One way to think about a startup is as a machine that prints money. Or a machine that could print money, if you configure it well. If you run the machine just so, it will make you rich. If you run it exceptionally well, it can print enough money to make you one of the richest people in the world. If you run it badly, it will send you to jail.1 It is important to configure it well.
The machine, however, is hard to use. You can’t just buy one by sending a couple hundred dollars to Delaware and turn it on. No, you have to carefully run the machine, and the machine is complicated. It is controlled by thousands of knobs and levers, and every switch will cause the machine to print different amounts of money, or burn different amounts of money, or commit different amounts of crime. Again, it is important to configure it well.
Unfortunately, in addition to the machine having lots of knobs and levers, nothing on the machine is well labeled. You don’t really know what any of the knobs and levers do. There is a wire coming out of the back of each one, but they are all tangled together. Some of the wires might be severed, and might not be connected to the machine at all. Some of the wires conduct electrical signals very well, and the machine responds immediately when you turn the knobs connected to them. Some of the wires are full of tar, and conduct electrical currents exceptionally slowly.2 If you flip a switch connected to one of those wires, the signal may not reach the machine for weeks or months. You don’t know which wires are which. All you can do is experiment.
But even that doesn’t always work. Some wires are spliced together with other wires; switches connected to those wires will do different things depending on the configuration of the other switches. Some wires are frayed: Turn some knobs too far, and their wires become fuses. And some wires are influenced by outside forces, like the seasons, and the federal funds rate, and what’s trending on TikTok.
The switches themselves are also complicated. Some of them move on their own. Some periodically reset themselves to their old positions. Some buttons can't be pushed until other buttons are pushed; some buttons can only be pushed once. Some knobs can be turned easily, and some take a tremendous amount of effort to move.
When former founders and venture capitalists write their tell-all blogs and LinkedIn memoirs about startups, this last point is often the central theme of their narrative: That running a startup is hard. It takes intensely hard work; it requires doing intensely uncomfortable things; it demands intensely long hours; it is intensely all-consuming. Last year, Jensen Huang, the founder of Nvidia, which is currently the second-best configured machine in the world, famously said the work wasn’t worth the $35 billion he made from doing it:3
Building a company and building Nvidia turned out to have been a million times harder than I expected it to be—any of us expected it to be. And at that time, if we realized the pain and suffering, and just how vulnerable you're going to feel, and the challenges that you're going to endure, the embarrassment and the shame, and the list of all the things that go wrong, I don't think anyone would start a company. Nobody in their right mind would do it.
I’ve brought up this quote before, to make the same point that everyone makes: Work is every startup’s only enduring competitive advantage. Effort is the defining input to a startup; exhaustion is the defining feeling. It is important to configure the machine well, and the machine takes a lot of energy to configure.
It’s a simple, enticing truth—but perhaps a misleading one. These sorts of stories suggest that starting a startup is, first and foremost, about committing yourself to doing a lot of work. They suggest that the question you should ask yourself, if you’re deciding whether or not to ignore Jensen Huang’s warnings, is if you’re willing to put in the time. Will you be in the gym at 5 a.m.? Will you quickly respond to every email? Will you clean the toilets? If you’re ready for the grindset, you’re ready for the machine.
But that, I’d argue, is not its essential character. The thing that makes the magical mystery money machine a magical mystery money machine is its mystery: You don’t know how it works. It isn’t hard to operate because the switches are hard to flip—though they are—it is hard to operate because you don’t know what the switches do. And the prevailing emotion isn’t stress, from knowing how much work has to be done; it is the stress from not knowing what to do.
There will be moments—days, months, the entire lifetime of the startup—when the machine doesn’t work. Perhaps the machine once worked, and now it’s gummed itself up, printing less and less money every day. Perhaps the machine is in crisis: it is overheating, it is on fire; it is hemorrhaging fuel; it is hemorrhaging vital parts. Perhaps the machine never worked, and no matter how its configured, it lies lifeless, unable to burp out a single dollar.
