I try to encourage my team to be wrong at least 50% of the time. If you're right too often, it means you haven't tried things that are sufficiently interesting.

The first two hours of the day are generally hours that people have high cognitive capacity ... 9 to 11 are usually very good hours for people. And if you come to the office at 9, and you have kind of high capacity, and in the first two hours you do email and Facebook... The hours that you have high capacity are so precious. You have a few of those a day, and they're so much more productive than the hours after lunch where you can hardly function.

I think there are a lot of areas where startups benefit a lot from overconfidence ... Being overconfident about your own intuitions about human behavior, I think, is very unhealthy.

I think that information does three things: It's about accuracy, it's about helping people understand the relationship between cause and effect, and it's about motivation. And it turns out, you can't always have all three of them.

Dishonesty is all about rationalizations. It's all about the question of 'What can you do and get away with?' And not get away with from the perspective of not being caught, get away with from the perspective of you not thinking of yourself as a bad person.

There's actually really interesting research showing that when people try to fall asleep, they interpret the difficulty as an indication that something is wrong.

If somebody is asking you to do something a year from now, ask yourself whether you would do it next week.

There's this belief [with startups]: Get out early and iterate ... People do get out early, but they don't iterate until it's too late.

Nobody would ever do a startup in biology without having a biologist, but people have no problem doing a startup on behavioral change without somebody who knows anything about behavioral change.

The obesity epidemic, in general, it's a 200 calorie a day game. If you can cut 200 calories a day systematically, basically people lose a pound every two weeks.

Implementation intention is the idea that we have very fuzzy plans for life, and we make concrete plans only in very specific cases or only when we get close to the event.

I think a second big mistake is that people don't work enough on their users' mental model of their product.

I think good leaders are able to separate core assumptions of the product, and figure out what to test and what to make assumptions about.