After leaving my job to build apps in 2009, life took on some hard math. I only had so much money saved up. While I was making around $1,000 monthly from the App Store, the gap between my income and expenses was leaving me in the red.

So my maximum budget for a meal was $3. If I could get under that number, I tried.

This meant clever use of bulk buying at Costco and lots of meal prep. Going out for meals was an occasional indulgence, but the three dollar figure remained a constraint. Spending more than that meant having to bring home enough leftovers to break even.

I’d figured I could make some extra money working part time while I got the details of my app business figured out. To this point in my life, part-time work had been easy to come by. I took it for granted, like water from the tap. It would be there if I needed it.

But the financial crisis meant stiff competition for any work. I’d moved to Bend, Oregon for its low cost of living and beautiful scenery. Unfortunately, no one was hiring, and as the new kid in town, I had no local network to lean on. When a Kohl’s opened up, it made the evening news: newly laid-off job applicants wrapped around the block hoping for a chance at even 15 hours of work per week.

I wasn’t just competing with part-time job seekers. I was competing with people who needed any work they could get to keep a roof over their families’ heads.

It was a scary, precarious time. Unemployment peaked around 10% nationwide, even greater than in the dot-com bust, with millions of workers unable to find jobs. I came within weeks of losing my housing.

This desperate climate catalyzed the fuel for the last technology cycle. Central bankers around the world slashed interest rates nearly to zero. Without traditional means generating returns on their money, pensions, institutional investors and the wealthy sought new vehicles for growth.

The Venture Capitalist had just what they needed. On behalf of these limited partners, a VC would secure a packet of investments in technology firms, one or more of which might just burst into a multi-billion dollar valuation.

They called these “unicorns.”

Cheap scale

Software solves a problem through automation.

When you build a spreadsheet to solve your reporting needs, you’ve just made yourself some software. It’s got decent leverage: instead of manually filling out a chart, some formulas summarize your data and nicely format it, saving you a few hours a week.

You make a trade for this leverage. Making the report manually maybe costs you an hour a week. Building automation into the spreadsheet might take you much longer: a couple of afternoons. But within a few months, the leverage has paid for itself in time savings. Within a year, you’re ahead.

This simple example captures the economic value of software, and its potential scales well beyond an individual’s problem at the office. Whatever costs you sink in creating software, you get many multiples back in whatever work it accomplishes.

You can make office productivity tasks more efficient, you can build social interaction that promotes a particular style of engagement, you can deliver entertainment through games or other media, and you can even make cars show up in particular places where people want them.

Any task that software can do once, it can again as many more times as you want. On paper, this happens at near-zero marginal cost.

In practice, what this means is complicated. As Twitter’s Fail Whale demonstrated, just because scale is cheap doesn’t make it free. Architecting and re-architecting software systems to be resilient and responsive requires specialist knowledge and significant investment as your customer base transitions from one order of magnitude to the next.

Copying code costs almost nothing. Running code on the web, and keeping it running, is more expensive. It requires infrastructure to execute the code, and it requires human labor to troubleshoot and maintain it in flight.

The bargain of Venture Capital

This is where the Venture Capitalist comes in.

Writing code is expensive: it needs a full-time team of specialist engineers who know their way around the required technologies. The engineers also need to know what to build, so other specialists, like designers and strategists, are needed to document and refine the goals and end-product.

The code continues to be expensive once it’s written. Again, information is physical: it lives somewhere. The data that software handles has a physical location in a database and has to be moved to and from the user, who may be quite far away from the server they’re interacting with.

Thus, more specialist labor: engineers who can design technical infrastructure that gracefully avoids traffic jams as a web-based service becomes popular. Of course, there are more specialists after that. People to tell the story of the software, people to sell it, people to interact with the customers and users who are struggling to use it.

Code needs humans. It can’t be born without them, it needs them to grow, and it can’t stay running forever in their absence. Automation is powerful, but it’s never perfect, and it exists in a human context.

Humans are expensive. We need shelter, food, rest, recreation, partnership, indulgences, adventure… The list of human needs is vast, and like it or not, we meet those needs with time and money.

So the Venture Capitalist buys a chunk of a growing software firm, injecting it with money in trade. The money can be used more or less freely, but much of it goes into paying for humans.

Despite plunging tens or hundreds of millions of dollars into a firm, if the VC does their job right, they come out ahead. Software that can solve problems effectively, that captures a customer base that will pay for those problems to be solved, can produce billions of dollars in value. So long as the lifetime value of a customer is meaningfully greater than what it costs to earn them, you’ve got a powerful business.

So spending a few million to support an infant firm can turn into orders of magnitude more money down the road.

Not every startup achieves this success. These are outlier results. The majority will fail to achieve the scale and value needed for a win, in terms of venture expectations. But when the outliers are that lucrative, the overall model can work.

So long as you pick an outlier.

Going whackadoodle

GitHub was unique. Founded in 2008, GitHub was profitable almost immediately, without needing any venture funding. They are perhaps the ultimate modern success story in software tools:

GitHub simultaneously defined and seized the central turf of developer collaboration.

Their strategy was devastatingly effective. Git was an emerging technology for version control, which every software project large and small needs. Version control maintains a history of project progress—who changed what, on which date, and why—while allowing developers travel back in time to a previous state in the project.

Historically, developers needed an internet connection to interact with the history of a project, so they could connect with a server that stored all of the project data. Git’s great innovation was to decentralize this process, letting developers work in isolation and and integrate their changes with a server at their leisure.

Originally developed to serve the needs of the Linux kernel project, one of the largest collaborations in open source history, git had both enormous power and daunting complexity. Still, git would be the future not just of open source, but of software development altogether.

