Introduction

Patrick: [00:00:46] Our guests this week are Kevin Systrom and Mike Krieger, the co-founders of Instagram. I met Kevin and Mike a few months ago over a shared interest in business and investing. I found them both to be extremely good people who have a rare talent for finding and solving interesting problems. Indeed, problem-solving and jobs-to-be-done is a big part of our conversation. I realized walking into the podcast that Kevin and Mike also have a rare set of experiences, having both built and sold an extremely successful product from scratch, but then also operated and scaled inside of one of the largest businesses in the world. This means they have unique knowledge to offer, just about anybody interested in business and in products. We dig into all those lessons here. I'm working on hosting more founders and CEOs on this podcast, and can't think of a better pair to show you why I want to do so. Please enjoy my conversation with Kevin and Mike.Current Areas of Learning

So, gentlemen, we're going to talk about a lot of business-related stuff. We'll talk about Instagram for sure, because I'm fascinated by what you learned, both building the business from scratch and then also operating inside of a massive organization. But, I always like to begin with whatever has your interest and attention right now. So, you've had some time since leaving Instagram. The last time we were here, we were talking about all sorts of interesting companies that you had seen, and data science-related projects you were working on. And, you just strike me as two incredibly curious and interested guys. So let's just begin there with, what have you been tooling away at, playing with, learning about, since having a little bit more free time?

Mike: [00:02:13] Yeah, I think for me, it's been getting back into code, which has been very gratifying, a lot of fun. Last year and a half of Instagram, I was coding less and less, because I was still relatively in the technical details, but it's irresponsible to try to code too much at that level of company size. But, I think one thing that's been interesting for me recently is seeing how much data has become, going to say like, a first-class citizen in programming. And what I mean by that is, I remember when we were first spinning up, even some Instagram data pipelines. You have a bunch of ad-hoc scripts, they run every night hopefully, sometimes they fail. We can tell the story of how our first analytics pipeline ran, because it was seriously duct tape, a harrowing disaster.

And, going from that to the point where you have things like TensorFlow be readily available. I've been playing with some data pipeline stuff, where it's just much more declarative, much more thought through, much more built for a whole team working together on data, rather than that one data analyst in the corner. That's probably the thing that has most changed from when we started Instagram, where the new hotness then was web frameworks that let you spin up a site fairly quickly, like Django and Rails, to now I think, the same opportunity exists in data and pipelines.

Kevin: [00:03:18] I feel like when we started, we were beginners. And by the end, we went and mastered the game. You can never quite master the game, but by the end, we had figured out our game, and we had a great run and we created something that a lot of people love in the world. Over a billion people use it every month. We were really proud of the team we had created. It was over a thousand people when we left across how many offices?

Mike: [00:03:40] Three main ones and probably six or seven other ones. Yeah.

Kevin: [00:03:42] Exactly. So, we had all these amazing people around the world. And to be 35, and I guess Mike is 33, pressing the reset button and starting again, is this awesome, but harrowing adventure of becoming a beginner again.

So, I did what I think a lot of people would do in our situation, which is you take a few months off and you clear your head and you tell yourself you're going to go do whatever interests you. So Mike and I, didn't actually really talk that much for those first three months. I went skiing, I was in the mountains and I was like, "I'm going to get away from it all," and I grew out my beard. And by day six of skiing, I was like, "What am I doing out here?" It was a lot of fun for six days and then it was just cold.

So, I'm out there in mountains. We had moved out there. We were going to be there for three months or so. And, I decided I would open up some books. And, I would start asking myself what was interesting. And, I remember Mike had said, "He was taking a course at Stanford." I guess you were auditing or taking the online version of the Machine Learning course. And I was like, "Ah, Machine Learning, I've heard a lot about that." "I should probably look into that." It helped us a lot at Instagram. It was phenomenally useful for making the site what it is today, but I never really understood the details.

It just caught my eye and I sat down and started working on it. And then, Amazon shipment, after Amazon shipment of books coming to the house. I was reading all this stuff and I texted Mike and I was like, "Mike, this AI stuff's really cool." So I mean, if you look at the last six months or so, I guess, we've been gone about a year now, most of that time has been understanding the math behind modern machine learning and then implementing experiments and trying it ourselves. But, Mike was my inspiration there.

Patrick: [00:05:22] What is your early opinion on the places that what you're learning can be applied most usefully, whether that's in a business context or some other context?

Kevin: [00:05:30] There's an engineer at Instagram who is our machine learning guru. And, he had this amazing phrase which is, "Machine learning, it's just math." And, I love that. By the way, it is just math. It's not magic, it can't tell the future. Well, in some cases, it might be able to tell the future, but we'll get to that. But, it's just math. And, if you just understand that it's just math, then you understand that you can't hope to just apply math to problems, and have it solve those problems overnight. I mean, of course, it's world-changing, and of course, you can get to better outcomes. At the end of the day, it's just math, so you have to have a healthy dose of humility when you approach new problems. I mean, Mike has been experimenting as well.

Mike: [00:06:10] Yeah. And I think for me, what stands out is you need to have a reasonable hypothesis about why it is that throwing a bunch of data into some kind of prediction system is going to yield a useful prediction.

Patrick: [00:06:19] Yeah.

Mike: [00:06:20] Which sounds like maybe a trite phrase, but you see it. I mean, I saw it at Instagram too. When you are building a machine learning system, the way that you try to increase your accuracy is add features. So for Instagram might be, where was the photo taken? How many likes does it have? And, you start with the obvious ones. And then all of a sudden, you find your team is mostly doing feature engineering on this long, long tail of stuff. Like, which direction was the camera pointed? I'm making some of these up. What was the weather that day?

And, they might actually have some predictive power in the short-term, but you're probably over-fitting the data. And then, you get really excited and it leads to all sorts of weird effects downstream because you haven't actually created one. So, I guess to directly answer your question, I think it's places where the quantity and quality of data has gotten to the point where it actually is useful to throw the, it's just math approach on it and try to yield a generally useful prediction, in a way that might be different than having hundreds of analysts try to extract some meaning in the same data. It's not, I think, useful in the magic sense of, "Oh, all these patterns that we never knew were there." They probably weren't there to begin with.

Machine Learning

Patrick: [00:07:18] Do you have a most interesting experience that you've had thus far diving back into data and code, maybe with machine learning specifically, something fun that you played with or something that surprised you?

Mike: [00:07:27] I think the feeling that you have when looking at some of the computer vision stuff, and its applicability beyond the set it's trained on. And, what do I mean by that is, there's always this interesting question of, are machine learning systems trying to emulate some kind of physiological thing that's happening in the brain? You hear the phrase, ‘Neural networks’ a lot. And actually, I think the neurological side or the neuroscience and the computer science, I'd probably diverge in the 70s and haven't really reconverged until there's been some interesting conversations recently. But, one thing that I found really exciting was you train a bunch of computer vision algorithms on a set of images for one particular problem. And weirdly, it works really well for other sets of problems, sometimes not even vision-related. So, I think there is this convergent that might happen now in the next five, ten years where we say, "Hey, all right, it is just math, but it is looking at things in a way that could have some parallels to how we see it."

And, what's beautiful about the human brain is how plastic it is, and how much we're able to say, "Yeah, I know how to see in this domain, but now I'm interested in applying that somewhere else." So, that applicability was the moment I was like, "Huh, there might be something more to this than just a particular domain vertical sort of thing." And, it kind of comes through when you start looking at, all right, how can we analyze time series? Is that sort of analyzing an image, and how is it like analyzing an image or a sound wave, and thinking about the applicability across domains, even for a given model.

Patrick: [00:08:42] ML seems like one of the many incredible new tools for leverage in the emerging business person's toolkit. One of the things that I was fascinated about with Instagram is, when you sold it, it was I think to 12 people. It was a small team, that even at that size had achieved serious scale. And, that was almost 10 years ago when you guys started. I know you've also looked at some startup founders, maybe even made some angel investments. What is your take on the most significant differences between the stack that you guys dealt with when you started, and what you're seeing either hands-on personally or from the startup founders that you've talked to recently?

Kevin: [00:09:15] I'll go actually all the way back to my internship in college. It was at a company called Odeo, which not many people know. But, you know the company it became, which is Twitter. And, we were maybe eight people when I joined Odeo. And, I remember part of my job as an intern was to go to the colocation facility in San Francisco. And, I remember the general area of town. I don't remember the building, but we would walk in and you'd have to wear these ear covers, because it was really loud and you'd bring your jacket. And, we had to like go reboot the servers and we have eight on a rack. And anyway, when we started Instagram, I guess, I just thought that's how you did things. So, Mike will laugh at this. Because at the beginning, we both thought it was normal. We bought a server in a colo facility.

And by the way, this was interesting because we had never seen the server. Whereas at least in the Odeo days, we knew our servers. When Instagram launched, the flood of people came in and everyone wanted to use the service and we couldn't keep up with the demands. So, we had to figure out, what are we going to do? And, I like to say, "The only time Mike and I have ever really gotten into an argument was whether or not we needed another server." I didn't want another server, because I was penny-pinching. We were a startup and who knew what would happen? Mike was like, "We have to get another server." Mike was right. And, we had to get a few more after that, it turns out. But my point in telling this story is, it used to be that you had to have these physical things.

And then, we were at a VC's office and they were like, "Well, why don't you just use AWS?" And we're like, "What's AWS?" And the second week on AWS, it was like, "Wow, we just scale efficiently." And by the way, the things we were scaling efficiently were all libraries that had been open-sourced by many companies before we had started. I remember in the Odeo days, what did you use as a caching layer, maybe you had to roll your own thing. And then it was memcached for us, and then it became this other thing, and this other thing. And, just the number of open source libraries that we could use for free to start a business, AWS cost something, but it was effectively scale for free. You could build a business for what? I think we spent 65k total by launch day, when we launched Instagram.

And, that was a $500,000 raised, maybe we should have raised less, I don't know. Shoulda, coulda, woulda, right. But now you look at it, and you realize that... I bought a book back in the day. And, I think it was called AI in Python or something. It was an O'Reilly book, but it was back in the day. And, it had an entire chapter on neural nets, but it was like how to create them from scratch. There were no libraries. You couldn't plug into anything. There were no server types that you just spin up with everything installed. And now, we're in the place where actually you can do that. Scikit-learn exists. And, not to say that, that's production-ready, but it exists and you can use it.

XGBoost, we were talking about the other day. It exists, you can use it. TensorFlow, all the flavors around TensorFlow exist. Now, you're in this place where for zero marginal cost, you too can have these things at your fingertips. And then it becomes, I think, not about access, but about the knowledge of how to use these super tools. Because man, people misuse them. They throw them at problems they shouldn't. They break all sorts of overfitting rules. And, that's actually where I see the opportunity going forward, is who knows how to use this stuff. Not just who has access.

Patrick: [00:12:24] Yeah.

