All actually good startup ideas are bad ideas when they are started. Everybody thought twitter was retarded, and nobody had any idea social networking was even going to be a thing when facebook launched. If you only work on things that people agree seem like a good idea, you are probably the 5th team that is working on it, and you probably started about 3 years too late.

This actually ends up giving many real successful startup ideas or just good ideas. It also gives a lot of bad ones, but here are some I noticed:

OKCupid, for music. (Pandora) Github, but for knowledge. (Wikipedia) Hacker News, but for knowledge (Stack Overflow, Stack Exchange) FourSquare, but for photography (is that Instagram?) and a bunch that are more of a stretch.

So, here are some ideas it had that I think are are actually good:

OKCupid, but for shopping (better, hyper personalized product recommendations).

LinkedIn, but for beer (more like "it's Just lunch" for professional relationships, set up one on one beers and such, foster mentor relationships, etc).

Github, but for fashion (imagine something like paper dolls, where users can mix and match shirts, pants, accessories and share the combinations with each other, and then purchase complete outfits).