In fact, Sze said the firm conducted a study of partners’ voting patterns and decided to change the process by which it decides on on investments. He didn’t go into too many details about those changes changed, but apparently the team observed that when everyone thought backing a company was a bad idea, it was indeed a bad idea. More surprisingly, if all the partners loved the startup, “the returns turned out to be very mediocre.” The problem, Sze suggested, is that those ideas are probably “too easy.”

The investments that resulted in the biggest returns? Those were “the ones in the middle, where there’s a lot of debate and it’s not completely clear.” In those cases, Sze and Hoffman said they’re looking for something to tip the firm in the startup’s direction, like passion and experience.

These results tied into a broader point that Hoffman made, that “almost every great investment has a contrarian check.” In other words, the best investments are the ones where “a lot of people will think that’s a bad investment but you will think that’s a good investment.” A good example is LinkedIn — when Greylock backed the professional social network, Hoffman (who co-founded the company) said it was dwarfed by other social networking sites.

Intuitively this makes sense. If everyone loves it, it invites much competition.

Whereas if very few people understand it (*cough* PandaWhale *cough*) then you can become huge before anyone has any idea what you're doing.

Still, I like this concept that venture partners don't have to all be in agreement to make an investment.

That's unusual, I believe. No wonder I think so highly of Greylock.