Mike: [00:12:25] We had this phrase at Instagram, which we used religiously, which is, "Do the simple thing first." Kind of a spoke to the approach, both from a business side, but also from a technology side. And at the time that meant, don't reinvent your own web framework, use Django. Pick Python, because it's a simple language, we can get it up and going. But, we had a conversation recently with this private equity person and he said, "A lot of their downsizing..." What they do is they take companies that have millions of lines of code, and they're like, "Yeah, this is actually just eight Amazon services we can compose, and put 3000 lines of glue around and done." And, that's it.

We had this phrase at Instagram, which we used religiously, which is, "Do the simple thing first." Kind of a spoke to the approach, both from a business side, but also from a technology side

And, this interesting question around this question of not invented here, it's even less of what you're inventing is, core technology at this point. Because, they're great open source, or Amazon, or Microsoft provided components. Now, it's what you do with it, and now you differentiate on what data you can get in, what data you can provide or what problem you're solving for people, which I think it's actually exciting. Because, you sit around and you're like, "Is the world well-served by hundreds of companies all reinventing the same exact technology?" No. It's like, "Let's do something interesting with it."

Patrick: [00:13:22] Yeah. What is in common between the most interesting younger founders that you've talked to, in terms of the markers of someone that might use this stuff responsibly? Can you sort out any difference between those that you think are going to do a good job, versus those just copying and using powerful tools just to use them?

Mike: [00:13:39] I think the founders who get interested in what their data actually looks like underneath. It's like an underappreciated thing because you're like, "Oh, I have a million examples of this, I'm going to throw it in the system." And, I like the founders that were like... Oh yeah, and I spent a day just looking at the examples and they're like, "Oh, we realized half the images came in backwards or upside down." I'm like, "That would have been a problem," but you wouldn't have seen it if you actually weren't diving in. So, I think what we always try to do is bridge the things that were really unscalable, handling support tickets ourselves at the beginning, or still coding for a really long time, even though the team was scaling. And then obviously, trying to build a company for scale. And, I think that's a balance you need to strike as an early-stage founder too, which is, don't get so enamored by building something super scalable, and just the millions of things you could throw in. Just have a deep understanding of the problem you're solving and what it actually looks like in the details.

Problem Solving

Patrick: [00:14:25] There's this really interesting thrust to startups that are general-purpose tools, productivity tools, for lack of a better term, that can do a million different things, but they don't seem to do any one thing that you actually need. Which is a good excuse to talk about your guys' interest in this jobs-to-be-done framework. I know this was an important part of how you thought about things at Instagram. I'd love to hear just kind of from the beginning, why you find this idea compelling? And, how you think other business people can use it to their advantage?

Kevin: [00:14:50] Well, from the beginning, We always said, "You have to solve a problem." So, if you're going to build a product, the only reason it should exist in the world is because it solves someone's problem. And when we started Instagram, we listed out what the top three problems were with mobile photos at the time. So number one, was that most people felt their photos were terrible. Because by the way, they were. They were with a blurry lens. It was on an old camera. Or not an old camera, but the sensor was terrible. It was just at the beginning of digital photography. The second was that people wanted to be able to share their photos everywhere. They were fragmented between a bunch of different services. So, they were on Twitter, they were on Facebook. They were on, at the time, it was Posterous and Foursquare. You were across all these services and you just wanted to take one photo in one place.

And, the last one was speed. It was so slow to not only take a photo, but also get it onto one of these services. Because, I don't remember what transfer speeds were on average, but it felt like it took a couple minutes often. And, this was when you... I can't remember if at the beginning of Instagram, you were allowed to background apps. Because remember, you couldn't actually background in an app, and then it couldn't upload the photo unless you were staring at it. Exactly. So, all the things we take for granted now, in terms of solved problems, were unsolved problems there. So, we wrote those three down and we said, "We were going to crush each of them." So, we made filters as good as we possibly could, to make up for the fact that the camera wasn't that good. We made a one-click sharing feature that allowed you to tap a bunch of services and auth with them and share all at once. That turned out to be great for growth, by the way.

And then, the last one we were most proud of was, as you were captioning your photo, we would actually upload the photo in the background, regardless of whether or not you completed, because... Okay, we'll upload it and if you say, "Cancel," we'll throw it away. But if you say, "Hey, here's my whole caption that I spent five minutes figuring out what to say to sound clever." It turns out, all that time your photo has been uploading in the background. So when you click done, within a fraction of a second, it says done. And you're like, "Wait, but I thought I had to go upload it now." And, that little, slight of hand meant that Instagram felt enormously fast compared to all the other ones. So, by focusing on those three problems, I think... And by the way, a bit of luck and timing, Instagram was at the right place at the right time and it solved core problems around something people clearly wanted to do.

So later on in Instagram... So, we always had that phrase which is, "Solve a problem." I came across Clay Christiansen's book, Competing Against Luck. Actually, I came across one of his writings in a different form where he mentions this, but then found out he wrote a book about it. So, reading about Jobs to Be Done Theory was one of the most transformative experiences for us at Instagram. Because, we realized that's what we had meant when we said, "Solve a problem." So, jobs-to-be-done is effectively, people need this product in their lives to solve a problem for them. So, what are they hiring it to do? It sounds a little weird to hire a product because usually you hire people. But, imagine all the things in your life or employees, instead of just things. What do you want them to do in your life for you?

And, if you can explicitly state the problems that it solves, not just the functional ones, but the social ones, the emotional ones, then you understand why people have this product. And, when we started to pivot around that notion saying, "Each of our products needs to be very clear, what problem it's solving in someone's life?" "Why is someone adopting this thing?" Then you started to see this enormous amount of success. I mean, the number of products we launched and the success of things like stories, eventually on the Instagram platform was all because we had adopted this framework.

Mike: [00:18:24] We found jobs-to-be-done has this interesting application, not just in the products you ship, but how you think about your different teams. So, I like to think we ran our infrastructure teams like a product company would. So, I had all of our infrastructure teams actually write up what job they were doing for the other teams, because you can often end up with these teams that go off into sort of gold-plating land or build technology nobody needs. And then a year later, they're complaining because nobody adopted their technology, when that's not everybody else's job, other than them to build actually useful things for the company.

And at first, people kind of grumble. They're like, "Why are we writing these strategy documents and jobs-to-be-done for a team that's focusing on machine learning tools." And then, a year later they're like, "Oh cool." I know that my job, our job as a team is to build the fastest way of deploying models for other Instagram teams, so let's throw out the other half of the projects that wouldn't solve that job, and let's do the stuff that actually matters.

Patrick: [00:19:14] I'm really curious how your ability to identify problems or jobs evolved through Instagram. So one of the things, for example, that we've noticed in Instagram is very much a platform, people do with it, unexpected things. And so, I'm curious whether more of the jobs were discovered by unusual ways that the product was being used, that you didn't intend for, or you coming up with a thesis and then identifying it beforehand, or afterward is kind of what I'm interested in.

Kevin: [00:19:39] It’s kind of both. By the way, there's no secret here. You don't wake up one day and have this vision, you're on top of the mountain and you say, "Team, we're going this direction and no one's ever seen this thing before, and it's going to be amazing." That doesn't happen. I'm sorry. The romance is out of the story. Instead, what happens is you note these trends that happen in the world, that you see it popping up and maybe even in your data. One of my favorite stories is about this tool called Fiddler internally, where effectively they said, "Hey, I want to know how people are using this product." So for public posts, whether it's on stories or on regular Instagram and feed, I want to select maybe it's men age 13 to 21 in the U.S., how are they using it? And, it just samples a bunch of their posts. And, it shows it to you in a grid.

Okay, females in France who are 24 to 37, I'm making these ages up, let's see that. So, you're seeing a sampling of the world and a specific demo. And, what'd you very quickly see is that although Instagram's primitives are very basic and can be used in many different ways. It turns out groups of people use it in very specific ways. So, when we were designing stories and we asked ourselves, okay, what tools might be super useful? It turns out that all those teen boys loved taking to stories and asking questions of their friends. So, they would color out the background as a solid color. They wouldn't even take a photo. And they'd ask a question. One of our engineers had this insight that he was going to build a polling sticker for Instagram Stories. And, I still remember seeing that and going, "Ah, I don't know, am I going to use that?

And they're like, "Trust me, the teens are going to love it." And they launched this thing. Day one, it was a hit and it just skyrocketed. And usage of stories skyrocketed. That's a simple story of a developed product. How you can look into the data, see the trends, and say, "Hey, how are people hacking our product? And, how can we meet them in the middle to make that way easier? Hashtags on Instagram, people were using hashtags way before we allowed people to actually-

Patrick: [00:21:43] Track them.

Kevin: [00:21:43] ... yeah, track them or click them. Or in fact, that was one of the harder technical projects early on, because we were trying to figure out how to make links. We were really early in our technical careers back then. But, there were probably other examples too.

Mike: [00:21:55] We had one where we found... We were just looking for anomalies in our accounts, which is important when you're fighting spam and just making sure that your community is healthy. We found one account that had deleted thousands of images. Clearly, they were posting bad stuff and then deleting it. Let's go take it on the account. Like, "All right, we should look at what they're posting first." And, we realized it was an account in Indonesia and it was a store, sort of independent business. And, they would post an item for sale, and then when they sold it, they would delete it. So of course, they would delete a bunch of things. This was 2013, years before we had any inkling of a commerce product inside Instagram. And you're like, "Oh, these people are, quote, unquote, misusing the product, but they're using it for this very real use case, which is they want to transact."

But, the comments would say, "How much is it?" And, their response would be, "It's 50 and WhatsApp me." And, that's how they would get payment. And, that whole completely external Instagram infrastructure was getting created by people who found a need, because they had the audience there. So they can ask, "All right, is there a job-to-be-done, that's adjacent to the jobs we already do?" Because, I think that's important. I feel like companies fail when they are getting hired for one job, but they're jealous of another job, they wish they got hired for, that's way out in left field. And, they launch a product and they wonder why, "Gosh, we already have all these people using the product." "Why don't they just use it for this other thing?" I think you always need to build in these adjacencies. So, people were already getting inspired on Instagram and seeing interesting products, now you can also buy them. That feels like a pretty normal link. Whereas, if that behavior didn't exist on Instagram and you launched a commerce product, it would probably flop. Because there's no existing behavior that you can divert or build upon.

Patrick: [00:23:19] Did that change a lot over the years where... I mean, it sounds like there's just tons of examples of observing behavior, and realizing there's a job happening that you can amplify and provide a better toolkit for. Even when that was happening, were you also just coming up with the classic jobs thing, no one asked for an iPhone? You have to sort of anticipate the needs of your customers, versus just observe and accommodate them.

Kevin: [00:23:41] I mean people were asking for an iPhone. That's the crazy part. If you look at it, the Nokia smartphones with little games and everything. The terrible, what was it called-.

Patrick: [00:23:50] Wap.

