JAMES: Welcome to Happy Paths, where we unpack the human stories behind some of the most used, most influential, and most famous software from the internet era. I'm James Evans. I’m the CEO of CommandBar, and the host of Happy Paths.


Gmail is one of the most used pieces of software ever, with 1.8 billion users today. In many ways, Gmail defined modern email, transitioning us from the [Playing AOL Greeting] "You've Got Mail" era into the present. Pretty much everyone has used Gmail or another email client that's been inspired by Gmail. 


I talk with Gmail's creator, Paul Buchheit, to learn more about how what he built came to be the juggernaut of the email world. In particular, I was curious about how Paul became interested in email in the first place and how Google gave him the opportunity to launch Gmail. Let's set the stage for what the web was like in the 1990s.


PAUL: I got my first sort of, like, real computer [laughs] in early '90s. What year would it have been like? '91, maybe, and got interested in BBS. You'd dial up with a modem. It was, like, a single-user internet. So it would be, like, a computer sitting in someone's closet or something like that, that you would dial into. And you could play games, or, like, send messages. So there was actually a messaging system called FidoNet that was almost like a Reddit or something like that, these communities. So I probably sent some messages there. The first time, I guess, I actually had real Internet email was, I think, when I went to college in 1994.


JAMES: And Paul was already thinking about ways to improve email back then.


PAUL: I had actually tried to start a web-based email thing when I was in college in 1996. I had this idea for trying to make email that kind of lives on the internet because, back then, you would have an app on your desktop computer like Eudora that would actually just download all of the email onto your machine. And then it lived there for normal Windows users or something like that. 


I was into Linux at that point, which was still...Linux was still early years and still kind of, like, this weird hobbyist thing at the time. But I had Linux on my computer, so I could, like, Telnet into my computer and run Elm. Elm was the email program of choice at the time. So I kind of had an advantage versus other people in terms of thinking of email as being a service on the internet versus a thing that lived on your computer.


JAMES: I tend to think of email as a concept born in the 1990s. But the first email was actually sent way back in 1971, using the ARPANET, which is what became the internet. So Eudora and Elm, which were both developed in the 1980s, were really second-generation clients, but they were desktop apps that you had to install on your computer. If you didn't have a laptop, that meant you couldn't get your email anywhere but home. 


This, of course, would later give way to the first web-based email clients like Yahoo Mail and Hotmail, which made it possible to get your email anywhere with an internet connection. This made them extremely popular, even though their UX left a lot to be desired based on today's standards.


PAUL: That stuff didn't exist yet at that point. But that launched around that same time of kind of that first-generation of internet stuff. I would have to go find the exact dates, but I think that they launched around 1996 as well. That was all first-generation, like web 1.0. There was no JavaScript or anything like that. So every time you would click on a link, it had to go back to the server and reload the entire page. 


And so it was never a great experience that you got from these web-based email things. The reason they were popular was because it was free, so anyone could sign up. And it was great because you could access it anywhere. You could just login from, like, an internet cafe, which used to be a thing. 


[Playing Audio Clip]


MAN: Oftentimes, people would come here from Europe and send email to friends as a postcard.


WOMAN: So I've got quite a few friends back in England who, you know, they have their email addresses.


PAUL: And so it was actually kind of this exciting idea that people could access their email anywhere. But the interface that you would get with that first-generation of Webmail was always pretty substandard. And so people who were serious about email always considered Webmail kind of just, like, a toy, not something that you would use for serious business. 


And even when we were working on Gmail, a lot of people thought it was a mistake and that we should build a desktop app or something like that because there was this idea that a web-based experience will always be second-class and just, like, not very good.


JAMES: This arc reminds me of so many stories that have played out in software. The browser is constantly underestimated as a forum for real applications; think about Photoshop and then think about Figma. Over time, the browser has transitioned from software used for surfing the web to what is essentially an operating system on top of an operating system or, in some cases, a literal operating system like a Google's Chrome OS. Getting back to Gmail, when Paul joined Google in June of 1999 as their 23rd employee, there weren't any plans from Google to try to revolutionize email.


PAUL: So, to be clear, email was already big. [laughs] People like sending messages. Messaging is, like, a fundamental human desire. [chuckles] So anytime there's a platform, there's something like email on it. So email was already super popular. It wasn't like we were inventing something new. You know, there had just been kind of a reorg in engineering in 2001 where Larry decided that Google...this is kind of funny in hindsight now that it's 2023. 


But, back in 2001, Larry felt like Google wasn't moving fast enough. [laughs] He decided the answer was to get rid of all the engineering managers and just kind of have each of the engineers working on a project and have it be more project-oriented. So it was kind of just, like, more of a free-range engineering structure. 


