I just read a fascinating book - Gut Feelings by Gerd Gigerenzer. It's by the guy who did most of the research that Malcolm Gladwell's Blink is based upon. The central thesis was that people have evolved heuristics for decision making that let them quickly make snap decisions more effectively than gathering full information, but a couple chapters were on an interesting corollary: in the absence of hard data on their performance, most people's decision-making heuristic falls back to "Will I be blamed for this decision?"

He provided a bunch of examples from different professions. For example, doctors' decisions often aren't based on solid evidence-based medicine (which is often contradictory), but rather on "Will I be sued if I do or don't perform this treatment?" If you want the doctor's actual opinion, you should ask "If it were your own mother, what would you recommend?" instead of "What would you recommend I do?" - the former shifts the doctor's perspective so that they're thinking "How can I provide the best care?" rather than "Will this person sue me?" A study of UK magistrates found that 92% of their bail decisions could be predicted by the following heuristic: "Did any of the prosecution, police, or previous court recommend bail?" If not, and the defendant commits a crime, it falls into the category of unforeseeable events and the magistrate can't be blamed for it.

A similar effect may be at work with VCs. It's usually impossible to know, even with hindsight, what the opportunity cost of a lost deal was. If a VC turns down the next Google, chances are nothing happens and the company just fizzles. If a VC invests and it goes bust, however, everyone knows. So it pays for the VC to invest like everyone else does: then they can blame any failure on "Well, this was completely unforeseeable: everyone else was sure they'd be a success too." It's the combination of risk aversion, self-interest, and lack of feedback that drives this. If VCs actually had solid conceptual frameworks and good data for evaluating possible opportunities, they could rely on that rather than on their peers' evaluations. (Maybe this explains Sequoia...)

Whenever I see something that doesn't quite work I can't help thinking about why it doesn't work better and how to resolve the problem. This includes how to shorten the lines without using extra personnel in my local supermarket, how YC news recommendation system could be made to work better, and how to do 3D displays. Anything really...

The keys to getting the good ideas are:

1) have lots of them - my ideas are no better that anyone elses, I just have a lot. Statistically a good one will pop up every now and then.

2) When you encounter a problem think through all the possible ways of resolving it. Don't mind if some of them are outrageous, just be sure to discard them when you have checked that they don't work. A friend and mine and I once actually thought that according to the laws of physics a certain combination of magnets and superconductors should make a perpetual motion machine. We set up an experiment, and of course it didn't work. We tried it and discarded it, and we weren't afraid of trying something crazy.

3) Know about different disciplines, and have broad knowledge. Innovation often comes from crossbreeding different disciplines. If you put a physicist and a web designer in a room you will get more innovation than if you put two physicists in a room.

Just my two cents, hope it's useful.

I didn't think that. I remember telling the powers that be at Yahoo in 1999 (I worked there then) that they ought to buy Google, and it was the only company I ever suggested they buy.

Over the years, I've found Amazon becoming more and more cluttered. I find myself becoming blind to useful new features because they are hidden in amongst blogs, plogs, video testimonials, recommendation, lions and tigers and bears.

In aerospace, there's the notion of new displays "buying their way into the cockpit", which is not about money, but about being worthwhile enough to share the valuable real estate with other displays. There is an acknowledgement that the existing displays will either get smaller or viewed less or both.

I'd be interested in hearing how he thinks about the fracturing of attention that occurs when there is so stuff on Amazon's pages. There's a real cost associated with that, and it feels like Amazon is not paying attention to it at all.

Edit: just confirmed that it works.

Basically let's you search users by id or email then give you ability to change their email/reset password.

Admin dashboard: http://img.skitch.com/20080415-ef6k9c8hpbasi9g137j6u4iqrs.jp...

Clicking "password": http://img.skitch.com/20080415-day5r87i8tbt9iaqrtf73tpnnd.jp...

Clicking "email": http://img.skitch.com/20080415-bg12j2y7wxmeqn44pmctcs69e3.jp...

And clicking "edit" tumblog: http://img.skitch.com/20080415-cg7iwrmc7gu2a58qkmcnfre6ka.jp...

Don't worry, I didn't do anything!