From Founders At Work:
No, it was never clear that we were on to something huge. You never know anything. The hardest part in a startup is that you wake up one morning, and you feel great about the day, and you think, “We’re kicking ass.” And then you wake up the next morning, and you think “We’re dead.” And literally nothing’s changed.
When you’re working on a startup, you’re in the trenches, the front line of a whole market. It’s the “bare metal programming” of business. No HR, no meetings, just product.
But when you’ve got your head down because you’re working on product, you have no information. Are you doing well? Doing poorly? Solving a problem one person has? Solving a problem one million people have? Doesn’t matter because we have this fire to put out, we have to ship, we have to launch.
And so basically the signal that you operate on–the only signal that you have–is the information that somehow blasts its way through the front door of your office building. And only the strongest signals reach you. Acquisition offers or customer complaints. Runway time. Occasionally press coverage, if you can get it. That’s it. So if you get an aquisition offer, great week! If a customer complains, bad day! But 90% of the time, neither one. Does your product make a difference? What’s happening with the people who almost bought it but didn’t? Good luck with that.
If you’re a big company, you have the luxury of finding out. You can do focus groups, hire some survey companies, and figure out whether you really matter. But if you’re small, everybody has to be building the product.
Developers hate meetings, because they’re time spent away from “real work”. But I think just a tiny bit of this hatred might be misplaced. Meetings are where you find out if what you’re doing actually matters to anyone. If you’re actually doing real work 100% of the time, you would never find that out, because you’re too busy working.
I think a lot of the “bipolarness” of running a startup is because we’re all operating on bad signal. The signal can be biased away from the true average either positively or negatively, it can have excessive variance, and it can have an awfully low sampling size. It’s almost worthless.
Part of the paradox is that (good) developers are really highly trained people and do things that 99% of the population simply can’t do. But 99% of the population doesn’t have skill X either, for many worthless values of X . Is this specific application of my development skill the best? Let me count the bug reports today and get back to you.