Kevin: [00:23:51] Yeah, thanks Wap. It's a specially formatted webpages for your mobile phone. They were. In fact they were screaming for an iPhone. And, you just had to be at a point where you could say, "Oh, interesting. Everyone wants to browse the web on their phone, that just makes sense, it just sucks when we do it right now. So what if we made it not suck? There is a brilliant idea."

And to their credit, I think everyone saw it. Everyone knew that apps were going to exist in some way, everyone knew that web browsing was going to exist, but not everyone knew how to do it. So, touch screens, giving up the keyboard. I mean, so many companies get in their way from solving a problem or being hired by consumers, because they just assume too much is going to stay the same. They assumed people would never use a phone without a physical keyboard. They assumed no one would ever use a mobile-only photo-sharing service, because, of course, you're going to want to upload from your camera that you hook up to your desktop with a cable. You spent so much money on it, why wouldn't you just stick with that thing?

The companies that question a fundamental assumption, during an enormous transition of technology, meaning, there's this wave of mobile that comes and no one quite knows how it's going to play out, but everyone knows that... I don't know, there's a camera attached, if you can take advantage of that wave and say, "Listen, there's an unmet need of sharing photos with your friends on these mobile devices. No one's touched it yet. We can do it in a slightly better way." I think that's where companies end up taking off.

Instagram's Product Principles

Patrick: [00:25:20] How early in Instagram's history where you planting... I don't know what to call them. Core values or principles around product?

Kevin: [00:25:28] Oh, from day one.

Patrick: [00:25:29] As a user of Instagram one thing that stands out is visual simplicity. I've seen some of the... Is it a WeChat in China? With seven million things going on. Now that works for them, it's a huge, huge product, but Instagram is extremely elegantly simple. So how early did you plant some of these principles and how did you arrive at the principles themselves?

Kevin: [00:25:46] That's funny. I don't remember the word principles ever coming up early on, but I remember Mike and I being very principled. We had, not arguments, because they weren't between us, but they'd be about, "Should we do something or not?" And we'd always come back to these few simple ones. It was, one, it had to be community first. So if we were going to do something to the product, was it to our benefit as founders or a company, or was it to the community's benefit? So community first was an enormous principle of ours that I think kept us focused on building stuff for people, again solving their problems.

The second was, keep it simple or do the simple thing first. We had just been bitten too many times when we tried to make things really complex and then we realized, "Wait, that wasn't the right thing to do. It took us many months and actually we could just do this simple solution and it's fine." And then, speed was a huge one too. I mean, early on we were so incredibly focused on making the app work really well on my terrible iPhone. Was it a 3G?

Mike: [00:26:49] 3G, yeah.

Kevin: [00:26:50] And Mike had the new version, he was fancy, I had the old one.

Mike: [00:26:53] No, he had a diseased 3G. I think-

Kevin: [00:26:55] Easy, easy.

Mike: [00:26:55] We know now that when the battery gets low the phone gets slower. So I think now I know what was happening with that phone. But it was slower than every other 3G in the world, but it was great, because Kevin would come in and be like, "Man, it still doesn't work well on this phone." And I'd be like, "Oh."

Kevin: [00:27:09] Mike wanted to move on and it was like, "Ey, we've got other projects to do man. I don't care that your feed's stuttering." And I was like, "No, we need to make this like glass." And the fun part about Mike, is he looks that... He might be frustrated for all of a half a second, but then he's like, "Oh great, this is a huge technical challenge." And then when you get to that point where you open up the phone on this 3G and it's like glass when you scroll, now on every better phone, it's the best experience ever. So anyway, you asked if we had principles from the beginning. Yes, and then some. Speed, it was craft, it was simplicity, it was being community first. What other ones do we have?

Mike: [00:27:43] We talked often about unlocking people's creativity as well. Building tools that let you feel like, "Oh, I wouldn't have shared this before, but now I actually feel like I captured something new and interesting and different." And that obviously evolved throughout the years. Because the nice thing about all of those, they're evergreen. What they meant evolved or how we explained it evolved, but they're kind of core to the product to this day, I think, actually.

Patrick: [00:28:02] Do you think that if you were to start another business today, that right away you would start with those same principles?

Mike: [00:28:08] I think the truth underlying that, for sure. The simple thing first, we still have just been noodling around and hacking on stuff. I still get caught. The other day I was like, "Man, I spent two hours with this really complicated solution. It turns out I could have used a one-liner from Matplotlib or something," it was like, "Ah." And then you're like... Again, it's almost like the more you learn, the more you have to remind yourself of that one.

I think the community first one when it gets at is... and I think we've seen this broadly. I don't think people get enough credit as consumers for kind of sniffing out the intentions or sort of the underlying ethos of the companies that they're using. I think they're getting even more attuned to it now. So I think saying community first and being community first, for us was, ultimately, people can tell who they're kind of using products from. They can tell what the ethos is. And I think people really could tell for a long time.

As me and Kev running Instagram and whether that meant we would post announcements to the Explore page, because that's the only way we could communicate with people about outages that we were going to do or improvements we're going to make or seeing us on public, meaning like when we're doing talks. I think it kind of cuts through. So the truth underneath community first is people can tell who's behind the company, so be authentic and talk to them as real people rather than as an abstract user base.

Kevin: [00:29:16] To answer your question about, would we do it again? I don't know any other way, that's the problem. It's like you meet athletes... I'm reading this book, this fascinating book, I think it's called Analytic Methods in Sports. Don't get me wrong, I'm not into sports at all, I don't love sports, and this was the one way I thought maybe if I apply math to it I'll get into sports. But it was talking about the different types of... You guys are both laughing. I love that. That's a very Kevin way of thinking about the world, by the way.

Patrick: [00:29:42] Hack friendship?

Mike: [00:29:42] Yeah exactly.

Kevin: [00:29:43] No, not hack friendship. But, find something to love in something. "Oh, the math, it's great." Anyway, it was talking about different types of... I don't know whether it was running backs or something. I don't even know what a running back is. Anyway, depending on your body type, are there different distributions for performance for those different groups?

And I was thinking to myself, there are sprinters in business, there are marathoners in business. I don't know if we're a sprinter or a marathoner or whatever or a high jumper, but whatever we are, I don't know how to be anything else. I'm never going to be good at any of those other things. I can try to be something else, but the values that we have internally are kind of like our DNA, and all I know is keeping things simple.

Maybe out of necessity, because when we started Instagram, I was the best in a lot of these things. We raised money for an HTML5 web app. Not an app. Why? Not because I thought HTML5 was the future, I didn't know how to make an app. So I had to keep it simple.

So, again, when you ask if we did it again, would we keep all those same principles? Absolutely, yes. But not because we choose to, but because we have to.

Patrick: [00:30:48] So the community thing is fascinating to me. How do you manage a community when the community, at the point that you left, is the world. It's just a slice of the entire world? So, I understand cultivating community. I think that's super important when you can sort of draw boundaries about what makes a community a community, but now you're dealing with the entire human populace and that remains an important thing. So how did you manage that challenge?

Mike: [00:31:09] I think realizing that.. We had the team for a while, it's called The Community Team, and we often wanted to call it The Communities Team, because ultimately that's what you get. You get this sort of amalgamation of a bunch of communities, and then I think the interesting sort of challenge becomes, how do you, at different points in time, go deep on one or the other and understand how they're using the product and whether there's something you could do to cultivate that?

So I mentioned commerce is one of them. Independent salespeople. I went back home to Brazil and I got to meet a bunch of entrepreneurs for whom Instagram was their storefront. That's an interesting community of people. Can you connect them to other people doing the same thing so they can share insights? Or a community might be geo-specific. So what we would do is send research teams out, and I would always send engineers with them, because I think having insights viewed firsthand was always interesting and they'd come back and say, "Hey, just spent a week talking to French teens. Here's what's interesting about that community and let's change the product in this way potentially, or let's make it work better in this way."

So you end up realizing that the joy of Instagram is that it could scale and not become just one huge living room with a billion people in it, but a whole collection of different spaces, and then it's about, "Can you let people transit through the interesting spaces that they might have for themselves and let that evolve over time too? Because you might pick up a new habit or a new interest and now it's time to evolve your usage of Instagram." So I think that's how it changed and one of the challenges internally was helping people evolve past our initial community, which is, people who were super interested in iPhone photography and that was sort of a new thing at the time. It was called iPhoneography, which nobody has ever called it that after 2010, because, of course, it's just photography, you just take it with your phone, to understanding that we were now dealing with a huge group of communities rather than just individual sort of initial user base.

Patrick: [00:32:47] What were the hardest or some of the hardest challenges in that community piece of the equation, through the entire experience? The hardest things you had to deal with?

Kevin: [00:32:56] I mean, the list goes on, because of a lot of what you just explained, which is you're not dealing with one community, you're dealing with many communities. So maybe I'll give two broad examples or two broad patterns. One is changing the app. So maybe the first example of this was expanding to Android, which, of course, everyone now is like, "Of course you're on Android. It would be crazy not to." But when we decided we were going to launch Android, there was almost a community of revolter's. Like, "No, we're iPhoneographers and we have these beautiful phones and you're going to ruin Instagram if you let those Android users on." Fun fact now I guess Mike's on... I'm looking over, he's got an iPhone. But for a while there you were an Android user.

Instagram could never get to over a billion people if it just focused solely on iPhone. So part of it is negotiating with your community in a way that isn't patronizing about a big change that's happening. We know we have to get to this place. We know at some point, yes, Britney Spears is going to be on our suggested user list, because it turns out a lot of people really care about Britney Spears or Kim K. It doesn't matter, very large pop figures.

I know you adopted this app initially because you wanted to just follow the best photographers in the world, but Instagram is larger than any of us now with our own viewpoints. It's this platform that should be universal and not just unique to our own wants and needs. It would be selfish as leaders if we kept it to just our thing, because what we want to do is build a platform, not just a little app. So those changes I think were hard.

The second is, I would call it the relativism issue, which is, in the U.S, nudity on public platforms is generally not accepted. You look at the app store terms, etc. Read a newspaper here, there's not nudity in it generally, or magazines generally, but in Europe, it's a little different, standards are different.

But it turns out, as a platform, you have to make one rule. And how do you deal with the fact that France might see it one way and the United States might see it another and you're one group? So explaining to people why certain things can be on the platform and certain things can't, that's really tough. Because, you're caught between different views on a subject. But when we say community first and we're that big, we don't mean the community runs the platform. Instead, it's, "Okay. We wouldn't be here if these billion people didn't use this app." You can either treat it as if you are a dictator in charge of the thing that they use, you have your views on policies, etc, or you can view it as, "I'm really thankful I have this job. I'm really thankful all these people showed up and for some reason they continue to show up and then some. How do I make it so that it's best for them?"

And I think that the companies that do well in the long run, understand that balance and really serve the community and say, "Hey, I'm going to do my absolute best, but it's going to be hard. I'm going to do my absolute best." And that continues to, I think, play out today on Instagram. It continues growing and people love it.

