It hasn't been an especially felicitous year for the founder of file-sharing site Megaupload — his domain name has been seized, his assets have been impounded and Kim Dotcom faces potential extradition to the US on criminal charges of copyright infringement.
Daniel Raimer, RapidShare's general counsel, told CNET, "We try to come up with new ideas to accommodate the industry, but not at all costs".
(Credit: Declan McCullagh/CNET)
That's a fate that RapidShare is determined to avoid. The Swiss company said that it wants to be a legitimate hosting service that not only responds promptly to removal requests from copyright holders, but goes far beyond what the law requires.
RapidShare's "responsible practices" policy may have pleased Hollywood when it was announced in April, but it nevertheless remains controversial. The US advocacy group Public Knowledge responded by saying that the policy "implies that cloud services that choose to merely comply with copyright law" are "somehow morally deficient or in favour of copyright infringement".
RapidShare said that it employs over 50 people and has over 400,000 files a day uploaded to over 1,000 servers by its users.
CNET spoke this week with Daniel Raimer, the company's general counsel, about the techniques RapidShare uses to detect piratical material and how far it's willing to go.
What do you use to try to detect pirated files?
It's a software component that has two parts. There's a 'bot or crawler that is constantly reading websites. The second component is a matching algorithm that is a piece of software that, wherever it finds a RapidShare download link, it tries to make an estimate of what that is.
If there's a website saying "Avatar the movie — illegal download", our crawling algorithm would find the download link and the matching algorithm would say that the content behind the download link [is infringing]. It creates Excel spreadsheets with the URL where the link was found, the URL that was found and the estimate of what the content is. That's processed by our abuse department.
The Web is a big place — how many sites do you look at?
We have already come across 400 major websites where there was a lot of copyright infringement that had taken place. We've programmed our crawler to look at them.
Can you give an example?
When people hear that we have a crawling system, they assume that the crawling system is really fast and can reach millions of websites in a short amount of time. It's not that our software has been programmed in a bad way. But we have to slow it down on purpose, so it's not recognised by the operators of these websites.
If you were to make 1,000 requests per minute, then the administrators and operators of these sites, such as Warezbb.org, would recognise it as a 'bot and block it. As a result, it's really time consuming.
Why do all of this work if you're not legally required to?
It hurts our reputation to have all those copyright infringing sites. We believe it's a much more interesting market to have the legitimate customers upload important files that they want to have for long periods of time — a reliable cloud computing service that they can trust. These are the types of customers we want. Legitimate customers don't really want to argue whether your service costs $4.99 a month or $50 a year.
Copyright pirates are different. They really want everything for free. They're definitely not the long-time customers. Plus, they give you a headache. You get sued. You have to have an anti-abuse department.
Does your system satisfy US content holders? The Recording Industry Association of America?
It's their business to not be happy. We take down 2.5 times more files (on our own), than those in reaction to takedown notices. This number is going to go up ... Today, we find 2.5 times more files than all the content industry together.
What kind of requests do you get from the porn industry?
It's been really broad. Let me say that some people in the porn industry seem to be less reasonable than the movie industry or the music industry in their requests. Their suggestions seem to be less thought through.
They are people who will call us, yell at the phone, "Shut down the service!"
What else could large content holders like you to do?
What they want us to do keeps changing. It's been changing over six years now. First they wanted a word filter. Then they wanted us to have no anonymous users any more. Then they wanted us to stop the reward system.
Currently, it seems to be content recognition filtering. At least, that seems to be true for the music and movie industries.
Is anonymous use protected under EU or German law?
I'm an attorney, and it's a tough question to ask an attorney. In my legal opinion, German law is absolutely clear. A provider must allow for anonymous users. The thing is that, even some courts have ignored that statute.
Do you currently allow anonymous users?
We allow anonymous users.
Have you tried a word filter?
We've tried it a few times, and found the results terrible.
Because of generic terms?
Generic terms. We did a test run when the new Harry Potter movie came out. We thought that putting "Harry Potter and the Sorcerer's Stone" on the word filter should be a sure-fire thing.
It turned out that the word filter triggered all the time, whenever people uploaded video trailers. What I didn't know, and other people in the anti-abuse department didn't know, is that people have a thing for video trailers. But we always had to sort out the illegal movie copies from the video trailer that people were allowed to have. Sometimes, it's fan merchandise.
Why do all this work if you're not obliged to under the US Digital Millennium Copyright Act?
The simple answer is that German law is even stricter and broader.
We try to come up with new ideas to accommodate the industry, but not at all costs. In addition, content recognition is not going to work on a RAR file that's encrypted.