User:Andy/Idea/Internet music chart

The currently available music charts do not reflect what people are actually listening to. Even if you ignore the fact that the UK Singles Chart is based only on singles sales and not by singles actually listened to, the biggest problem is that more and more people are downloading their music.

Without getting into the argument about whether sharing copyrighted media is acceptable or not, it would still be interesting if we could somehow measure what people actually listen to.

What to do

So let's make plugins for Winamp, XMMS, Windows Media Player and other common music players, that voluntarily reports what people are listening to, to a central server, which then makes a website out of it.

Problems

People don't get track metadata correct

We can make the plugin strip off the metadata and then submit a SHA1 hash of the rest of the file, storing both in our database. Then we'd simply tally the counts for the hash and report the corresponding metadata. We could have a team of volunteer editors who manually merge results as and when they see multiple copies of roughly the same file.

You will never get support from the music industry for a chart that seemingly supports illegal behaviour

We don't have to give support to illegal behaviour, we can just accept that it happens. Who cares what the music industry thinks? It is only an interesting concept, not a serious competitor to Gallup.

Why bother?

Interesting challenge. But it could be too hard (expensive) to do properly, admittedly.

There's already CDDB

This isn't a database of music, it's a chart of what is actually being listened to.

What if I'm skipping through all my tracks? Will it vote for each even though I didn't listen properly?

It could be based on total listening time as a percentage of track length; a 4 minute track listened to for 2 minutes scores 0.5 but a 1 minute track listened to for 1 minute scores 1. Any play under 30 seconds could be discarded.

It's an invasion of privacy and people will be too scared to enable the reporting because of the threat of the RIAA tracking them down and suing them

Aside from the fact that there are plenty of ways to legally listen to music through your favourite player, if we could engineer the database so that not even we could find the identity of people who are giving the data then we should be okay.

I don't want my player reporting every single file I play; some might be embarrassing

Plugins could have settings to only report from certain directories and file types. If you were ashamed of your main music collection then presumably you would not want to enable reporting at all.

People can game it too easily

How?

Listening to the same track over and over

Can accept at most one result (the longest) for each track per IP per day, will still be a valid measure of how many people are listening to whatever, and it isn't possible to make a super-long track without changing its SHA1 hash.

Some tools that might help

• MusicBrainz Client Library - open source lookup routines for an existing music database. Could be used as-is or reverse-engineered to start a new database.