Mike: [00:35:57] You have to be occasionally okay, changing the product in a way that might actually sort of break the product in some way for some group of people. Memorably, for us, so Instagram Direct was a product we evolved many, many times, and one incarnation was the way it worked is you would post a photo, add people to that private photo and then people would have a conversation about it. Which it was okay for a first pass, but it turns out-

Kevin: [00:36:19] I think okay is very nice.

Mike: [00:36:19] Yeah, okay. It did not do well. It turns out that we eventually-

Kevin: [00:36:24] Did not meet our expectations.

Mike: [00:36:26] We fired that product and we evolved it to eventually to be like threads that were more permanent because it makes way more sense. You generally have some groups of people you keep talking to every messenger app has eventually evolved to that. So we're like, "Okay, everybody's going to think this new version is way better. That's easy." And it turns out that some people were using the fact that femoral groups get created to tell stories. There were storytelling groups online. It's very niche.

Kevin: [00:36:48] Wasn't it a role-playing?

Mike: [00:36:49] It was like a role-playing game thing. I forget-

Patrick: [00:36:52] LARPing or something?

Mike: [00:36:53] LARPing, totally.

Kevin: [00:36:54] Yeah. It was connected to that community.

Mike: [00:36:55] And they would post a photo and then they would all role play, which is not a use case that anybody on the team had, and it was very small. But I remember the first day we were getting bombarded while small, you can have a platform of billion people, if it's a small percentage of someone percent, it's still a lot of people. And the team was really upset they were like, "Man, all these people are really angry about this change. Can we find out why?" And they were like, "Oh, it's all these LARPers." I'm like, "All right." Well, you have to be okay sometimes changing the product in the way that might break some use case and just trying to understand, was it a bug or did you just make a decision and then be principled about it?

Kevin: [00:37:28] But Mike, agree or disagree, every major change that came with a ton of pain was either A, the best decision we ever made, or B, the worst decision we ever made.

Mike: [00:37:39] Exactly.

Patrick: [00:37:40] I guess that was an agree or agree and a multiple-choice in one. So go for it.

Mike: [00:37:43] I think what's interesting there is you can't treat every major disruptive change as being necessarily the right thing for the product, but if you automatically assume it will be the wrong thing for the product, you'll just be paralyzed and never actually make any of these changes. And we had this conversation with the team before we launched stories. So, sort of notably for us, we did a bunch of incremental A B tests with the product along the way, but for stories, we launched it to basically everybody in the world at the exact same time.

Patrick: [00:38:08] What year was that?

Mike: [00:38:09] 2016.

Kevin: [00:38:10] We had a small holdout and that was it. Maybe even not a holdout-

Mike: [00:38:13] We had like a half a percent of people didn't get it. Not because we thought it wasn't the right move, but we thought it would be at least useful to be able to measure what effect it had, at least, among some group of people, and we tried to keep that holdout very short. Like a couple of weeks and then they got the product.

And I remember the team being really freaked out they were like, "But what if it breaks Instagram? What if this..." We kind of had to look at the team and say, "Look, we can always turn it off. I mean, wouldn't look great. Kind of be embarrassing." But I think there's also this element of sort of conviction that you need to have in the major changes that you're making, that if you sort of tiptoe your way into it... Again, it goes back to the community first point. I think people can tell if, "Do you guys mean it? Are you guys really launching this? Or you..."

Kevin: [00:38:50] It's all like, "My friends have this, I don't."

Mike: [00:38:52] Yeah-

Kevin: [00:38:52] What's going on?

Mike: [00:38:52] It's all sorts of weird social dynamics if you don't go all in on it. So, it could have been the wrong decision and then it was one of the best changes we ever made to the product. But I liked that at Instagram, we tried not to sort of half launch things or make these half decisions and instead say, "Let's go for it and we'll deal with the consequences of figuring out on the other side, if it didn't work."

Patrick: [00:39:11] How extreme were some of the evolutions? So, maybe at the end, what percent of usage on a high defined usage like FaceTime or screen time was video and stories versus the original filtered photos?

Kevin: [00:39:22] Not sure I remember the exact time spent on all of them, but let me put it this way. Regular Instagram continued to grow, time spent per user is the way we measured it. So how many minutes per day, daily active, basically? That continues to grow over time and it was doing okay, and then we launched a couple of things. One was machine learning ranking for our feed. So it was like, okay, now you have too much in your feed. If you get bored after the third thing, because everyone's posting all the time, you turn off. So we're going to filter through and basically say, "What do we think you're going to enjoy the most." That had an enormous impact on time spent.

The other one actually, was our pivot towards focusing on your friends. So as Instagram grew, it was easier and easier to see stuff by, like I said, Britney Spears or Kim K or one of these really large accounts like Taylor Swift or a band you like. But it was much harder to find your friends. So if Mikey posts, maybe it would get lost in the shuffle. So we both had to optimize for time, but also optimize to see your friends, and it turns out that pivot and optimizing towards your friends had an enormous boost to the usage of the app as well. And we thought that was pretty good, and then it turns out stories came along.

And stories was... There's some trade off, but it was highly additive. And you look at Instagram's time spent today and I'm not internal so I don't know where it's at anymore, but we had no idea that we could have multiple chapters of success in terms of expanding the time people spend on Instagram by, again, focusing on what we talked about before, solving the problems that people have in their lives. They want to know what's happening with their friends, they want to know what's happening with celebrities and these public figures, and they don't want to have to trade off between the two. So if you just put two products that are kind of optimized for each of those groups in the product, they'll spend more time and they'll enjoy it more.

So, again, coming back to solving problems, every single step of the way was just, "How does this product stink for this group of users and how can we fix it?" That helped increase time spent significantly by the end.

The Impact of Instragram

Patrick: [00:41:23] I'm curious in the Jobs-to-be-Done lens, what stories was accomplishing so well? Because, my own subjective take on this is I often will just look at those and not even do the feed, and it seems like the volume of my friends putting stuff. It's way easier to know what's going on in their lives by looking at their stories and their pictures. Is that what it is? That it's just more ephemeral stuff?

Kevin: [00:41:44] Well, initially, we would see people post photos of whether a Snap code, or they would link to their Snapchat profile in their bios, and we asked ourselves, "Hey, what are people trying to hack here?" And we realized what people want to do is express their lives to their friends. They don't care what platform it's on, whether it's Facebook, Instagram, Snapchat, Twitter, you name it. They just want to be able to express. And people are trying to tell us, "Hey, your product's missing a feature. We want this thing in here. We want our friends and our networks to know that we have this type of ephemeral content that's lying around." Because if they linked it from their profile, it wasn't true that there was this completely other social graph. They were telling their social graph on Instagram that this stuff existed.

So the thesis was, and no one agreed at the time, I certainly didn't, that it would work, but through our experiments, first we try an internal version, hack on it a little bit, then you see employees start using it you're like, "Huh. Maybe there's a there, there." So you double-click on it a little bit, make it a little bit better, and what we saw by the end was that people just wanted to express themselves creatively to their friends and family or fans, and it doesn't actually matter if that's something that lives on your profile permanently or whether it goes away in 24 hours and it's full screen. It doesn't matter. Maybe there are new products that could be launched based on that insight. We just got two in and then we're not there. So.

Mike: [00:43:05] My take on stories was that, when you think about Jobs-to-be-Done, Instagram was being hired for a job and it had gotten... As an employee that might degrade in performance over time. Instagram had just gotten worse for that job. When you look at early Instagram, it's actually a lot more like stories. People would post four or five times a day, whatever they wanted. Over time, you have sort of what I think of as sort of the natural laws of physics of a feed product, which is, people start being more selective about what they post. We would interview a bunch of people over the years and they would kind of express what the rules of Instagram were to us, which was really, really funny to me. And they would say, "You never double Insta." What does that mean? "You don't post twice in a day, maybe at prom." I was like, "Okay." We've not built intentionally, but we've evolved a product that is self-limiting in a way that the original version wasn't.

If you scroll back all the way in my feed or Kevin's feed, I was posting multiple things a day and even I got more selective over time. So I'd like to think that we just got worse at the job that people originally hired us for and stories was the evolution we needed to make in order to recapture being really good at the job that people hired us for originally, which is, "I want to really quickly share what I'm up to with my friends and then engage with them on it when they see what I'm up to."

Patrick: [00:44:15] What was the most interesting thing you saw as Instagram penetrated more into developing parts of the world? I'm assuming that your success sort of followed better bandwidth and better mobile penetration or something like that. Any interesting observations? That must have been really neat to watch that happen.

Mike: [00:44:31] All sorts, especially in the last couple of years of Instagram I spend a lot of time thinking about, we've talked about Instagram emerging markets and how we make that work. Kind of things you notice in the US, you can rely on people having generally the same quality of network connection throughout a day and that can be very wildly in other countries. So we sent a research team over to India and they would say, "Oh yeah, I try to load up as much of my feet as I can at home because I have WiFi there, and then I go out and I'm really rate-limited everywhere else." So we realized our product was actually not great for that because the second you went somewhere new, it would try to refresh, and if the refresh failed, basically the product would stop working. So there's things you can do around thinking about people's day long experience of Instagram versus sort of single experience of Instagram.

The other thing you start realizing is how cost plays a factor. One of our largest growth drivers ever internationally, was, there's a concept called Zero-rating where basically a carrier wants to get more customers so they'll say, "Hey, Instagram is free for a month." Or, "Instagram is free up to this amount." Or, "Instagram is free for photos, but not videos." And this cell provider in India launched that feature for Instagram and it was incredible what a rocket ship it was. Where if that happened to at AT&T here, I don't think we'd see much more than a blip. Most people are either on lacks enough data plans or have universal WiFi. It's just not an issue. So you start realizing are there impediments to using your product different in different countries? Are you limited because most people are using an Android device that you don't support yet? "Okay. Well, we should probably go support that."

Also, an interesting example we had was, we often look at what version of Instagram people were signing up with. Usually it should be a pretty recent version because if you think about what happens when people sign up, they go to the play store or the app store, they download Instagram, it's the latest version, and they install it. And we're getting all these sign-ups from a version that was a year and a half old. We were like, "Ah, those spammers are at it again." And we dug into it and it wasn't that. It was phones being sold, I think it was in Nigeria, and they came preloaded with Instagram, an old version of Instagram, and it wasn't a deal they'd done with us. They just were doing this because they wanted that. And so we had to make it work well even for that use case where the bandwidth wasn't wide enough for them to just instantly download the latest version and you have to support the one that came on their phone.

So, all sorts of things, and these only manifest themselves as you grow. But that's where it starts becoming very important having teams that are tuned in to what's going on internationally.

Patrick: [00:46:40] It's fascinating to think about the ways in which Instagram has shaped culture, literally square-shaped things that show up that people can take pictures in front of. How much did you think or care about that? A specific outcome is like a medium is the message kind of thing?

