A classic Microsoft interview question was: "Why is a manhole cover round?" The right answer, of course, is, "Who cares? Are we in the manhole business?" (Followed by twisting in your chair to look all around, getting up, and leaving.)

Andreessen's critique of hiring purely for intelligence via logic puzzles.

Drive is self-motivation—people who will walk right through brick walls, on their own power, without having to be asked. Motivating people who are fundamentally unmotivated is not easy. But motivating people who are self-motivated is wind at your back.

Look for what someone has actually done—not been involved in, part of, or hanging around for. The business they started in high school, the nonprofit in college, the open source project they led. If a candidate has just followed the rules their whole life—right classes, right tests, right career opportunities—they probably aren't driven, and you're not going to change them.

Curiosity is a proxy for: do you love what you do? Sit a programmer candidate down and ask them about the ten most interesting things happening in Internet software. If they love the field, they'll have informed opinions. If they don't, their day job isn't keeping them current—and the kind of person you want wouldn't have tolerated that for long.

Test for honesty by asking increasingly esoteric questions until the candidate doesn't know the answer. They'll either say "I don't know" or try to bullshit you. If they bullshit you during the hiring process, they'll bullshit you once they're onboard.

Beware people from hugely successful companies—many were just along for the ride. The old IBM rule: never hire someone straight out of IBM. Let them go somewhere else and fail first, then once they've realized the real world isn't like IBM, hire them.

Little tells in interviews blow up into disasters on the job: never laughs (hard to get along with), constantly interrupts (egomaniac), claims to be friends with someone you know but doesn't know what they're doing now (bullshitter), nonlinear answers to simple questions (disorganized on the job).

Even with a scrupulous process, expect a 70% success rate on individual contributors and 50% on executives. Most startups fire too slowly, not too fast—and your great people are watching, happy when you keep the bar high. You're usually doing the person a favor by releasing them to find a role where they can actually star.