To stare at these problems is to stare at a router, when all the lights are green, but your computer refuses to say anything other than something went wrong, it can’t connect to the internet, try again later. It is to stare at your broken router when your friends are over for the Super Bowl, the game is about to start, and people are here because they believe in your wifi and your leadership. It is to stare at some malfunctioning machine, simultaneously needing it to work while also knowing next to nothing about how it works. It is to stare at a problem that is urgent, vague, and fatal.
More concretely: People aren’t buying the product. Or a competitor just launched a rebrand, and they’re stealing customers. Key employees are leaving. Marketing campaigns can’t seem to bring in new leads. The new social media initiative flopped. The mobile app is chronically buggy, and new features are chronically behind schedule. The trajectory isn’t sustainable, you know it’s not sustainable, everyone knows it’s not sustainable, and the machine has to be fixed.
But…how? Though the Silicon Valley content mill will tell you how to solve these problems—for finding product-market fit, talk to more customers; for flagging employee morale, buy people pizzas; for slowing growth, up up down down left right left right B A start—these instructions are unreliable. Everyone’s money machine is different. Are the wires from these dials connected to your machine in the way the blog posts say they are? Are they connected at all? How far do you turn them, and how long do you wait to find out if they work?
This, I think, is the answer to the riddle of why we’re surprised that startups are hard. It’s not the amount of work that surprises us, but the nature of that work: Sitting in front of a mixing console worth of controls and not knowing what to do. It is important to configure the machine well, and you will have no idea how to do it.4
In hindsight, it was a machine that I was perhaps constitutionally unfit to run. If you’re given a sea of mysterious levers to pull and your response is to cautiously analyze the situation—if you do things like, say, I don’t know, repeatedly write thousands of words about niche software trends—you will compact yourself under a Shepard tone of relentless pressure.5
Though that one probably should’ve been obvious, what’s less obvious is that the machine also can’t be run by people who are prone to perpetual action. Button mashing in response to problems—and there will always be problems—rarely fixes the machine, and tells you nothing about it actually works.6 The machine will change, but based on what? Was it because two switches were flipped at once? Has the signal from the knob you turned six months ago finally passed through its long wire? Button mashers will have done something, but they still won’t know what to do.
Instead, the people who are best able to run the machine are probably psychopaths, for better or for worse. They are people who are convinced that they know how the machine works. They don’t look at a bunch of levers and wonder what to do, nor do they try to experiment with every switch. They look at the dials and just decide—based on intuition? Delusion?—how to turn each one. To people of this sort, the machine isn’t stressful, because to them, the levers are labeled.
That, of course, blows up a lot of machines; you are unlikely to simply guess the right way to configure it. But both indecision and erratic action blow up a lot of machines too—more, probably, because at least guessing a plan and sticking to it could work. Plus, if you’re going to commit to running the machine for a while, you can save yourself a lot of anxiety by picking a set of switches and doing whatever it takes to flip those. It is important to configure the machine well, and you might as well tell yourself that you know how to do it.
I don’t think this is how electricity works? Electricity and magnetism, man.
Someone should ask him if it was worth $108 billion. Which, on one hand, how could his answer have changed? How could anyone say, “All that work wasn’t worth Pizza Hut, Taco Bell, and KFC, but it was worth UPS?” How could he possibly comprehend that difference? On the other hand, how could an extra $70 billion not change his answer? How could $70 billion not matter? Big numbers, man.
This is why startups are different from more mature companies. The older a company is, the more it has untangled some of its wires. It’s figured out what some of its controls do, and which ones to pull. For GE, the layoff lever is pretty well understood. For a startup that’s never fired anyone, what will happen after it lays people off for the first time? Who knows.
In a way, data is how people like this attempt to cope with this problem. We analyze the numbers, and hope that it reveals how the machine works. But it doesn’t always work.
Shoutout to Tony Hawk’s Pro Skater though, which is the one place I very distinctly remember button mashing working exceptionally well. (I think it was because of combos? You got huge combo multipliers for tricks, and button mashing caused you to do a whole bunch of tiny tricks. Which, somewhat ironically, scored far higher than doing something like a straight 900.)
Don't forget that in addition to all that, it also matters where you place the machine. Put it in the freezing cold, and no matter how much you tinker with it you won't get what you are hoping for.
Put it in just the right environment, and it's possible it'll work really well with less tinkering.
Loved the write up .