Kevin: [00:46:56] We didn't care. Not that we didn't care about the outcome. Obviously we cared about the outcome, but how do I put it? There's a great venture investor here in town that we met with the other week, and he said, "Often the companies that basically start off searching for a great startup idea never find one. The companies that start off with, "Hey, I had this itch. I want to just scratch it. I think other people have this problem too. I'm going to create something really awesome that solves this thing for me." And to be very clear, the first version of Instagram was literally for me and Mike. It was all the stuff we wanted in a product-

Mike: [00:47:32] It's called Burbn, right?

Kevin: [00:47:32] But the future version of Burbn that would become Instagram when we decided to just do photos only, we focused solely on what we thought was an awesome product for us. And thankfully, that resonated with a larger group. Spur photos came... I mean, in large part, because I think a lot of companies were doing it at the time, like Hipstamatic and Camera Bag were kind of square photos. But I had taken this class in Florence photography class when I studied abroad, and my photography teacher made me use a Holga camera that only took square photos with... by the way, a pretty terrible lens. So the idea of how do you make terrible lens plus square photos work on the iPhone was something that I went back to my experience in that darkroom saying, "What makes these things more interesting?" Toning them, overexposure, light leaks, all this stuff, borders were a big deal. None of which, by the way, matter today on Instagram. When's the last time one of your friends added a border or put an enormous filter with light leaks on there. Think about that. But what got us there, didn't get us here. And just looking back at that experience. You have to just see what people say they want at the time, what problem it solves at the time, knowing that those problems get arbed away pretty quickly.

And then it's onto the next thing. But imagine Instagram today has live streaming. There's not a world in which I thought live-streaming could happen on Instagram V1 because cell speed was so slow. The devices were so slow. When did it become possible to take a video on the iPhone? I mean, maybe it was always possible, but I don't think you could share it. I don't even remember.

Mike: [00:49:08] If it was the iPhone 4’s is what jumps in mind.

Kevin: [00:49:10] Right? We should go back to the books to figure this out, but the idea that video was going to exist and then you could stream it and that someone could go live and stream it to the world. That's crazy.

Mike: [00:49:20] Seems like pie in the sky. Yeah.

Kevin: [00:49:21] No. And maybe I'm just not too much of a futurist or a good enough one at that. I consider myself a hyper pragmatist, which is just what problem exists right now. How do we solve it best? And then that'll earn us the ticket to go solve the next version and the next version, the next version. If you amass a community in the meantime, and you're nice to them, they will pay you back and then some. And that's how Instagram got to where it is today.

Mike: [00:49:46] That gets at a management challenge. I think we had over time, which was, I think we share this characteristic, Kevin and I are very complimentary in many ways, but share, I think a couple of set of values of ways of looking at the world. And one of them is that sort of what's available today or soon, like in the next six months. And how can we make that mainstream and make that a behavior. Then all of a sudden people... it unlocks in people, which I think was taking photos, it was shooting video. It was live streaming. And our teams would sometimes look at it and be like, "Yeah, but what are we doing in five years? What's the five-year vision."

Kevin: [00:50:14] I'd be like, "Promise me you'll still work here in five years."

Mike: [00:50:16] Five years. Yeah, exactly like average tenure was not five years. And we had a year I remember where we were like, "All right, well maybe we should be doing more of this." And we spend some time prototyping hardware ideas that never saw the light of day. And then I don't think our heart was in it for either of them. And then the end result of all of that iteration was saying, "Okay, let's solve a problem that we have today." Which is people want to post longer videos on Instagram, but it doesn't really fit in feed, which is where IGTV came out of.

But I think we eventually got to the point where we got the team to trust us in saying that the jobs are the thing that carry you forward three years from now, you're actually probably going to be solving that job that you're good at now plus some adjacent ones. I'm like technology, we will continuously be looking at and saying, "Alright, that can now be taken mainstream. That is now available on the iPhone." We can bring it to everybody. Not so much, "Wow, we're all wearing AR headsets. Let's build that today." Because nobody's going to use it. And by the way, you have a billion people that would have benefited from a different thing you could have been building on instead. I don't think enough teams think about the opportunity costs of being super pie in the sky. And obviously it's important to have R&D teams in some companies developing that technology or else it wouldn't exist. But I think folks often confuse very product heavy companies for very R&D driven companies and lose sight of solving problems that could be solved today.

Kevin: [00:51:27] I like to think of this. So I'm not a surfer, but I've seen surfing on TV once... I told you, I'm not a sports guy. There's an optimal time to start paddling to catch a wave. You start too early, you look like an idiot. You start too late, you'll look like an idiot or like you're trying to catch up and you don't quite catch the grasp. If you start at the right time, everything aligns. And Instagram started at the right time. It was right when the iPhone four had come out. So people were just beginning to think, maybe I'll use this camera on my phone instead of bringing along a camera on this trip. That was not a foregone conclusion at the time. But because we were just there right at the right moment with all the right ingredients, the thing took off.

And I think for whatever Mike and I do next, the thing we talk most about is what waves are happening right now. And it's rare that if your head's down in a company or if you're just staring at every other company and trying to figure out what waves are happening.

Mike: [00:52:19] Can't get a higher perspective.

Kevin: [00:52:20] No, you don't get a higher perspective. And I think machine learning is one of these waves that is absolutely happening. The question is how does it get applied and to what industries and what problems will it actually solve? Because you can't just, I mean, I guess some companies have actually founded just machine learning companies and sold them for a lot. But I think those are limited. And whenever-

Mike: [00:52:38] It's a tool not a product.

Kevin: [00:52:39] - Yeah. We always talked about, are you technology in search of a problem or are you problem-focused, searching for what technology will best solve it? And the honest answer is it's okay to say, "Hey, there's this new technology like the iPhone. What is it on lock?" But unless you then identify the problem that you're going to go solve you’re paddling too early.

Patrick: [00:52:58] What are some of the other potential waves that you've observed in the year since leaving Instagram?

Kevin: [00:53:03] We'd tell you, but then we'd have to delete the podcast. In all honesty I think machine learning and all flavors of it is the big one that I've seen. I mean, there's crypto, there's revolutions happening in personal transportation. Whether it's self-driving cars or electric vertical takeoff, and landing, I guess you could call them planes. The one that I get most excited about that feels like a wave, a real wave, technologically is the idea that we now have the ability for when I say the every man, I mean, literally just me with my laptop, to go apply machine learning to problems where maybe you haven't applied it before. And as long as you follow the rules and you're buttoned up, you can make some pretty revolutionary stuff happen. What about you, Mike?

Mike: [00:53:45] I was a cognitive science major in college. And I think it's interesting how you have cycles in your life. I ended up focusing more on the human-computer interaction, design UX side, and built that skill out, but have been reconnecting recently, like what happened in neuroscience in the last 12 to 14 years and in some cases, not a lot in other cases of vast sea change. So when I was at Stanford, you wanted to do an EEG measurement on somebody's brain. You're talking 10 to $20,000 of equipment. You can get a pretty beautifully designed off the shelf Muse headset for $200 now or 250. That's obviously the turnaround miniaturization and all these different things. But actually it's a story about machine learning, which is the only reason that the Muse, which basically helps you meditate, is useful is that they've trained a bunch of ML models on brainwaves.

And they could say, this is what calm looks like. And this is what stress looks like. And I'm grossly oversimplifying it, but now we can tailor an experience around it. So I'm excited about that combination of things. I just got the Oura band that measures sleep. And in some cases it's telling you what you already know. It's like, "Yeah, I did sleep very poorly last night." It's like, that’s why I feel bad. But also to me, it was interesting to say, "What does it look like when I am more stressed out before bed in terms of heart rate variability." So I think that combination of devices getting smaller, more powerful, the fact that the Oura can pack all of that into a ring shape is pretty interesting. The fact that the Muse can basically just look like a headband from star Trek. It's not something that you'd wear in public, but it's still portable. I can travel with it and generate useful insight is a combination that I'm pretty excited about.

Lessons Learned

Patrick: [00:55:11] You guys had an unusual, maybe singular experience of being kind of the prototypical early-stage garage-style founders from bootstrapping something interesting like Instagram through managing a very big team, but also doing it inside of one of the largest companies in the world. There's just got to be really interesting insight to tap from your guys' brains because of this experience. So the first is the ways in which your thoughts on leadership and management of teams and people maybe change the most or evolve the most in the time in which you were at Facebook. I don't know how big the team was when you joined, I guess 12 and a thousand when you left. So what major chunky lessons do you take from the leadership part of the job while at Facebook?

Kevin: [00:55:51] So there were... how many chapters? Well, I'll lay out the chapters and then we'll go through them. The first was the me and Mike phase. It was just us. We were hacking, we were getting stuff done. I mean, the big thing I think we realized there was enormous prioritization of your time working on things that matter. So again, identifying problems that people actually have, rank-ordering what you're actually focused on and focusing only on the top one or two things to the detriment of everything else. And it's okay.

Mike: [00:56:17] We had a no external meeting policy for four months before launch and it drove investors crazy because we're like, "What do you mean no? Who do you guys think you are?" And it was not born of arrogance. It was literally born of the fact that there were two of us and every minute spent outside our tiny desks on a pier was minutes that the product wasn't getting built and we've stuck to it basically until we launched.

Kevin: [00:56:36] But also, Mike remember early on, it was just the two of us. So scale and leverage was an enormous lesson that we entered into this death spiral, where we launched. And I think we might've been, no, we were two people when we launched. And I think we hired Josh, our community manager right after that. And then we got Shane to come on. And so it was four people, but the problem was, it was growing like a weed and we were fixing it and not hiring people. So we went into this death spiral of not having enough time to hire people. Thus the problems were getting bigger because it was accelerating beyond our control. And then we wouldn't have the time to go hire those people. So it just-

Mike: [00:57:12] Compounding problem.

Kevin: [00:57:13] - Yeah, it compounded. I don't even remember how we got out of it, but the point is, hiring people earlier for something that's working. I think it was both enormously helpful to be small and nimble when we had to be. But then when that thing starts going, don't hesitate to hire up. I mean, Mike did a lot of the early hiring early on, but also did a tremendous amount of the server fixing at 1:00 AM. There are pictures of him on a camping trip where he's got his laptop open with a wifi card, rebooting the servers, but scale on leverage, I think at that point was important. What about maybe once we joined Facebook?

Mike: [00:57:46] Yeah. I think what was interesting about that transition is we were, I think at that time 16, we'd hired all of four people during the time the deal closed and we needed to accelerate. And one of the benefits of being inside Facebook, is they've put a ton of work into sort of core engineering management training. So I think a lot of companies look around and say, "Oh, that person's been around the longest. Maybe they should manage." And that's often the wrong approach. So I think the two things from there that we adopted, I think were absolutely the right thing. Number one is, make management a parallel career track to individual contribution. And it sounds obvious, and I would have assumed that most companies do this, but I was just speaking at a dinner for growth stage companies. And I was talking about this point and be like, of course everybody does this.

