CYBERSPACE—Earlier today, TorrentFreak posted an article about takedown requests sent by Takedown Piracy (TDP) to Google that erroneously targeted suspected infringement on the web-based Git repository hosting service, Github. But what TorrentFreak has not yet done is update its article, adorned with a face-palm graphic, to reflect the actions that Takedown Piracy immediately took to rectify the mistake as soon as it was informed whyat happened.
According to the article by Andy, "Acting on behalf of porn outfit Wicked Pictures, TDP sent Google a notice containing thousands of URLs targeting dozens of well and lesser-known file-sharing sites. Sadly, however, the notice also targeted coding site Github—over and over and over again. And Google complied." Andy then provides a half dozen specific examples of mistaken takedown claims made in the single takedown notice sent issue by Takedown Piracy to Google on December 22, about which he notes at one point, "It makes disappointing reading."
Andy also feels obvious sympathy for Google, about which he writes, "Every single week thousands of copyright holders and anti-piracy companies demand that Google removes links to allegedly infringing content.
"The effort required to deal with this deluge is considerable," he adds. "Google has received as many as 11 million requests in a single week and in 2014 alone the search giant processed some 345 million URL takedowns."
Andy concedes that "it’s believed that most takedown requests are accurate," and takes note of Google's efforts to "ensure that erroneous notices don’t negatively affect legitimate online services" by "regularly [rejecting] overbroad and inaccurate notices," but he also states that the company isn’t perfect," meaning that mistakes like the one with Github sneak through.
However, he offers far less sympathy to the issuers of takedown notices, referring to the act of sending large numbers of such takedown requests to Google as "carpet-bombing," which he says sarcastically at the end of his piece is "much easier, quicker and cheaper" than sending a notice directly to Github.
The problem is, for companies dealing with the reality of mass infringement, it is precisely the "easier, quicker and cheaper" aspects of using Google that allows them to be as efficient as they are in the infringement-thick world we live in. That said, "false positives" that result from automated spidering of suspected infringing URLs has been an issue in the past, and will likely continue to present a challenge to anti-piracy outfits as they continue to react to the cat-and-mouse game of infringement and detection.
That appears to be precisely the situation that Nate Glass at Takedown Piracy was faced with in the Github situation. When AVN reached out to Glass for comment, he was quick to respond with a thorough (and thoughtful) email explaining in detail what caused the false positives to occur, and he did to rectify the situation. We reprint his explanation here in its entirety:
We initially flagged a few URLs on Github because of repositories on there, mostly dealing with The Pirate Bay, that were using torrents leading to our clients works via Bit Torrent/The Pirate Bay. However, our system then flagged related URLs because it matched a domain we had flagged infringements from - that also were matches for the titles of works of other clients. This is not common practice for us, as we usually don't get hits on 'related' URLs that just happen to be on the same domain.
Once we were alerted to the issue, I honestly can tell you I was sick to my stomach. I really really hate it when something like this happens. We really pride ourselves on not making these kinds of mistakes. However, the margin for error in this field is very thin. In reality, we're like any software company, we have to tweak the software and fix bugs and glitches, we're just held to a zero tolerance policy by certain groups of people. That's fine, it's the industry we've chosen. The first thing I did when we found out about this was immediately send a notice to Google to reinstate the Github URLs (which as of this writing has been done by Google). The next thing we did was to fix the bug so that this can't happen again. We've tested it and our system is now whitelisting all Github URLs.
We like to think that in the ever increasingly difficult industry of anti-piracy we shouldn't be judged by .00025% of the total URLs we've submitted to Google. It means we need to scrutinize our software closer. That being said, it's certainly a face palm worthy mistake and one we won't make again.
It should be noted that TorrentFreak, an excellent site despite its generally anti-copyright stance, throws a bone to Takedown Piracy, noting in its defense that "the company has sent 39.6 million URL notices to Google since 2011," but then quickly adding, "However, that will be of little comfort to the many legitimate projects which are now harder to find due to the company’s errors."
Ironically, or not, Glass confirmed to us what we suspected, that TorrentFreak did not contact Takedown Piracy before posting its story, and, as mentioned, have thus far failed to update it. No wonder most of the comments to the article are screaming for Takedown Piracy's blood. The impression they are being fed is that TDP is either willfully sending out false positives or doesn't care that it is.
But as Glass makes clear in his words above and also noted in a separate communication to us not marked for publication, he took the incident with the seriousness it deserved, and took it squarely on himself.
"I couldn't sleep last night," he wrote AVN. "After we sent the notice to Google to restore the URLs, I went into fixing the issue. Once that was done I just laid on the couch for hours pissed off at myself, worrying about how and why it happened, and just overall so frustrated that we didn't catch it. Grrrrrrr."
Maybe you have to know Nate Glass to believe his expression of regret and determination not to let something similar happen again, but while it may be the case that TorrentFreak will never care to make an effort to be more open-minded about outfits like TDP, it certainly seems like the pot calling kettle black for them to generally admonish the "wide range of copyright holders" for not bothering "to send a notice to Github itself," when they, too, did not bother to contact the subject of their scornful article, which also makes for disappointing reading.