And so the project that they gave me was they wanted me to make something with email because they knew I had an interest in email. And I was pretty good at hacking together interesting new products and ideas. But there wasn't really a specific direction. But, culturally, there was a notion inside of Google that, you know, one of the keys is actually just to make a really great product. Because, at the time, like, the other internet companies were being run as these media companies that really had no interest in product. 


And even, like, sort of the origin story of Google, I think, is really telling in this regard, which is that you know, Google was originally a research project at Stanford University with Larry and Sergey, and they actually didn't want to start a company. They were [chuckles] just doing a research project. So they had actually tried to sell Google to, you know, the existing internet companies, which were...back then portals were the big thing, so Yahoo, Xcite, Infoseek, Lycos, all of that generation of 1990s Internet companies.


They kind of shopped it around, and they wanted a million dollars. [laughs] So they tried to sell Google for a million dollars, and no one was interested. And the story that I remember, which I think is most telling, is the CEO of one of these big internet companies, which probably no longer exists, was like, "Look, you know, we've looked into this. And the fact is that search just isn't that important. Like, people can't really tell the difference in quality anyway. And then it's just a commodity, like, you know, build your search engine. Maybe we'll license it, but we'll license something different a different time.


JAMES: I would pay a substantial amount to travel back in time to experience what it was like to work at Google in those early free-range days. Today, most people don't think of Google Docs or even Gmail as particularly modern or experimental. In fact, many people view them as dated. But, like Paul said, back in the late 1990s, Google was one of the first companies to recognize that interaction speed mattered, both in search and in applications, and they were, of course, spectacularly right. 


So while Google search was starting to take off, the company saw other opportunities to create better software for the rapidly growing internet user base. That's where Gmail started.


PAUL: Really, the foundation of Gmail was just, like, hey, let's re-examine email and see what we can do from the perspective of, you know, we have these giant computer clusters. We can store and search lots of data. So, obviously, like, search is going to be a key element of this. And then we're just going to make it really fast. But even just the question of whether it would be web-based was not established at that time. That was actually something that got debated. And there were some people saying, "You know, we should build a client or something like that." 


I always believed pretty strongly in the web-based thing. And I had been interested in this mobile code technology for a long time. So, when I was originally thinking about building an email product in the '90s, it was actually because I was excited about Java. And it's kind of forgotten at this point, but originally when Java was announced back in, like, 1995, it was about applets and this idea that your web browser would download this little chunk of code and run it in a sandbox. And I thought that was a really great idea. 


As it happens, they never really got Java applets to work very reliably. But meanwhile, this thing called JavaScript, which is a completely different language but, for marketing reasons, it has Java in the name, took off. And, you know, the web browsers just kept getting better and better. So, by the time we were building Gmail in 2001, 2002, 2003, the web browsers were getting to the point where you could actually do really good stuff, but it wasn't always easy.


JAMES: A bit of internet trivia. 1995 was a colossal year. Java launched like Paul said, but so many other software and hardware projects launched that year also: Internet Explorer 1.0, PHP 1.0., SSL, even the DVD. 


[Playing Audio Clip]


MAN: DVD isn't just better than a VCR. It's amazingly better. 


WOMAN: Here's a scene from a movie on VH -- 


JAMES: Going back to the browser as operating system analogy, often what enables a new web application is a new API or a set of APIs that make their way into mainstream browsers. WebGL, for example, is a JavaScript API that enables graphics rendering in the browser. And this is what Dylan Field and co took advantage of to build Figma. Back in the 1990s, some of the fundamental APIs that underpin so much of web interfaces today were being published.


PAUL: So, one of the early problems that we would have when we would push the JavaScript really hard is that we were crashing the web browsers. And so there was also just a question of whether it was possible to make a full JavaScript-based application without just blowing up the web browser. 


Ajax is a term people made up after we launched Gmail. That stuff all came once people were, like, disassembling Gmail and figured out how it worked. [laughs] And so that whole design pattern, like, we basically popularized the concept. The underlying component was the thing called XMLHttpRequest or something like that. It was actually something that Microsoft had put in into IE. They actually had a web-based version of Outlook though it was, like, very heavyweight and clunky, but they had actually put it in there, ironically. I believe it was specifically for Outlook. So it was put in there to enable email things.


The kinds of stuff that we did with Gmail was really only becoming possible around the time that we were doing Gmail. And there had been a prior generation of people trying to do some of these, like, JavaScript-based applications. There was something called Desktop.com, I think. But none of the stuff really worked very well. So one of the things that was fundamental with Gmail was actually that we spent a lot of time making sure that it worked really well.


So a lot of people had tried to do JavaScript before but ended up making it too slow or just clunky and weird in various ways. So one aspect of this is, like, you really have to make it seamless and fast, and so we figured out how to do that. And then, you know, when people experienced it after Gmail launched, it seemed kind of magical because you would just, like, click on something, and it would be instant. And no one had had that experience with web apps before. And so that was when people got really excited about the possibilities of JavaScript. And they kind of made up this term Ajax.


