Understanding the customer, feeling what they are feeling, seeing what competitors are doing, you end up having a richer sense of the marketplace.

Artificial intelligence, machine learning... cell therapy, immunotherapy. There's just a constant stream of investment ideas we could pursue better in that fashion.

First and foremost, localization is a customer strategy, it's not a cost arbitrage or whatever.

Going back to the past is not productive.

It's a wonderfully democratic method, publishing a text on the Internet.

The Internet promises to open new channels for worker-firm communications. What are the consequences of this opening?

I've been writing a lot more, I believe, because of the Internet. I've been posting stuff that I've written and I've just been writing.

I don't understand computers. I've been unable to construct a working mental model of how they do what they do. I can break software by looking at it. I can blow anything up. Without trying. It's sort of like being a dowser. And this extreme elaborate clumsiness on my part is actually something people will pay me for. It's quite wonderful.