And somebody came up to me afterwards. He's like, "I was about to roll out our new career ladders and they don't do that. And management was a promotion. I'm so glad I came to this dinner tonight." But it's essential because you don't want people to self-select into management just because they see it as a promotion or the way that they're going to get more authority in the company. But as a calling or at least an interest, if like, "No, no I'm more interested in the people side and building out the teams." And that's something we adopted very early and we actually let people transition from one to the other. In one case, one person went into management did that for two years, was excited to get his quote-unquote hands dirty. Again, did that for another year, then back into management. So it can be fairly porous on either direction.

So I think that was a pretty interesting set of things. And then another mistake I think companies make is that they're like, "Now you're a manager just because we called you that, you know everything that you're doing, or like maybe here's a book." Actually that's a real transition and requires real mentorship. So giving people that extra set of resources for that first year, whether it's extra coaching, either internal or external, doing skip levels, meaning I would often go talk to their direct reports and say, "How's it going with this new person?" And they'd be like, "Oh, it's great, except for this one thing, or it's a disaster." And you're like, "Okay, well maybe we need to hit the reset button and do something different." But just a lot of intentionality around that practice of core management was key.

And I think that's how we got from 16 people with no managers to basically a year later, we were almost 80 and we already had probably 12 managers managing that. And you sort of start figuring out the right ratios that you wanted. So you make a great point. We grew up in a very interesting way where our first managers, other than me and Kevin happened once we were already inside Facebook, which meant that we were able to look at what was working well about their internal core engineering management discipline and say, "Great, that works really well. Let's adopt that as much as possible."

Kevin: [01:00:04] I think one of the things that maybe we both caught on towards the end was don't forget that your main job is to build great products for people. You can get into this state where you've got career ladders and you've got all these tools for managing at scale. When's the last time you sat down and said, "What product are we going to build for people? What do they care about?" Or are you just sitting there having people bring it to you that don't actually subscribe to the same philosophies that we did early on. And I think it's really important to remember that the job, yes, you need to scale, but the job is not to just manage people. The job is to build great stuff. And if managing people is in the name of that, then awesome. So Mike and I would try to spend a tremendous amount of time just bringing people back to that core problem.

It's like, I know you care about your promotion. I get it. But can we please just talk about the product for a second? And I think that you can get to a certain scale where people forget that the fun job is not just the politics of managing a large organization, but instead you got to work on the product. I also think one of the lessons we learned throughout our time at Facebook was, do not pat yourself on the back ever. We got hit in the face so hard, multiple times launching products that didn't work. And thankfully, very few people remember the name Bolt. It was a little app that we launched, never worked. I'm surprised I'm bringing it up right now. But my point is people forget your failures. They remember your successes. So forever people will remember our stories pivot. Very few people will remember the 15 projects that came before that, that failed.

My point is, sometimes when you're a successful company, you can get through these chapters and you just think you are the hot S, like you're amazing. And what ends up happening is you get lazy and you focus on the management stuff. And what you have to remember is that there's someone out there who really, really wants your spot in line, and they're going to fight hard and they're going to launch an amazing product, and if you're not hungry and if you're not paranoid, and if you're not focused, like I said, on building those products, you can pat yourself on the back to getting this level and assume everything that got you to this point will get you to the next point and you are going to be dead wrong.

And I just think we had enough failures early on. And maybe because we sold the company and people were like, "Oh, you're done." We wanted to prove them wrong. I think we stayed really hungry. And we tried really hard. And we learned not to just sit upon the perch and assume that everything was going to go well. And I think that's part of why it continued to go well.

Mike: [01:02:36] I think we would learn this lesson. We'd run by annual employee surveys and obviously check in with people in between, but almost every metric in the end was correlated with, are we doing great work on the product? Like I remember it was like, "How optimistic are you about Instagram's future? Do you feel like you're growing at Instagram? Do you see yourself here? What tenure do you expect to have at Instagram?" And the pre and post of thing like stories or thing like ranked feed or thing like pivoting and making direct work was stark. We would do initiatives internally to do the management side of things. And they we're important too, because often they'll fix some block and tackle that you're dropping the ball on, but then ultimately that rising tide of, "Oh yeah, we are excited. Ah, Mike's kind of annoyed that this org is not exactly structured the right way and yeah I'll keep raising it, but I'm really excited to get to tackle this big thing."

Patrick: [01:03:23] Listening to you guys. There are two major things that I extract. And I'm just curious if you think these two explain things best. So the first is this community slash product centricity, which you've heard other great modern business leaders talk a lot about this principle, Bezos, most notably, maybe with this kind of customer obsession idea. And then the second, which is really interesting is this humility tied to pragmatic observation, that you're not trying to predict what the problems are going to be five years from now. You’re trying to look around and see what they are today. And the humility piece is the willingness to disrupt your own business. Do you think that those are kind of the right two pillars? Am I missing a major one?

Kevin: [01:03:58] I totally forgot. We had this principal early on, or we called it a value, which was humble, but confident. It's like, "You should be ripping confident about your idea, but humble enough to know you might be wrong, figure it out pretty quickly." And I don't know why we focus so much on that humility. I think it was just because the fear of being wrong, the fear of missing this great opportunity. I mean, let's go back to this context. We were two guys who had pretty good jobs in tech, and we both really desperately wanted to try to take a crack at starting a startup.

And this was our one chance if this didn't go well, I mean, probably not going to raise money again, anytime soon, we'd just be those guys that burned cash. And this was our chance. So I think the only way to face challenges like that is to say, "We are going to put everything possible into making sure these people who use our product are happy. And that we're going to try to be so humble that every moment we're going to maybe question ourselves four times before cutting or before shooting or whatever you want." And when we do that, that just raises the chances of success at every single trial along the way. Do you agree?

Mike: [01:05:04] Yeah. And I think almost to a fault, we were very intentional about every sort of decision that we made along the way. And I think that comes from that balance of the decision has to matter, so you have to be prioritizing correctly. So I think that comes from the product centricity and user focus. But then ultimately, you also have to be willing to say, "All right, let's debate this and then let's also commit and go forward." One of the phrases I really liked was from a case study, we both studied independently, which was, "We may not be right, but we're not confused." And we loved that because it really captures... You can't basically try to diversify your efforts when you're two people, you got two people, you should probably focus on one thing, but you also have the humility to say like, "well, we might not be right."

A story I like to tell is, Burbn.com, which was the product that we'd built before Instagram. That site and product was active and we were ready to pivot back to it after we launched Instagram, if that hadn't worked. Four hours until the launch of Instagram, it was clearly working we had to turn it off because the server was overloaded and we were just turning off left and right anything that we could. But I woke up in the morning, we launched Instagram and I was like, "I hope this works, but if it doesn't, I'm not done. I want to keep going. We still got money in the bank. Let's figure out what's next."

Patrick: [01:06:06] The two other areas that it's funny to talk so much about a business and not touched on these two areas yet Instagram again is unusual in this sort of build it. And they came distribution sort of handled itself. I'm sure that there were a lot of things that went into thinking about its growth, but it seemed to grow pretty fast, just virally from the jump, which is unusual, obviously for a business. So distribution is the first growth. And then the second is the business model. So this is a free product. And we know now how it makes money or some of the ways in which it makes money. But how often did you guys think, or do you think about growth distribution, building those channels intentionally versus just letting it be, letting the best possible product handle growth on its own because it's that good?

Kevin: [01:06:47] It's a little bit of both. How do you manage it? And then how do you not get in its way? How we manage it I mean, Mike and I would spend a ton of time early on optimizing the landing page. To be clear when I said the three problems that we solved, one of them was sharing on multiple platforms. And it turns out we both loved Mike was like, he has hundreds of thousands of users on Twitter and that was before Mike was the founder of Instagram. And we looked at that and we were like, "Wow, if we help people share to these platforms and their links back to our platform, that can be enormously helpful. So we optimize that landing page to make sure that when you clicked on a photo from a friend, you would see that this was the app they use.

Turns out the filters and the borders were actually calling cards for us. Some people were like, "Oh, how'd you do this app?" So we actually did architect it very well. At the same time, we had no idea what was happening underneath. So we didn't know how many new users we were acquiring versus how many were turning out versus how many were resurrecting. We didn't really have those metrics dialed in, but once we did get those metrics dialed in, we realized very quickly that often when growth wasn't going well, it was because of something we had screwed up. So we were like, "Get out of the way of the thing growing. We're screwing it up." I mean, Mike has a bunch of stories about push notifications, not working, or how do we screw our growth up?

Mike: [01:08:01] Yeah, no, exactly. And we would look at it and say like, "Wow, this growth went off a cliff. How did people change? Is there a competitor?" 99 out of a hundred times, it was the push notification system broke. That was a very, very common one or some other engagement thing broke. Memorably, once we shipped a version that crashed every time you tried to upload a photo. If you were in Japan, you ended up getting this very specific, crazy region-specific bugs. Did need to make some changes. As we grew, we're like, "It's hard to find a username." That seems like a trivial problem, but you start running out of usernames.

How do you help people through that signup process? Is not the same way you would have helped them through it at the very, very beginning. But I think going back to the surfing metaphor, the way you distribute and grow is also very much tied to when you're building it. People come to me all the time they're like, "Hey, I'm launching an app. What should I do to make it as big as Instagram?" I'm like, "Time travel back to 2010 and we can have a conversation." Cause it's a different thing. The avenues of growth we have, for example, we shared a lot on Twitter and Facebook and then you would get attribution and you would share back. I don't think that's the case for either of those platforms anymore.

Kevin: [01:08:59] Mostly because Instagram grew so much I was like, "Wait a sec."

Mike: [01:09:03] We grew off of the Twitter and Facebook social graphs when you can connect and find your friends and that went away. And at the same time, if you're building a startup three years ago, the way to do it was raise a bunch of money and try to outspend everybody in-app store or app install ads on Facebook. I think that's changed too. So I think iteratively, you have to figure out what are the available mechanisms of distribution. We were playing with an app that tries to distribute itself by being a sticker on Snapchat and hope that people tap the sticker and go back to their app. So that playbook is obviously new and different. So that always was evolving and always changing and also became a function of scale too.

For us what we were trying to acquire early were, people who were into Twitter and Facebook and using photos when you're acquiring your billion, first person, you're thinking, "All right, they're in a different country. Have they heard of Instagram? Maybe they already have heard of it. Is it about making sure they know their friends are on there?" It's about thinking more creatively, but never forgetting that block and tackle basics of growth because that was often where we would have to turn to and say, "Ah that thing. That is what we need to focus on."

Thoughts on Business and Investing

Patrick: [01:10:01] I'm curious since it was a free product, how early on you were thinking deliberately about the economic model for whether it's ads or whatever else, what role does that play in this whole story?

Kevin: [01:10:11] I'm smiling just because in order to get Mike to work at Instagram is Brazilian was Brazilian-

Mike: [01:10:17] It's all Brazilian.

Kevin: [01:10:18] - I guess you'll always be resilient. Yeah. But we had to get them an H1B visa and they didn't believe that the company was real. They were like, "Is this a shell company? Are you just trying to get..." So Mike, what was your first assignment at Instagram?

