My Y Combinator interview.

This applies equally for being talked out of good ideas.

And if your idea is good to start with, then there would be no point in changing it until you've pursued it long enough to see where it leads.

The thing to remember about YC is that it becomes a sort of self-fulfilling prophecy. For example, let's say that Cal State Chico (not a top school) somehow convinced all the top students to attend next year. Pretty soon, it would start to be known as a top school. And let's say Stanford only got applications from the bottom 5%... you see where this is going.

Paul Graham's evangelizing means that, regardless of the utility of Y Combinator, if he convinces the top start-up prospects to join the program, it will make the program look good. They are looking for people who would be successful even without them.

Y Combinator could be minorly helpful and they would go on to success; it could be a wash, and they could go on to success; it could be a slight negative, and the strongest candidates would mostly still go onto success.

People could also FEEL that it's an incredible experience, even if this isn't objectively the case. Many people join various kinds of self-help or self-empowerment groups and enjoy the events and believe in them strongly, even if there is demonstrable practical harm.

Y Combinator could also have a net negative effect on the startup landscape, because NOT getting in can be so disappointing that it can lead people to give up (I wrote that before your post, but the link expired; I reloaded, and made this a reply instead of top-level comment); in other cases, they've found people who applied and got in, who considered getting in to be the accomplishment. Paul's viewpoints have been increasingly dominating the national conversation on startups the last few years, so the hypothetical indictments are meant to address its success as propaganda. Obviously if it really WERE the best approach, criticizing it for being too successful in its results would be disingenuous.

There's also a huge emphasis on getting funded being an accomplishment, which is a big distraction from the real priority for a business, which is making something that makes a profit. (A nice big funding round also means now you need an even BIGGER exit. This can actually reduce your chances for personal success. Xobni couldn't sell to Microsoft for $20 million because it wouldn't have been a big enough ROI for its investors. Perhaps they really thought they could be a billion dollar company and didn't want to sell anyway, but you see the point.)

Paul's talk about angels and super-angels and valuations focused on the main point being "the percent chance that the start-up is Google," if I recall correctly.

To me it looks like Y Combinator has the wrong model for that. The resources and timeframe favor much smaller ideas. According to the unofficial YClist.com, the top exit was 280 North at $20 million, which means not only have they not had another Google, they haven't yet had another ViaWeb.

It would be interesting to find a list of angels and their investments for the last 5 years and see which ones had bigger deals than that. I'm guessing a lot. So, that covers instincts and judgment, at least so far.

As for changing your idea, there are market segments that have been goldmines which I have yet to see a single YC company delve into. Thus, I'm not so sure on the advice portion.

There is a pretty big generation gap with the hot web properties and I don't know if any of the YC principals really grok it.

If PG reads this I expect to be told I'm wrong on every point except perhaps for a token concession for decorum. It's hard to talk people out of bad ideas when they're stubborn ;)