Ask YC: are you a UK-based hacker, or working on a startup in the UK?

Clickpass not Sitepass! :P

Huh, I hadn't seen Songkick before, it looks great! I had the same frustration with finding shows from bands I like, and wanted to make something similar. I'm sure it'll be a hit if you get the data and some half-decent recommendations.

Thanks man! Hope to meet you in person soon.

What do you think of our recommendations right now? We're taking a different approach to the 2 most popular paradigms: collaborative filtering on your userbase (e.g. Last.fm) or expert hand labeling (e.g. Pandora).

Instead we consider any implicit or explicit relationship between two artists expressed on the web as a data point for our recommendation engine to analyse. This could for example be that Arcade Fire and LCD Soundsystem were mentioned in the same review. We also incorporate data from our users' listening histories. You can see the recommendations at work on artist pages and on our Band Manager.

Our goal is to generate recommendations that harness the wisdom of the biggest crowd possible: all music lovers anywhere on the web. We're not there yet but would love to hear what you think of the first results.

I had a look around the artists pages; the recommendations are good, each set seems fairly solid with mostly nothing too irrelevant, the main associations I would expect are generally there, the results seem fairly consistent across genres, and pretty good for niche/non-mainstream bands too. If this is all hand-rolled from web crawling and that sort of thing, rather impressive!

A couple of thoughts/ideas:

It'd be good to have more recommendations for bands who have dates coming up. As someone wanting to see shows, I'd generally like to see at least a couple of recommendations for specific tours, as long as they're fairly relevant. Though I imagine the main issue with this is tour data, rather than recommending. Also, If I'm looking for touring bands, it doesn't seem quite right to get several recommendations for bands I know I'll never see, like Nirvana and Biggie Smalls, so maybe you could provide separate result sets for bands that are touring and bands that aren't.

Another suggestion is to have a little more randomness, or a little more tolerance for less-close matches, or different types of closeness. The results tend toward big names/popular bands, but it may be good to throw in more that are more closely related by genre/style, and less by prominence. You may even consider randomly removing some of the biggest bands from results, otherwise you end with e.g. Metallica recommended for every band when browsing around any vaguely rocky bands.

There are a couple of cases where it seems maybe you've sucked in some related but irrelevant data, and end up giving word-concept recommendations, e.g. for Underworld I got [Gods, Hades, Orpheus, Doom, Demon, Witch, Terror, Curse, Warrior, Dawn]

Of course all this depends on the user and context. A disclaimer then: I'm a huge music fan, but my comments are probably coloured by my (generally, somewhat) less mainstream tastes. Sorry to go on so much, but I'm very interested in both music and recommendations, so I'm finding it hard to stop myself. But good job, it looks to be a very useful site that I would use. I'd very much like to hear more about it, and what you plan on doing with it when this meetup happens.

I am in Devon hacking away at my ideas. Doing client work to pay the bills at the moment.