JAMES: I've heard many founders attribute this, quote, unquote, "100-millisecond rule" to Paul. The 100-millisecond rule states that every interaction a user has with a piece of software should be faster than 100 milliseconds. Why? Because 100 milliseconds is the threshold of where interactions feel instantaneous to our human brains.


PAUL: Our threshold of what seems instant is somewhere around 100 milliseconds. That's just kind of, like, in our hardware. And so what was in the Google culture was this obsession with speed. It used to be when you do a search on Google; it would say, you know, found 10 million results in, you know, 0.01 seconds, or something like that, and emphasizing the speed was a part of the branding even. 


But it was also, just internally, something that there was a tremendous amount of monitoring of latency, again, much more so than I think a lot of other companies were just, like, well, whatever. It takes a few seconds. That's fast enough. And Google was obsessive about even just the size of the homepage, like how many internet packets. How many packets will it take to deliver the homepage? 


So I don't remember anymore, but it used to be there's a specific number of bytes, you know, people are keeping track of the size of pages. So we did a lot of work to try to keep things small and fast. And so that was certainly a cultural value that everything needed to be very fast. 


Going through one of the earlier prototypes at Gmail, like, with Larry Page, and he was complaining. He's like, "Ah, it's so terrible." You know, he clicked on something, and he's like, "Ah," you know, he would just complain about how slow it was. And he said, "I felt like it took at least 600 milliseconds." [laughs] And I was like, "You can't know that." [laughs]


But, you know, so then I run back to my office, and I checked the logs. And it was something like 640 milliseconds. He was that accurate. I was, like, really impressed. So then I actually wrote just a simple little JavaScript thing for myself where it would just time things. And with just a little bit of practice, I learned that I could train my brain to perceive time with 100-millisecond resolution.


JAMES: Paul is also known as one of the first partners at Y Combinator, which famously encourages startups like my own to ship quickly, talk to users, and iterate from there. Turns out that advice precisely reflects Paul's own experience building Gmail.


PAUL: The very first version was actually shipped in a day, written and released in the first day. So the way I was able to do that is I took the code from a previous project, Google Groups, which was a Usenet search. And Usenet was kind of like an old internet forums thing. But it's very similar to email in terms of the structure of the data. And so I just took all of my email and mangled it to look like Usenet data and just, like, loaded it into this Usenet search system that I'd been working on. 


And then I just launched that to engineering, and I said, "Hey, look, I built an email search tool. Give it a try. Let me know what you think." And so the feedback I got was, "This is kind of useful, but it would be better if it had my email instead of yours." And so that was my first feature request was like, okay, people want to be able to search their own email. 


So, you know, version two, I made a thing that would, like, crawl through people's home directories and pull in all of their email and partition it by user and whatever. You know, that took another couple of days. I launched that. And so version two would search each person's individual email, and then I would get a feature request, like, I want to reply to an email. [laughs] And so then I added the ability to reply to emails. 


And so, literally, the entire product was just iterated that way. And the goal that we had in terms of when it would be ready to launch externally to the world was we wanted to reach the point where there were 100 happy users.


JAMES: Gmail officially launched on April Fool's Day in 2004. In hindsight, it might have been the best April Fool's Day joke that wasn't a joke of all time. And if you're looking for inspiration to launch before your product is 100% fully polished, look no further.


PAUL: The launch was a little bit of a disaster. We weren't quite ready, honestly. The code wasn't even all written yet when we launched, but we were [laughs] working on a deadline. We were aiming for April 1st. And then what happened was; actually, someone inside the company had leaked it. And so the New York Times was going to publish a story about it. And so we wanted to get it out before that story. 


So if you go back and you look at the original press release, it's actually dated midnight UTC, which is, I don't know, much earlier in California. I used to go look up the time difference. So we launched, actually, like, the afternoon or evening of March 31st. The problem was nothing was ready yet. We didn't even have DNS. So if you would go to, like, gmail.com, it wouldn't resolve. So we finally, you know, after some period of frantic action, had a landing page. And so, actually, no one had a Gmail account at first. 


JAMES: And I think it was invite-only, right? 


PAUL: It wasn't until sometime later on April the 1st that finally I finished, like, the account creation code and created my first account. So I only got a Gmail account, let's say, 12 hours after it launched, [laughs] and then, you know, that worked out okay. So I invited everyone within Google to get a Gmail account.


JAMES: Today, we tend to think of invite systems as a FOMO tactic, creating artificial scarcity so people get more excited about scoring a coveted spot. But, for Gmail, the invite system was actually a rate-limiting tactic.


