to work with and be supportive. So I don't expect to generate the next great idea. But maybe that random rant that I just went on

sparked something in somebody's mind and says, generate the next great idea. But maybe that random rant that I just went on sparks something in somebody's mind and says, well, I don't want to do that. But there's something kind of similar to it that I've been thinking about. Oh, that's great. So if that person's listening, how do they reach out to you? Oh, yeah. Lee at root.vc. Or on Twitter, I'm T-E-R-R-O-N-K. That's Taryn K, which is a handle I've had since I was 15. It's like my AOL handle. The story is relatively embarrassing, so I won't go into that. Thanks so much. This was great. Yeah, it's been a lot of fun, Lee. Yeah. If I'm ever in the area,

let's get together or something. Yeah. Yeah. Please hit me up. I'd love to take you out for lunch or something. All right. Take care. All right. Bye. I hope you enjoyed the interview. Brian Cantrell introduced me to Lee and I've gotten a chance to talk to him a couple of times now. And it's always been super insightful. If you have some crazy idea for a VC-backed business,

maybe reach out to Lee or, you know, if you know me, hit me up, maybe I can introduce you guys. Thank you for listening to the podcast this far. I'd like to take some time to thank some people who have mentioned the podcast online. So Pierre Angelo, who did a talk inspired by our Ray Tracer episode. Very cool, sir. I'm looking forward to watching your talk.

I haven't seen it yet, but I'm excited. Spinning Slate mentioned the podcast on Hacker News. Thank you so much. J.S. Toja, J.J. Molina talking about the podcast on Lobsters. Thank you, Platts on Reddit and everyone on Reddit, our Haskell and our programming for taking time to discuss the podcast. This podcasting stuff has been a fun journey for me. Sometimes it takes a lot of time, but it always makes me feel like I'm doing something good when I see people reacting to the content I put out there. So

thank you, everybody. Also, thank Bob Thurialt and Brandon Brown for helping out with the Slack community for this podcast. It's not the most active Slack community in the world, but it's got to be one of the friendliest around. They take time to welcome everybody who joins, and there's a lot of interesting discussions that happen there. So if you haven't checked it out, maybe take a chance and join until next time. Thank you so much for listening.