Mike: [01:10:31] So basically they're like, "This visa we can't approve it until we have more information. Just send in us your business plan." Thinking, I'm sure the government was like, "Of course they have a business plan, they’ve raised capital"

Kevin: [01:10:38] So we look at each other and we’re like, "What business plan?"

Mike: [01:10:42] We'd raised on a hope and a dream and an app demo which Kevin had raised because I wasn't even involved in it. So I wrote up a business plan to capture our hopes and dreams about Burbn, which was the product we were working on. And the business model of the time is that it was all about local discovery. So we would sell local ads, local business could run local hyper-local ads and pay to basically promote themselves on the platform. Obviously, as we pivoted away from Burbn. I’m not even sure that business automated since the beginning. But it definitely didn't make sense when we went towards Instagram.

And then I think you look at the nature of the product and it's super visual and the best ads have always been visual. So there was clearly a plausible path to that working, but exactly what form that took in terms of, is it programmatic, versus is it a new magazine type model where you just spend a bunch of money on one big moment? I think that was all left to be figured out. But I think we both always had the intuition once we were working on Instagram rather than Burbn, that this thing is monetizable. If it's large enough, we'd be idiots not to be able to monetize it in some way.

Kevin: [01:11:40] I do remember our pitch to Benchmark Capital who that was our series B, I guess.

Mike: [01:11:46] Series A. Yeah.

Kevin: [01:11:46] Series A, you're right. So we had our seed, our Benchmark A. So man, if I could go back and re pitch, I would try so much harder. I think we walked in with maybe three slides, maybe a user growth slide, a business model, a question mark slide.

But I remember the business mark question model slide, which it featured Instagram and we had two screenshots. And I remember it had a picture of Banana Republic, and I had a photo and it was an ad by Banana Republic effectively. And we didn't have ads at that time. And I don't think we had Banana Republic at that time. But we got to a point where they did sign up and that did come true. And I'm pretty sure they advertise on Instagram now. So, that was our series A. I mean we understood that you could have ads on Instagram. I think our pitch was, "Listen, what are the most effective ads in the world today? Where does people spend all their money? It's print and video advertising on TV, magazines, etcetera." It's not that different. This is just on your phone.

Mike: [01:12:39] Yeah. And then it evolved and figured it out over time. And at the beginning it was highly manual and we thought of looking at every ad. I showed one of my favorite, do the simple thing for stories that we ever had. So the first ad product you could buy on Instagram, it was a pretty big deal. You maybe be the only advertiser on Instagram that day. We're starting basic and small and very controlled. And you were a tech company. You would have built this great auction system from day one to figure out who ran on each day and made sure they got the right impressions. View one of our auction system was literally a whiteboard with a calendar that was drawn in Sharpie. And basically, maybe March 5th would say, Banana Republic, they're running the ad that day. In the morning, the engineers would look at and be like, all right, we got to make sure Banana Republic is running that day. Which, on one hand, is what we'd call clown town.

But actually it was like a great example of doing the simple thing first. If that ad system didn't work, meaning people weren't interested in buying those kinds of ads on Instagram, why would you have spent a year building the perfect auction model? And when it worked, then you go build the thing that lets thousands and thousands of different kinds of ads be running on Instagram at any given time. But that evolution happened over time. And what I always loved, like I would look at the whiteboard and be like, "Is it Banana Republic time now?" They’re like, "Yeah, yeh it's today. We're running the ad."

Patrick: [01:13:43] I'm curious how you guys think about your own kind of business and investing ambitions as an investing show. I haven’t talked about investing at all. I know you're interested in companies being built and problems being solved. And so how do you think about what to do as investors, as allocators of your own capital now? You don't have to spend as much time thinking about how to allocate capital and resources within a much bigger business.

Kevin: [01:14:05] I'd like to say we were always venture investors. We just hired our own entrepreneurs. And again, at every moment you're deciding how to spend time and also resources when you hire people on what products and projects, they just all happen to be related. Maybe that's why some of the best investors, especially in venture tend to invest in verticals and get it right, because they understand the space versus spreading themselves too thin. I don't know. I mean, maybe listening five years from now to this podcast, I'll be like, interesting, that's how I thought about venture investing and maybe I'll have a completely different view. I mean, the world of investing is giving ideas the chance to take flight, hoping that they work and basically just having different levels of risk.. At the seed stage, you're like, I have no idea if this product is going to become this world-changing thing.

And in Instagram, I like to say whoever invested in Instagram early had zero insight into what Instagram would become, because they didn't invest in Instagram, they invested in Burbn. Maybe there's a view there on the team. And maybe there's... I think we all got very lucky back then. And then there's once it takes flight you have to decide, okay, does it have a chance? And then once it's really big, the question is what are the growth avenues? And maybe it becomes public someday. And then you have different juries all along the way. But as an investor, myself, what I look for are great companies that understand their customer in and out and are solving a really unique problem that no one else has quite figured out how to solve. And those companies tend to do very, very well.

But I have very little track record. I mean, I have a little spreadsheet. I said, Mike, I was like, here are all the venture investments I've done. And I'm like the multiples it's looking good. It's not great, but you got to be humble because you never know. But looking at it more focused in the next few years, I have this one mentor who's like, "Yeah, you figured out how to be a great entrepreneur." But being a great investor, it's a completely different game. And unless you approach it with that humility of, “I might actually be pretty bad at this. So how can I learn as quickly as possible?” I think that's the only way to figure out this next chapter investing. But again, I don't know if it's 10% of our time or 90% of our time, but it's a fun game to figure it out because it's so different. As much as it was similar.

Mike: [01:16:13] We walked into a meeting with a VC a month or two ago, and he's like, "The biggest thing I've learned in my career as a VC, product doesn't really matter, go to market is everything."

Kevin: [01:16:22] Both of us were like, "What?"

Mike: [01:16:25] And that could be the right thesis, maybe within a category of things. And think if that were the case, I don't think I’d enjoy investing because if there is any amount of edge or unique view we have on the world, I think it's what problems are solved, how well the product is solving those problems and what team is going out and solving those problems, and how are they tackling? What's their unique view of the problem? And I have to believe that’s an interesting view on the world and will lead to some ability to spot things, although maybe it doesn't lead to the ability to spot things very early.

So I think we're in the total discovery mode of understanding where we add value as well. So far, my intuition is not so much the seed stage, where you feel like very lottery ticket-ish. There often isn't a product. So I'm not sure there's a lot we can even get into in terms of like what problem they're solving yet. Whereas actually the sort of underappreciated, maybe part of our journey was going from that 16/24 to 1000 employee base where you're saying, how do I motivate a huge team? How do I think about building a repeatable product process? How do I think about having multiple bets in play at once and make sure that we still deliver on the core of the product? All of that stuff is really interesting. And in many ways the companies I enjoyed spending the most time with are in that series B, have traction, but don't really have a company yet around it.

Patrick: [01:17:35] How about mentors for you guys? You just mentioned a mentor of yours and you also mentioned this idea of being thoughtful about training people inside of Instagram, whether it'd be the individual contributor track or the management track. What were some major mentors that you guys had that made an impact on your thinking?

Mike: [01:17:51] One of the people I met in college actually, because I was taking a class on digital design and building things, was Reid Hoffman. And at the time he was still mostly involved with LinkedIn. And hadn't yet thought about... was maybe doing a little bit of investing, but wasn't full-on venture yet. And the thing about mentors, there's probably a theme that you have learned from each of them. And for Reid, I think it's leverage. When he goes and says, “Hey, I want to make this podcast happen. I want to make this book happen.” He gets somebody great to partner with and then make that happen. And I think as a self-reflection, I'm the kind of person that tries to do too much myself. I'm like, of course I can figure this out. And the answer isn't that you couldn't have. It's that, one, there's probably somebody else out there who's been thinking about it for much longer, who could do that faster and better. And two, that incremental minute could be spent thinking about some different things. So that question of leverage and how to scale yourself as one that, on self-reflection need to be even more mindful of now that we don't have a team anymore, because quite literally there's no delegation that can happen right now. What are we working on tomorrow is all that's getting done right now in this partnership.

Kevin: [01:18:46] So for me, it actually came later on an Instagram. Because early on I was like, yeah, I mean, everyone looks up to Steve Jobs and everyone looks up to Bill Gates. And I think they should because they've been enormously successful in what they do. But it wasn't like a personal relationship or we'd go back to being specific athletes. They were all very specific athletes in their own arena, but it never quite clicked. And for me, I went outside of venture startups and I said, who just thinks the way I like to think? Very logically and principled. And I got introduced to Ray Daleo and I remember my first conversation with him. We were talking about some of the ways we ran Instagram and some of the ways they run Bridgewater, and man that couldn't be more different. But having a very specific view and holding onto it, I thought it was something that was really special, regardless of whether you're right.

I think I learned a lot from him to think a little bit more principled about what we were doing and what are the rules of the road inside of Instagram? What do we believe? What are the things we think are true, universally, throughout building product? And the more we started coming around and writing those values down, some, I call them principles, I call them values for Instagram. The more I realized actually you can apply that over and over again. So he was an enormous influence on me. Still is. Looking outside of your own world to other industries, whether that's global macro hedge fund or whether that's a neuroscience for Mike. I think I get the most inspiration when you step outside of your world. Because that's what really inspires you. And you find people that think differently.

Patrick: [01:20:19] The idea of not being confused is such a... I never thought about it that way, but it's such a clarifying, the values and principles things through the lens of not being confused makes so much sense to me. I'll definitely take that away. In closing, a couple fun topics that I know you guys are interested in or passionate about. One of them is flying. So I know you're learning to fly or have learned to fly. What drew you to that and what have you most taken from it thus far?

Kevin: [01:20:42] So number one, I'm going to answer the question in a roundabout way, which is, I think everyone should have a someday, maybe list. A friend, way back in the day, gave me this book, Getting Things Done by David Allen. And most people skim it, and they're like, yeah, okay. I have to-do list and triggers and action. And they kind of subscribed to it. They don't really... And then they forgot about it. I went pretty deep on it, but the one thing I came away with was you should always have a list of things that you think are really interesting, but you don't quite have the time to do right now. And I have a long list of those, and then it keeps growing. But when I left Instagram, I said to myself, okay, what do I want to do? And I was like, well, you know, I could go venture investing.

I could go hiking. I did the skiing thing we talked about. I looked down at my list. I said, Oh, that's right. I have this list. And I went through it and I was like, I think people who sometimes leave their companies get a little, I don't know, the word might be dull. They just get a little slow. They're not as sharp because they're not managing something every day. And I was like, I want something really hard to learn where my life's on the line. And that could be like thrilling and exciting, but also very technical. So I was like, Oh look, I could either do my trip to Japan, which isn't those things or I could learn to fly. So I looked up a local school, ask them to learn to fly. And I went up and I was like, okay, this is definitely going to be the thing that keeps me sharp.

