Show HN: We made a spam-free lyric search engine

STOP CENSORING MY COMMENTS YOU FAGGOT ASS PISS NAZI HITLER JEW-HATING COCKSUCKER

Back in 1997 when I and three others ran a record label called Jeepster, we had a few great bands on our label: Belle & Sebastian, Salako and Snow Patrol. (OK, Snow Patrol were less than great then, and perhaps now.)

One of the things I figured I should make was a web site in which fans were able to communicate with each other, the theory was simply that:

1) If fans kept the interest high between releases they'd be there still when we came with a new release

2) If someone was looking for a lyric or something they'd heard and we helped them find it, we might convert a stray passer-by into a fan

So I built something that you'd now call a crowd-sourcing site.

For every track, people could submit:

* Lyrics

* Mis-hearings

* Personal memories the lyrics evoke

* Interpretations/Understanding of the Lyrics

* Tablature for different instruments

All of which were votable so that the best tablature rose to the top, but others (perhaps harder ones) were still available.

It was highly compelling, you came for the lyrics but stayed for the personal memories and insight.

As a label we felt it was hugely successful. It definitely paid back in keeping the interest of fans, increasing that interest and creating new fans.

It worked so well that NME nominated us for best web site in their awards ceremony in early 2000 because they'd never seen a site in which the content was mostly created by fans and that the fans loved.

I left the label in 2000 to do other things. The system ticked over for a while, but the label never replaced me with another technically inclined person and when they wanted to change the styling of the song system they broke the PERL script and left it broke before taking it offline. Such a shame really.

Since then, whenever I look up lyrics, I've wondered why there isn't a comprehensive crowd-sourcing of all of the information around a song in the same vein that I did it, but including things like when it's been performed live or recorded in radio sessions. A real database about a song, what it means, who's covered it, how to play it, what it means.

The business model would be to sell stuff related to the song, affiliate fees for MP3 downloads and CD sales, affiliate link for the purchase of the official tablature books. Keep it simple, and keep adverts off of it. A wikipedia solely for songs and everything about a song.

The cost per transaction and storage is incredibly low as it's just text. And you could probably negotiate with publishing companies as right now they don't even get credit for this stuff let alone increased opportunity to sell what they have.

The PERL script in 1997 wasn't more than a simple thing that used flat files as a database for the 200+ tracks we'd released to then. Nowadays, for the complexity we're talking about, a weekend project in Django would suffice as a first pass. The hardest thing is just talking to the publishers, but truly, it's not as hard as some people imagine, there are fewer publishers than there are record labels.

Minimalism is nice, but take a look how for example Facebook shows a link to this site. Too minimalist. (This of course does not matter if you don't want the link to be shared).

It doesn't have to be complex. I get around this simply by re-submitting if the hashtag is present: http://blurk.org/#tv-the+simpsons

Similar to your site in terms of "not storing, just search engine". I've put the javascript snippet for it at the bottom of the page source.

You should put a link to the original source. Even if their website is full of ads they deserve credit for the lyrics. Also in case of mistakes we have somewhere to report to.