Seeing this, GitHub’s founders built a service that provided:

• A central server to host git-based projects, called repositories
• Convenient user interfaces to browse, change and duplicate repository content
• Collaboration tools to discuss, improve and ship code with both teammates and strangers

On the strength of this developer experience, GitHub fast became the most popular way to build software, no matter the size of your team or your goals.

Its power and value earned it leverage. Four years into its life, it raised money on wildly favorable terms. It was a safe bet: enormous traction, plenty of users, massive goodwill, and a whole enterprise market waiting to be conquered.

Then its cultural debts caught up with it.

There’s a situation here

GitHub built a replica of the Oval Office. They also built a replica of the White House Situation Room. Then there was the open bar, offering self-serve alcohol at any hour of the day. The rest of the space was of the typically fashionable vibes you’d expect of a startup, but these indulgences caused onlookers to raise their eyebrows.

Many clowned GitHub for the White House stuff—myself included—but here is a place where I will now defend their intent. America is lots of things, and primarily an empire. The symbols of an imperialist state have a lot of complicated (that is, bloody) history and it’s a peculiar sort of hubris to adopt them whole-cloth.

But America, in its most aspirational of modes, sees itself as a fabric for creating prosperity and shared growth. Having spent a couple years there, I really think that’s where GitHub was coming from in its office cosplay decisions.

They wanted to be a fertile landscape for shared prosperity, and picked up the most garish possible symbols of that aspiration. It’s not how I would have come at it, but I get it how they landed there. It’s impossible to imagine the last chapter of software without GitHub as a productive substrate for all the necessary collaboration.

This is a tidy microcosm for the cycle. Incredible ambition and optimism, tempered by social challenges few founding teams were equipped to address. An industry whose reach often exceeded its grasp.

GitHub’s blindspots, coupled with its incredible early power, would conspire to make it both a distressed asset and a notorious place to work, claiming one of its founders in the process.

The mess in its workplace culture made it challenging for the company to recruit both the experienced leadership and specialist labor needed to get it to the next phase of its evolution.

GitHub did the work to patch itself up, recruiting the team it needed to get back to shipping and evolve into a mature company. With the bleeding stopped and stability restored, Microsoft purchased the company for—you guessed it—billions of dollars.

This stuff was not unique to GitHub

Zenefits had people fucking in stairwells, Uber spent $25 million on a Vegas bacchanal, and WeWork… they did a whole a docudrama about WeWork.

With nowhere else for money to go, successful founders had a lot of leverage to spend as they pleased. VCs needed to compete for the best deals—not everything would be a unicorn, after all—and so the “founder-friendliness” of a firm was a central consideration of investor reputation. No founder wanted the money guys fiddling around and vetoing their vision and VCs feared losing out on the best deals.

So investors kept a light touch on the day-to-day management of firms, so long as they were pursuing a strategy that kept hitting their growth metrics.

Meanwhile, you’ll notice I keep emphasizing specialist labor. Some problems in technology are particularly challenging, with a limited number of practitioners able to make meaningful traction on them. Thus, startups had to work hard to recruit their workforce. Founders would argue here, with a certain degree of legitimacy, that differentiating their experience of work through generous perks was an essential lever in winning the competition for limited talent.

Thus was the frenzy for hyper-growth that continued for more than a decade. Money pumped into the system, searching for its unicorns. Founders with more technical acumen than management experience tried to recruit the teams they needed. Investors kept the show running, hoping to grow money for their limited partners.

But this was mere prelude to the wackiest money of all.

Crypto craze

The way venture historically worked:

• Invest in a young company
• Support it through relationships and more cash, if its progress was promising
• Wait around for quite awhile for it to reach maturity
• Get paid upon a “liquidity event”—when the company goes public (lots of work) or is acquired (a bit less work)

Cryptocurrency, non-fungible tokens, and other so-called web3 projects were an irresistible optimization on the normal cycle. Going public required stability, clean accounting, a proven record of growth, and a favorable long-term outlook for a company’s overall niche. It required stringent regulatory compliance, as a company transformed from a privately-held asset into a tradable component of the public markets.

Keeping a financial system healthy takes work, after all, and bodies like the SEC exist to ensure obviously-toxic actors don’t shit in the pool.

Less arduous was an acquisition. Sometimes these were small, merely breaking even on an investor’s cash. Still, multi-billion dollar acquisitions like Figma and GitHub do happen. To integrate into a public company like Microsoft or Adobe, a startup needs to get its house in order lots of ways. Accounting, information security, sustainable growth—it takes time to get a company into a level of fitness that earns the best price.

Cryptocurrency investment promised to short-circuit all of this.

Investors could buy “tokens” as a vehicle for investing in a company. Later, when those tokens became openly tradable—through a mostly unregulated, highly-speculative market—investors could cash in when they decided the time was right.

Rather than waiting seven or ten years for a company to reach maturity, investors in crypto firms could see liquidity within half that time.

With cloud services full of incumbents, while mobile matured, it was hoped that cryptocurrencies and web3 would represent an all new paradigm, starting a new cycle of investment and innovation.

More than a decade since the 2008 crisis, central bankers had to pare interest rates back down to zero once again, this time in response to the Covid-19 pandemic. This created a fresh rush of money once again in search of growth, and cryptocurrency-adjacent companies were happy to lap it up.

What comes next

Of course, we know how this ends. Covid-era inflation soared, and central bankers cranked rates back up to heights not seen since before 2008.

Crypto came crashing back to earth, NFTs became a tax loss harvesting strategy, and now we muse about what else was a “zero interest rate phenomenon.” I’d be shocked if crypto was entirely dead, but for the moment, its speculative rush has been arrested.

In the next and final installment, I want to take stock of what we can learn from all this, and what comes next.