Clay Tech

"clay-works make things real"

translated from clazytech.com

Don't fear failure, hurry up and fail

From Astro Teller’s SXSW talk at Google[x].

There was an interesting moment when he talked about the project to generate wind power by flying kites.

Ordinary wind power requires too much equipment and too much money. Going higher makes it easier to increase power generation, but of course the relative cost goes up too. Once they’d overcome that problem and got the prototype working reasonably well, Astro reported it to Larry Page. Larry’s response was supposedly, “Go build five more of those right now and test them until all of them break.”

The point was that breaking things speeds up learning.

This carries an extremely realistic implication.

You calculate the strength before you build a bridge. Of course you calculate it. There’s surely no exception to that. But in practice the next step is to build a miniature and test it — that is, break it. With bridges there’s plenty of precedent, and the theory is more or less settled. Any trouble nobody could have imagined has probably already surfaced, too. Even so, everyone accepts that you test before you build the real thing.

So what if the premise is “something nobody has ever tried before”? How much can you really trust theoretical values, simulations, and assumed cases? Working through them one by one, knocking each down, is certainly one facet of engineering, but it’s far too slow.

What’s the quickest solution?

Just go and do it, and fail.

You can learn an enormous amount from failure. Every engineer knows this. A bug that slips in right before production start, exception handling you worried about but left alone, inadequate heat-resistance evaluation, a use case outside the assumed spec, and so on. Failure teaches you a lot of “things you must not do” and the “typical solutions” for them, and that accumulation eventually sublimates into the irreplaceable asset called “instinct for where the trouble is.” How many years, how many products, does it take to get there otherwise?

So — forget it, just go fail already!

This can also be seen as an extremely logical way of thinking.

It’s often said, in various forms of argumentation, that proving 100% nonexistence is almost always impossible. Enumerating and verifying every exceptional case is extraordinarily difficult. In other words, “Show me that no counterexample to XX exists!” is in a sense a trap, a line typically wielded in discussions loaded with malice and aggression, an irrational, unproductive thing to say.

Something similar can be said about development too. It’s impossible to guarantee 100% that failure can’t happen. Pursuing that guarantee to its logical end is about as thorough a waste of time as there is. On the other hand, everyone also gets uneasy if all they’re told is, “Theoretically this should work, so, well… let’s just go for it!” “Theoretically” and “ideally” are, in most cases, weak against exceptional cases, and in practice there are plenty of pitfalls — people know this intuitively. What usually props you up at times like that is “experience” and “track record.” “It should be fine if we leave it to so-and-so.” “Given his track record, the risk is probably relatively low.” That sort of thing.

But then — what if the subject is an entirely new domain, where there are no true experts and nobody has ever succeeded before? That’s the same question as before. In that case, it seems like a perfectly reasonable approach to go ahead and fall into the pitfalls a few times first, then build the development process like this: “There’s a fairly serious pitfall here, a few small ones over there, and traces of what looks like a pitfall beyond that — so I plan to proceed along this particular route.” Of course, this means the exceptional cases haven’t been fully enumerated at all, and depending on the case, theoretical verification might be insufficient too. But once the notable pitfalls have been found, and the developers have learned how to recover after falling into them, or how to avoid them altogether, their ability to resolve new trouble when it arises — and even their ability to sense trouble coming — will be dramatically higher.

The result is that you end up building the best thing in the shortest time.

At first glance it looks reckless, like a crude argument. Normally nobody gets a budget approved by saying “we’re going to break things.”

But that’s exactly what becomes “how a newcomer beats the veterans,” “how you reach uncharted territory at the fastest possible speed,” and “how you create something out of nothing.”

This might be a relatively unfamiliar approach for a delicate people like the Japanese. But think about it: no child starts out “studying” how to use Lego. Yet after a bit of trial and error, they’re quickly able to build things. And the people who learned things this way go on to effortlessly surpass adult ways of thinking. Everybody knows this very, very well.

Finally, what Astro Teller said was necessary “to make the world not 10% more convenient, but 10 times or 100 times more convenient”:

“Creative, Productive and Failure”


Originally published in Japanese at https://clazytech.com/2015/05/97/. Translated with LLM assistance and reviewed before publication.