And by the way, I went through my regular license and then I got my instrument rating and the course of about six months of training, which was pretty quick, but I had a lot of time. And by the end I was like, that's exactly what I needed. It was both a distraction from the tech world just to get my head free a little bit, but also an expansive view of what the human brain can do if you just give it a little bit of time. That's why I learned to fly, but I've had a lot of fun with it.

Patrick: [01:22:24] What else is on the sometimes, maybe list?

Kevin: [01:22:26] The someday, maybe list. Yeah. I can't say. Only when the items come off, do we talk about them.

Patrick: [01:22:32] Ah, first rule of someday, maybe list.

Kevin: [01:22:34] Exactly. But obviously having a kid, and I'm about to have my second child. Having a family is a big one. But when you're founding a company and you're going crazy trying to fix things, you just don't have time for a lot of these other big projects. It's a list of things I want to learn to do. It's a list of places I want to travel. If Mike goes to a cool hotel somewhere and tries some great drink or something, I'll put it down. So it's a mix of different levels. Some of them are a little bit cooler than others.

Patrick: [01:23:00] I love it. I read that you guys are both very interested in coffee. What was the source of that interest?

Kevin: [01:23:06] Trying to stay awake.

Mike: [01:23:08] Coffee is funny for me. So I'm from Brazil. Brazil is obviously a huge coffee country. I actually grew up disliking coffee. And I think I really, in retrospect now, because you mostly drink coffee with milk. In Brazil, I've never liked that, and I still don't like it. And I remember, it actually goes back to this craft principle. I remember moving to San Francisco wandering into this weird dirty alley and I was just on a walk and there was a coffee shop in the wall. And I'm like, “What is going on here?" And having this awesome mind-blowingly good experience. I'm like, "Oh, this can actually be a real pleasurable, intentional experience." And that was Blue Bottle in one of their first locations. And I think it was actually a fun instance of craft at Instagram early on, which is, we got everybody... so Kev was a barista in college and understands espresso really well. And we taught basically every new employee how to pull reasonable espresso shots. Yet at any given moment you see somebody over there grinding and tamping it down and extracting and wandering back to their desk. It was like a good sort of onboarding experience that we both enjoyed.

Kevin: [01:24:04] I think the key here to understanding this story is that I didn't want anyone touching the espresso machine that wasn't trained. I mean, I'm like fairly detail-oriented. Some might say overbearingly detail-oriented, but the idea that a new employee was going to come on and touch this espresso machine and get it all dirty. I was mortified. So it's true actually. And I don't know if you know this, but one of our early employees, Shane, I was talking to him the other day. He opened up a coffee shop.

Mike: [01:24:28] Did he really?

Kevin: [01:24:29] Yeah.

Mike: [01:24:29] Oh, good to go.

Kevin: [01:24:30] Yeah, exactly. I mean, it's not his only thing that he's doing, but my point is it's traveled far and wide in the Instagram community, but early on, it was just like, we've got to stay awake. I couldn't really afford to go to the coffee shop all the time. So it was like, if you could have a really nice espresso... We didn't even have a sink nearby. It was a terrible idea.

Patrick: [01:24:47] What's the key to pulling a good shot?

Kevin: [01:24:49] If I were to give one answer, it's measurements. Absolutely all measurements. If you don't have exactly how many grams of bean you're going to grind in, if you don't know the coarseness or the fineness of that grind, if you don't know the time through which you're pulling that shot, is it 27 seconds, 28, 32, those are all normal, but you have to know. And then how many grams out do you want? If you have that recipe, you can go really far, and then you can get super geeky and you can look at temperature profiles, pressure profiles, all sorts of stuff. These days, with a kid, when I wake up in the morning, I just want fast coffee, and pulling an espresso shot is a great way to get your mind working in the morning. Because it's this little balancing act of getting everything set, but it's good. When you hit it, you hit it. And when you don't, if you're a little bit off, Oh, it's nasty. Punishes you.

Mike: [01:25:36] Yeah. There's probably four things that I think people in tech are naturally drawn to. I'd say coffee is one of them because you can be very precise. Photography is another one where I think this is actually gone away with the iPhone being the predominant way people capture or phones being the predominant way people capture. But for a while people would love being like, Oh, what's your ISO. Ah, I was able to get this great low light and do this long exposure. I've stopped three. And I love that stuff too. And at some point I realized I was more in love with the process and the technical side of photography, and the natural photography side of it. Flying is the other one, I think, where there's all these things going on all at once. And you're trying to figure all that. And then cooking is probably the last one where tech has managed to take over cooking as well.

Like I love doing Sufi and being like, all right, we're going to do this 36-hour carnitas recipe at exactly 1:44 and then I'm going to brown it. I don't know. I think it gives you an excuse to learn something new.

Patrick: [01:26:24] So you guys are roughly my age. I'm 34. You've had a fascinating set of life experiences thus far. What parting thoughts or lessons would you leave people listening with that you've gleaned from your experience, advice or otherwise?

Mike: [01:26:37] Never self-selecting out of things because you don't feel like you know them yet. When you look at us, we were not people with computer science degrees. I had a degree in human-computer interaction, Kevin had a degree in management, science, and engineering. We knew some programming. But when I think... I found a photo recently from week one and a half of Instagram and my friend, Sean, is sitting next to me and I think it's Kevin's photo. And the caption is like, working on the servers or something like that. And it was a friend of mine who had worked at the previous startup with, and he'd come over and try to help me with some of the scaling things. And I was like, I looked back on what I didn't know then about scalability, about servers, load averages, setting up good database design. I had no idea. And we had no idea about management.

We'd never managed a person. I would joke with our Instagram team. Every person we hired was automatically the largest team I've ever managed in my life. So you have to have the self-awareness to know that you don't know everything, and hire people that can teach you or get yourself that lesson otherwise. But also the open-mindedness of saying, I don't know that yet, but not knowing that is standing in the way of me proving something out or solving some problem. And then taking that learning mentality. That was us over and over and over and over again.

Kevin: [01:27:39] I think I have two connected to this. One is no matter what new idea you have, people are going to tell you it stinks. People are going to tell you it's not possible. It's been done a bunch of times. I mean, think about how many times when we said we were doing photo-sharing. People are like, ah, photo-sharing’s been done. When's the last time anyone innovated in photos. They're going to tell you this. And the goal is twofold. One is, first of all, if you're looking for some sense of congratulations or ‘go get em’, you're not going to get it. You're just not. The second is that doesn't mean it's a terrible idea, but also doesn't mean it's a good idea. The one common thing of all great ideas is that they usually look pretty bad at the beginning. Also really bad ideas also look bad at the beginning.

So it's your job, I think, to tease out quickly, are you on the good side or the bad side? And I think that's something we did fairly quickly. I mean, we had Burbn which wasn't working and it was on the bad side. There was some good things about it, but it was on the bad side, but we learned quickly, we isolated the things that were working and then out came Instagram. But everyone along the way, I mean, even the first couple of years, people were saying Instagram is the fad. So the ability for humans to foretell the future is terrible. So don't focus on that. Just focus on whether or not you have conviction and based on the data, do you have proof? I think the second thing, and this is the final thing is just to stay super curious and stay a learner. I think people throughout their lives will get to... I don't know if you're lucky you get through college and then people stop learning and then they stop trying and they go, they get a job at a desk and maybe they listen to a podcast to pass the time. Maybe that's you listening right now.

But think about it. When's the last time you picked up a book on a new subject, something that peaks your interest. I mean, maybe you don't have a lot of time because work is super busy. Maybe you're managing a family, etcetera. Find ways of staying curious and learning new things. Because, it's super sad to think that the last moment we learn is that last day in college. My goal is for college to last my entire life, even if I'm not paying for it.

Patrick: [01:29:46] I love both those answers. And I got this image, Mike, when you were talking, about anything that if you're going to be really successful. Almost by definition, it means you know 5% of the stuff that you're going to know at that end state. What could be more exciting than that? My closing question for everybody is to ask for the kindest thing that anyone's ever done for you.

Kevin: [01:30:04] I can go first.

Patrick: [01:30:04] Go for it.

Kevin: [01:30:06] This is the good thing about having two people. Okay. So kindest thing. It's hard because there are a lot of kind things along the way. But the one that stands out to me is my uncle, when I was, I'll say in fourth grade, so this would have been, I think, '94. I was getting into computers and I had this curiosity, and he ran this computer business, and he could tell I was excited about it. And I had learned some QBasic programming. Remember that Mike, QBasic?

Mike: [01:30:30] Yeah.

Kevin: [01:30:30] I was playing around with it and showed him that I was excited. We had one computer in the house and my dad was doing it for work. My mom needed it. The family shared it. And one time, I don't remember exactly when, he came to our house and he pulled out a laptop and I still remember it because the screen was green and black.

It wasn't black and white, it was black and green. He said, “Here, you can have this.” And I don't even remember if it just ran DOS or something. It basically let me start programming. And I had it in my room and I remember the table it sat on, and I would use it every single night for hours on end, instead of doing my homework and for him to take what probably was really expensive at the time. I mean, laptops were not cheap. And just say here, kid, take this thing, program your heart out. That was definitely the kindest thing anyone's ever done for me.

Patrick: [01:31:16] Wonderful.

Mike: [01:31:17] It's hard to pick one, but the one that came to mind. So my now wife, then girlfriend, for Valentine's day in 2011, I want to say, really wanted to do something different and special. And we've been dating for two or three years at the time. And she decided she was going to learn to code and build a thing called Lovestagram, which was around for a few years. Where basically you put your name and your Valentine's name in, and it would go through both of your Instagrams and find photos of you together basically create like a digital postcard of that stuff. And it was awesome. One, it's just a very thoughtful, cool thing to do, because we were deep in Instagram at the time, literally the most important thing in the world at that time, other than my relationship with her. But two, it was fun.

It was like the learning that she went through for that. And she kept it a secret until the night before the day. And then she was like, “Hey, I need to tell you this. Because I'm going to launch this tomorrow and I'm worried it's going to fall over.” And I basically spent the next four hours adding Memcached and making stuff asynchronous. And I was like, that's my way of expressing love back, is I'm going to optimize your gift now. But just the thoughtfulness, and the fact that learning has obviously been the theme of our conversation, but the fact that that was the thing that she chose to do. And she had almost basically no programming experience before that, to go from that to a fully functioning Django app with use of the Instagram API and web design was incredible.

Patrick: [01:32:28] Well, I'm really genuinely thankful to you guys for the time. We talked before I hit record about my idea to start talking to more and more CEOs. And I think I like lists and having like a to-be CEO list is really interesting. And I definitely would put both of you on that list in terms of how you think about business, how you think about community and product. It's incredibly refreshing, all the lessons we've talked about in the theme of curiosity is obvious. So thank you so much for your time.

Kevin: [01:32:52] Thank you.

Mike: [01:32:52] Thank you. It's awesome.

Transcript by @ScarrottKalani