Further to my last post.

In order to trade the rights to an idea independent of the expression of an idea, we need to be able recognise that expression.

In the audio world, we've been trying schemes involving, watermarks, embedded DRM, cryptographic hashes, markers in the start of red book CDs. All of these involve adding a marker to a specific expression. But this doesn't actually agree with the case law surrounding rights, patents and plagiarism. As humans we think we can identify a 32Kbps MP3 or a FLAC lossless copy as being the same song. We can probably identify a remix (the MTV Video version) as being the same song. With a bit of an effort we can identify all the samples in a bit of remixed hiphop. And so we have the example of the RIAA suing people based on filenames when the file is actually a Linux utility and not a top ten track. And their utility for checking P2P programs and audio/video on your machine makes no attempt to identify which files are licensed or even what they are and hence whether they need to be licensed.

So identifying a specific idea within an arbitrary expression of that idea reliably with automation is a hard problem. Perhaps we should be looking at technology like Shazam, that can make a good stab at getting a song title and artist from a snatch of music recorded by a mobile phone.


[ << Burningbird ยป The Clean Industry ] [ GP2X.co.uk official site - Buy the GP2X Linux console >> ]
[ 26-Sep-05 7:34am ] [ , ]