Marathon runners dread The Wall: for at some point 20 or so miles into a race, they often hit this all-too-real point of physical collapse. However, though it used to be thought that The Wall was a purely peripheral kind of fatigue (i.e. your muscles giving way), a body of modern research has shown that central nervous system fatigue could well be an even larger factor. That is, when your central nervous system notices that your brain and heart aren’t getting what they need, it shuts down the muscles to try to protect you from yourself – basically, to keep you alive. If you want to avoid The Wall, you move into the realms of what is known as ‘race management’ – mental strategies, pacing strategies, sometimes even focusing a kind of anger at a competitor to try to use a burst of adrenaline to override your nervous system’s homeostatic control.
Of course, much the same thing goes for startups.
You’re tired, you’re working late (yet again), your partner/spouse/children/dog are all unhappy, you have demos you desperately need to give last week, that carbon-like smell in the air is money burning… but despite all your best efforts, nothing is working. This is the Startup Wall. If it all sounds desperately familiar (but you’re managing not to run down the street at the very thought of it), then startups could well be your spiritual home. Yes, we startup people dread The Wall but, slightly perversely, we also love it – we have to, it’s part of the deal. In fact, it’s as central to the startup world as the water cooler is to the open plan corporate office.
I’ve just been banging up against a particular Startup Wall for a few weeks: wearing my software engineer hat (proper startup people don’t have job titles, they have a wardrobe full of hats to wear to get through each day), I’ve been trying – and failing – to get Linux up and running on Nanodome’s first camera. My bootloader was working fine; and I was using the latest version of Linux for a Samsung processor that has been stable for several months. But it was dying silently, apparently just after decompressing the kernel: there weren’t even any useful debugging messages coming out. You’re staring at something that’s broken, but you have no obvious tools to fix it with: pretty much a worst case scenario.
After a long, long period of floundering followed by an equally long period of grindingly painstaking checking, it turned out that the version of Linux I was working with – 126.96.36.199 – just happened to have a head.S file (the first file that gets called) that was broken for ARMv7 architecture processors when turning on the caches: upgrading to 2.6.37 fixed it first time. Even though I try to track the ARM Linux mailing list traffic for warning of anything as bad as that, I really didn’t see that one coming at all.
The wobbly moral of the story is simply this: don’t believe the VC hype that makes it look like startups are a sprint, to see how fast you can spend their soft money. Rather, they are much more like a marathon, carefully managing your (scant) money, resources and efforts to get you through the many Startup Walls you’ll run into along the way. Honestly, startup management is much more like marathon runners’ “race management” than you might think: though running out of money (peripheral fatigue) is terrible, running out of drive, determination and ideas (central nervous system fatigue) is arguably even worse. That’s when your race is really over.