PAUL: And the code was early and, like, not very efficient. And I was worried about losing data. So it would keep lots of copies of the data in different places, in different formats. So, you know, when a message would come in, we would end up storing it, I want to say, 13 times, something like that. Basically, we needed an invite system in order to contain growth; otherwise, the whole thing would have collapsed.


JAMES: Now that USB drives run up to one terabyte and our iPhones are rigged with hundreds of gigabytes, it's hard to imagine a world where storage was a key differentiator. But we've caught the Moore's law curve very fast on storage. Back when Gmail launched in 2004, they dropped a hammer on the email market by offering a laughable amount of free storage; laughable today because of how little it is, laughable then because of how unimaginably large it was.


PAUL: Hotmail offered two megabytes of storage, not two gigabytes, two megabytes, which is today not even enough to send, like, a single photo. [chuckles]  So they offered two megabytes, and I believe Yahoo was offering four megabytes for free. And so when we launched, we decided to offer 1,000 megabytes. [laughs] So, at the time, it was crazy enough that people thought it was like, I mean, also because we launched in April 1st, but people thought it was a joke. And a lot of people didn't believe that we could offer something like that because it was so much beyond what anyone else was doing.


JAMES: I love how swashbuckling this launch feels, kind of slapdash, kind of arrogant. Crazy to think this launch happened just a few months before Google went public. I was going to make a joke here about how Google would never sanction this cavalier of a launch today, but I guess we've come full circle with the Bard launch from a few months ago. I like to think Paul's Gmail launch was invoked for motivation while the Bard team toiled towards their own launch.


PAUL: It wasn't seen as, like, a super smooth launch, no. We definitely were kind of pushing it.


JAMES: In typical YC style, Paul continued iterating after launch, lazy loading features as user requests piled up.


PAUL: The product improved a lot after we launched too. Like, there were a lot of features that were pretty incomplete or missing. Like,  when we launched, there was no, like, rich text compose. You could only send plain text emails. You couldn't do, you know, kind of, like, colors, or anything like that. The address book was super primitive. I don't think we had a spellcheck. Like, there was just lots of stuff that, you know, we hadn't really quite finished. And it was just, like, pretty buggy and unreliable too. 


Like, I think we had at most one nine of reliability that first month. [laughs] We had to work our way up to two nines of reliability. Prior to Gmail coming out, it was pretty normal for people to have, you know, a low quota. So one of the groups of users that were most enthusiastic early on were newspaper reporters because, like, they would be working for, you know, let's say it's the New York Times or something. 


Their account that they would have as an employee would have, like, a 30-megabyte quota or something like that. And that was totally normal that even if you were working at a big company, you would only be able to have, like, 30 megabytes of email. And that's a huge problem if you're a reporter, and you want to be able to kind of, like, keep notes on all these different meetings and whatever. And so they loved Gmail because they could keep all of their email, and they could search it really quickly. 


So that also then spread to, you know, imagine if you're a corporate IT guy who's managing the exchange server, and you're trying to explain to everyone why, oh no, actually, you can only have a 30 meg quota because storing email is super hard. That story kind of falls apart when the person could just be like, well, why does Google give away a gig for free, and I'm paying you, like, a full time-salary, right? [laughs] Like, why do I have you? [laughs] And, in fact, you know, a lot of those organizations then did ditch, you know, they're no longer running their own exchange servers. They switched to Google.


JAMES: Gmail finally launched to everyone, no invite needed, in 2007. They also increased storage up to four gigabytes, making good on an earlier promise that Google would, quote, "keep giving people more space forever." In 2012, Gmail had grown to 425 million monthly active users, a 93% jump from 220 million monthly active users the year before. That was more than Hotmail and Yahoo Mail, making Gmail's popularity evident. 


With improved search, threaded conversations, and further expansion on mobile phones via Android, Gmail had effectively changed the game for email forever. Now, nearly 20 years after Gmail's launch, I asked Paul what he thought about Gmail's impact from his current vantage point.


PAUL: I don't have these amazing insights into email or something. Our approach was really one of iteration and believing in things like speed and actually just talking to people and seeing how you can make a really great product and, you know, what's slowing you down, what's getting in the way. And so it's really more just a philosophy of product creation. 


It's the same way now. It's just you don't always know what the future is; you know until it happens. It's what makes Silicon Valley and San Francisco such an attractive place is just that this is where everything is happening.


JAMES: That was Paul Buchheit, the creator of Gmail. While at Google, Paul also built a prototype for AdSense and came up with Google's famous slogan, "Don't be evil." 


Thank you for listening to Happy Paths. For more information about the show, visit us at commandbar.com/happypaths, or you could send us a Tweet @CommandBar. 


This podcast is executive produced by Maurice Cherry, with engineering and editing from Mandy Moore. 


Special thanks to Paul Buchheit and, of course, the entire team at Command Bar. 


I'm James Evans, and this is Happy Paths. See you next time.