Yes, the idea of a single if statement is over simplifying, but it's an example of how easy such a system is. If I was grooveshark and actually building such a system I would build in common song name matching (eg: If Rihanna is the artist name and the song is "Rude boy" then it's bad, if the song is "Hacker News is fun!" it's probably not!
The point is, a system for matching against names of popular artists is trivial to build, are you denying that? If someone tasked you with building a system that could take an mp3 + title + artist and tell if it was the song of a popular artist in a database could you not do that?
Searching for "usher.mp3" where the artist.name = Usher is pretty much exactly how Prof. Usher's lecture notes got DMCA'd. I'll give you credit that you could probably ban most of the typos of Rihanna, though. But the wider you cast that net, the more "dolphins" like Prof. Usher you're going to catch.
And you have to cast it widely to even put a dent into things, because you only need one copy available for everyone to copy it. Just one false negative out of millions of songs and everyone copies that one that slipped through. They need exactly one search result, not 100. Google does all that you say and more on YouTube. Let me know how hard it is for you to find infringing content there (hint: not very)? Sure, they play whack-a-mole with it. But that damned mole keeps popping back up and it's not hard to find a mole that hasn't yet been whacked.
Worse, the more you tighten up that code, the more Prof. Ushers you ban. We have Google blocking access to bird songs already. And I know you know how skilled their coders are.
You should have at least banned the cryptographic hashes of anything that got DMCA'd. At least that has few false positives (assuming few DMCA notices are false). It's vulnerable to deliberate infringers making tiny changes to the files, sure, but I'm unaware of any solution that isn't. And you're required to ban anyone who is a repeat infringer anyhow. But we're playing whack-a-mole. More moles always pop up.
You're misunderstanding how Grooveshark works and why this would matter.
Yes, a copy of Rihanna's latest album could slip through but it would have to be hidden under a different name for that to happen, then how are people going to find it? If you want to listen to Rihanna on Grooveshark then you type "Rihanna" into the search engine.
> Worse, the more you tighten up that code, the more Prof. Ushers you ban. We have Google blocking access to bird songs already. And I know you know how skilled their coders are.
That's completely different though, context is important. Grooveshark is a "free Spotify", Youtube is a video community and a video hosting platform. Grooveshark has one use case, Youtube has many. For example Youtube developers work to catch music that is a part of a video that isn't necessarily focused on the song, that would never happen with Grooveshark. Also external sites embed Youtube uploads and use that to host music, can't be done with Grooveshark.
The fact is you or I (or any slightly competent developer) could build a system that could block almost every single possible upload that labels/artists don't want uploaded, Grooveshark CHOOSE not to do this.
As a user you need to know what you're looking for, you need to tell the search engine what you're looking for, if you're looking for Rihanna you don't type "r1h4nna4534535" and so if someone uploads the latest Rihanna album with the artist name "r1h4nna4534535" who is going to find it? Hell, even if they did there are solutions to that problem.
Of course they don't, it's not their job to police someone else's copyrights. Do you think Escape's shareholders wouldn't sue the officers of the company for breach of fiduciary duty if they discovered they were spending company funds to prop up some other business? If you own a copyright, it's your own job to enforce it. Why should copyright holders get to freeload off Grooveshark's work?
Banning via cryptographic hashes are honestly better for this, even then. Yes, it's flawed, but I know of nothing better. Name filtering is just a mess. A long time ago, I read up on soundex/metaphone, Levenshtein distance, etc. and wrote what was essentially a search engine for names. When I tried to make it loose enough to catch most common typos, a search for "the jerk" gave my boss as the top result. That was not planned and he was never once a jerk to me.
That said, please understand that I do sympathize with you about artists getting screwed. I just don't want to see solutions where we merely pass the buck to someone else without getting at any of the underlying problems. Frankly, I believe that the solutions lie more along the lines of making sure that artists get their cut, rather than restricting who can listen to music and how. But I certainly don't have all the answers here.
No, assuming filters are easy and foolproof is intentionally stupid and ignorant. Google took years to develop the technology and it was because they had the massive capital at their disposal. But all it takes to get around the technology is apply a few audio/video filters to get around the fingerprinting technology and words and phrases don't work, what if its a celebrity gossip video on Rihanna? That's fair use. The technology Google developed is easily gotten around by those determined and hits enough false positives that people with legitimate content have their videos taken down for no reason.
As far as the suffering of the music industry. Music sales are up, not down. There are also plenty of studies that show people who listen to artists online for free have a higher likelihood of purchasing the album online or attending a concert. People are buying more indy artists, not just what they hear on the radio and the middlemen in the record companies no longer control the distribution channels or who is exposed to what artists. Artists are benefiting from this loss of control, so I don't exactly feel bad for Universal, a one of the many record companies that has fought tooth and nail against any progress in consumer friendly methods of distribution and a legit and easy alternative to pirating. It took Apple to get them to finally let people buy audio files, and even then it was locked down which hurt people buying the music, not the people downloading MP3's off Limewire. They created the situation they are in now.
They don't need to filter the audio files themselves, that's completely un-needed, all they need to do is filter the meta data (eg: artist name, song title, album name). That is not hard. If you have a list of artists with their songs and albums you could easily match the submitted data against this and work out if the submission is disallowed. If they upload with "fake" meta data (eg: fake artist name) it doesn't matter because no user is going to find that music.
Yes, the idea of a single if statement is over simplifying, but it's an example of how easy such a system is. If I was grooveshark and actually building such a system I would build in common song name matching (eg: If Rihanna is the artist name and the song is "Rude boy" then it's bad, if the song is "Hacker News is fun!" it's probably not!
The point is, a system for matching against names of popular artists is trivial to build, are you denying that? If someone tasked you with building a system that could take an mp3 + title + artist and tell if it was the song of a popular artist in a database could you not do that?