Skip to content

Astro Teller on planning, experimentation and innovation

This extract is from another great interview on Exponential View with Azeem Azhar.

Depending on how you count, [at Google X] maybe 1000 ideas turned into ten ideas.

I would put it this way, with respect to how we think about X: if you’re serious about innovation, you would be lucky to succeed one percent of the time. You don’t get a choice about that – that’s just how life is. The question is, how efficiently are we going to discover and throw out the 99% that eventually aren’t going to work out. That’s actually where the efficiency is. You can’t pre-decide which ones are the one percent. You can’t know ahead of time or you wouldn’t waste all that energy. Nobody’s that smart. But if we’re somewhat more intellectually honest we can find Achilles’ heels in our own projects faster. We’re constantly trying to innovate but weirdly what we’re innovating on is how wisely we can turn off the 99 percent. The one percent that’s left over? That’s a side effect of our process. It’s the efficiency of discovering which 99 percent won’t and shouldn’t make it that’s actually where we spend all our energy.

… [Our staff] all know that you could produce value much faster doing what I’ve been describing here. [The problem organisations face] It’s that no-one actually lives up to “Yes, we will celebrate you for the quality of the experiment that you ran rather than just whether you got a yes or a no.” Because as soon as you reward people on the outcome rather than on the quality of the experiment, innovation is dead.

We get a lot of people – world class people and particularly world class engineers – who are pretty dug in on the idea, “You build it right the first time.” And they have these sayings like “Measure twice and cut once.” And I get that. But nobody seems to have explained to these people that that is a really good attitude when where you’re going is well understood. But it is mal-fit to purpose when you don’t know where you’re going, and it’s super expensive and inefficient to have the hubris to think that you pre-know the answer and can design it right the first time. So it doesn’t mean that they’re completely wrong, but they’re wrong in this context. … whether this is a good habit is context dependent, and I just wish that people had more of a sense that you have to adapt to the context you’re in.

Astro Teller on Exponential View

I'd love to hear your thoughts and recommended resources...