Culture / DevOps

What a Former GitHub CTO Learned about Scaling

19 Jun 2022 6:00am, by

For the 447th episode of Stack Overflow’s podcast, GitHub’s former Chief Technology Officer Jason Warner shared some fascinating tales of tech transformation. The conversation covered both the massive scaling effort at GitHub from May 2017 to July 2021 and the story of Warner’s own “non-traditional” path into programming from “farm country” in Connecticut.

Maybe both stories offer a lesson in the power of a scaling and the importance of an engineering mindset, culminating with some projections about which technologies will shape our future.

Beyond Farm Country

Warner assumed GitHub’s CTO role after over three years as a VP/head of engineering at Heroku, and nearly four years as a head of engineering for Ubuntu Desktop. And Warner has now become a managing director at Redpoint Ventures, sporting an online profile that says “I get excited by companies where founders are trying to do something that seems unimaginable to most people.”

But how did this journey begin? In that Connecticut farm country, the mid-1990s was a time when IBM got tax breaks for hiring high school students into so-called “co-op” jobs (where students are given actual workplace experience as a complement to classroom instruction).

So in 1995 “I literally joined IBM to carry printers and computers around the building and hook them up,” Warner remembered on the podcast. “And they eventually said, ‘Hey, if you go to school and figure out how to program and learn computer science, we’ll give you a job when you’re done.'”

Or, as Warner says later, “The only way I got into programming is because I was able to pick up printers and carry them around and plug them into token ring networks.”

Warner still remembers writing that fateful first line of code in BASIC. “I was kind of diving into what was available to me,” he said.

Warner acknowledged that he got a late start compared to other programmers. “I had a little bit of worry about my ability to hang with other people.”

“The only way I got into programming is because I was able to pick up printers and carry them around and plug them into token ring networks.”

At college in the mid-1990s there was also C and assembly language, but the recently-created Linux operating system “wasn’t a thing yet.” Warner only discovered Linux later through “some subcommunity” at Pennsylvania State University – though it gave him a broader perspective on technology.

“That is when I started to realize that there was a lot of other programming languages available to me. And then the open source world opened up and I started doing things like Perl, as an example,” he recalled.

But even beyond that, Warner confesses that despite a B.S. and Master’s degree in computer science, “I genuinely thought I was probably an average developer.”

So much so that now, even with the benefit of hindsight, “I have since come to learn that I very much was an average developer,” Warner said on the podcast. “But I was an excellent, excellent architect… an excellent engineer. You know, systems!”

Later Warner said “I have come to peace with it because I think that I’ve taken a much more holistic view to what it takes to build successful companies,” adding at another point a self-assessment as a “distributed systems person.”

It’s interesting where this leads, since Warner sees human systems (like organizations) as “just lossier versions of computer distributed systems.

“I have since come to learn that I very much was an average developer, but I was an excellent, excellent architect… an excellent engineer. You know, systems!”

“So I happen to know how to manage those well, but weirdly, intrinsically I knew how to scale them. I knew how to operate them and I knew how to make them more efficient because of the stuff that I did on the systems programming side.”

And this would all play out when Warner became the CTO of GitHub…

Beyond Programming

Soon Benjamin Popper, Stack Overflow’s director of content, asked the obvious question: “what’s it like to be the chief technology officer of one of the most widely used developer tools and also consider yourself a mediocre programmer?”

But it was an even larger challenge, because the position keep expanding. Warner became a kind of de facto chief product officer handling a whirlwind of other responsibilities, including engineering and design, along with security, support, and infrastructure.

The consuming question was always what to build — and why — eventually leading to the creation of GitHub Actions, the GitHub Packages hosting service, advanced security analytics and “all of those types of things, and eventually even Copilot and Codespaces.”

By the time Warner arrived, GitHub’s developer community was already “thriving” — so the larger issue was simply scaling.

GitHub had had “around 20 million” accounts when Warner started his tenure in May 2017 and estimates the site had roughly 1.5-2 million daily active users, with another 10,000 signups each day.

But by the time Warner left in July 2021, that had jumped to 50,000 signups a day — and 7 million daily active users. For conquering that scaling challenge, Warner gives credit to “a stellar, stellar, stellar infrastructure group… just world-class.” (And, Warner adds later, it was “filled with excellent programmers.”)

Together they were able to brainstorm ways to scale further. But beyond that, they’d also discussed ways to “really take GitHub from a site where you could collaborate on software, to a full-fledged, end-to-end software development platform — that can scale all the way to the planet.”

This is where Warner’s engineering skills were the perfect match. “I could actually sit there and listen and actually contribute to the overall architectural approach.”

Knowing What They’re Building

The podcasters asked about making the move from technical roles to a financial role (as a managing director at Redpoint) — but the answer was surprising. After 12 years of experience with VC-backed startups, Warner had concluded that “honestly VCs don’t know what they’re doing!

“They don’t know what they’re backing. They’re looking at a spreadsheet — they don’t understand why a technology is important. The excellent ones do, and a lot of them get incredibly lucky, but if you notice the VCs’ track records there’s very few people who have more than one successful investment.”

So given that perspective, “It didn’t seem like it’d be that hard to step into it and have some fun.”

But more seriously, Warner adds, “entrepreneurs deserve better. They deserve to have somebody who understands what they’re building and why they’re building it and really what comes next.” Someone with ideas on building the company — and how to mitigate things that go wrong — while making an impact. “It’s all what CTOs and CPOs do on a regular basis, and I don’t think they had partner on the other side of the table from the financier side at all.”

“If you notice the venture capitalists’ track records, there’s very few people who have more than one successful investment.”

So when Popper asks which companies Warner is drawn to as a VC, the answer is high-tech infrastructure companies, ones that are “building fundamental plumbing layers of the internet or next generations of software, all that type of stuff.”

One of Redpoint’s recent investments was analytics insight platform Cribl, while Warner also invested in the crypto infrastructure company called Alchemy, a perfect fit for his strengths.

“A brand new thing has emerged in the market, emerged in the world — they’re doing fundamentally new and interesting things. You know, I invest in that and help them build that out and scale it to a degree that no one’s ever seen before,” he said.

While Bitcoin creator Satoshi Nakamoto may envision a decentralized system, Warner believes that’s impractical, arguing that “all great decentralized products need a centralized person to make it accessible.” Warner actually seeks out “the centralized component of this that I can invest in.”

Following up, Popper asks what trends and technologies excite Warner today — and where they’ll lead us over the next five to 10 years. And in answering, Warner argues that cloud data platform Snowflake “ushered in a new era of cloud-type computing and brand new mechanisms,” predicting an era with a proliferation of real-time streaming systems and “way more easy and facilitated data movement, which will unlock new types of applications out there.”

Warner imagines combining this potential with edge networks. Warner hopes to see edge networks thrive, maybe in health care and other “regulated spaces.” And this leads to a hope for privacy on the blockchain, to unlock its full potential.

Or, as Warner put it towards the end of the podcast, “the only thing that I ever wake up thinking about is what is next.

“So I talk about developers, I talk about infrastructure, and I talk about what’s coming.”


WebReduce

Feature image by Dean Moriarty from Pixabay