I had many opportunities at Intel to go to my boss and say I've been asked to do this feature by marketing for perfectly good reasons, but after we evaluated this technically it's my conclusion that you’re asking to inject too much risk into this project. Or the return on that investment isn’t high enough to warrant that much risk. And so I'm going to argue against it and the marketing guys are going to scream and call us idiots and say you didn't understand what we said. I don't know how many times I prevailed on that, but at least I had the terms to couch it in.

Looking back at it I think we did a decent job overall of hitting those compromises. Those were the hardest ones. It's easy to go to your boss and say they ask for this, we looked at it and we'll put it in. And it isn't going to cost you anything. Who wouldn’t love that? If you say, well, we're going to put it in but it's going to cost you 3 percent of the die, now all the sudden it's got a real price tag and the boss goes, "oh, I don't know. This hurts. Can you please make it 0?". Er, no. We often had those discussions. Who wouldn't want something for free? Anytime the discussion hinged around risk it got awkward really fast. Because engineers are quantifiers. Tell me a percentage then I can deal with it but if you just say I want to stick a feature in it that will make it this much faster, you can quantify that, it will cost this much in schedule, you can quantify that. It will use this much die size, I can quantify that. I get to sit and say, well wait a minute, it's going to be awfully risky in terms of complexity. There might be some bugs. I don't want to do it. You can quantify your answer. I can't quantify mine. Now how do I trade those off?

Those are very difficult conversations to have. I was lucky that people like Albert Yu, my executive VP at Intel, would sometimes say look, I don't completely understand the argument you're making, for good reasons because you're basing them on things that inside the chip that I can't possibly know, and you're comparing other things that I do know and I know the price tag of. But in a sense you're asking me to trust your intuition, because I can't do it myself. I can't do it based on what I know. I said, I agree with you. That's the right way to view this. But, if the day comes that you can't trust my intuition as your chief architect you need to fire me and find somebody else. Because I don't think there's another way to do this.