The Mininova verdict

Overview

After giving its own introduction to the case, the court gives its assessment of the claims, and this assessment forms section 4, the largest section of the published document. Quite quickly, from subsection 4.4 to 4.9, the court deals with and rejects the accusation that Mininova itself is committing copyright infringement, but it then goes on to look at the “alternative basis” for guilt, that of encouraging and profiting from copyright infringement. By showing the way that administrators (admins) and moderators (mods) run the site, and after discussing the contents of the site, the court decides that Mininova is guilty of this second charge, and proceeds to dismiss various defences put forward.

In section 5, the court hands down its ruling:

5.1. hereby duty Mininova acted unlawfully by giving opportunity to, to encourage and profit from infringements of copyrights and related rights of the holders represented by Brein,

and requires that Mininova no longer include on their site any torrents whose titles match a list from Brein, or, crucially, any torrents which relate to files for which there is “reasonable doubt” about the rights holders giving permission for the files to be downloaded. This requirement, it seems, was too onerous for Mininova to fulfil while maintaining with their relatively open nature, so the court case must have been a major disappointment for the people running the site. It may be some consolation, though, that the court only required them to pay €2147.44 to Brein.

The evidence

Going back to the beginning of section 4, then, subsection 4.7 correctly states:

It is the user that ensures that by clicking on the torrent and using a (not from Mininova) software program is brought into a process leading to direct transfer of the stock of the co-Internet users for his company.

It is a little worrying that the court feels it necessary to specify that the users’ torrent clients were not made by Mininova, as if it might count against Mininova if they funded development of an open source bittorrent client, for example, but this could just be a “for the avoidance of any doubt” clause.

By subsection 4.10, the court has decided:

The conclusion of the foregoing is that if these claims are based on actions of Mininova contrary to the Copyright and Neighboring Rights Act should be rejected.

which seems to mean that the court rejects the accusation of the plaintiff, Brein, that Mininova itself is performing the distribution which would break Dutch copyright law.

To show that Mininova is guilty of a crime under the “alternative basis”, the ruling starts by looking at the actions of the admins and mods of the site. Firstly the point is made that:

Mininova recognizes that the actions of the administrators, being the founders and directors of Mininova, can be attributed to her.

but then the court tries to show that the admins worked together with the mods of the site, as a sort of conspiracy. This raises one of the same questions that the Pirate Bay trial raised, which is when this sort of “conspiracy theory” becomes mere “guilt by association”, but I think there is a stronger case that the admins and mods did work together on a daily basis than that the Pirate Bay’s ISP was involved in the day to day running of the site (and thus that statements he made should be admissible as evidence of the other defendants’ motivations).

Most interestingly, though, the evidence it gives for this conspiracy is quotes from the forum, such as (from subsection 4.15):

“All other admins are also working hard to remove fake torrents or torrents that are against the rules” (administrator [administrator 2], Annex 10);

“We don’t need to download everything as we have PLENTY and I mean plenty of moderating tools to do our detective work ha so it makes life easier. “(Moderator” Moe1210 “Annex 23).

So, in terms of what Mininova did wrong, one could say that their mistake was to talk openly on a public website about how their site worked. If, in an extreme example, the guilt or innocence of a site owner depended solely on statements they had made on a website about their filtering rules, it starts to become a question of freedom of speech or self-incrimination, as the crime is equivalent to “admitting your filtering policy”.

The court then concludes (subsection 4.16):

The moderators use their powers in particular torrents referring to viruses, pornography and empty files. That is precisely the policy Mininova stands on its platform. Thus the moderators act as implementers of policies Mininova; … The “moderation” of the platform is seen as a shared responsibility of the administrators and moderators.

then (4.18):

Based on the foregoing Mininova held that the acts of the moderators such influence that its actions should be attributed. This means that the evaluation of the Mini Nova’s complaints should be taken as the basis that the actions of the moderators and administrators as acts of Mininova are to apply.

which seems to put the moderators into a rather difficult legal situation, although as no charges were brought against the individual administrators, the moderators also managed to escape a crippling financial penalty against them, unlike that faced by the Pirate Bay defendants.

Next, in order to invalidate the potential defence of being a mere service provider, the ruling explains that Mininova were actively involved in determining the contents of their service. One of the earlier quotes from subsection 4.15 was:

“Thanks for reporting. I deleted the fake version and uploaded the correct one “(administrator [administrator 1], Annex 9);

which seems to have been taken as proof that Mininova were occasionally influencing their platform both ways, not just removing files but adding them. Jumping forward to subsection 4.41, when the EU safe-harbour defence is considered, the court says:

From what has been considered, it follows that Mininova will engage in such information, ie including its involvement with the classification on its platform in categories where the torrents can be saved, remove the torrents that conflict with its own house rules, helping users to obtain useful files (including advising on the circumvention of security), and the occasional self uploading files with copyrighted material.

(If Mininova could prove there were such things as “self-uploading files” they would perhaps have got off completely, but I think this is just an artefact of the otherwise very impressive machine translation). Thus the court concludes (subsection 4.16):

Given the foregoing, Mini Nova not immune from liability to Brein described above for its conduct related to (the contents) from third parties, stored information.

Again the implications of this are somewhat chilling, in that it suggests that a company running an online forum could be liable for copyright infringement if a forum thread ended up containing a few pages of a recent novel, provided that the company had previously removed comments which contained swearwords (perhaps using a semi-automated process). Consider also Wikipedia, where the employees of the corporate body backing it can be Wikipedians themselves, and thus liable if anyone creates an article which plagiarises another encyclopedia.

Mininova’s final escape routes were cut off in the court’s analysis of the composition and internal structure of the site. Firstly, evidence was presented about how many of the torrents on their platform related to copyrighted works. According to subsection 4.21 of the ruling:

Brein TNO commission research into any torrents on the platform of Mininova are placed in the category Games, Movies, Music and TV Shows. The conclusion from these studies is that these categories significantly (on average 80-90% of the selected torrents) contain references to files with copyrighted material.

whereas Mininova didn’t come up with a figure. I don’t like the idea that a service can become illegal if a certain threshold of the activity conducted on it becomes illegal. Spamming, for instance, is illegal, but I don’t think that when email traffic became 50% or 90% spam, that should have suddenly made all ISPs liable. Fortunately the argument here does not appear to be that a service with illegal uses makes that service illegal. Instead, the purpose of the statistic may just be to show that it isn’t the case that, for example, 90% of the activity is legal, because if it were, Mininova could have more easily claimed they were unaware of any infringement going on.

Next, it was pointed out that the site had categories, such as “Movies”, which was given as evidence that Mininova knew that copyrighted material was being shared. Bizarrely, Mininova’s defence was that the categories were auto-generated, but were unable to give any proof of this. If they really did have an AI which could look at a collection of files and create categories for them, and assign the files to those categories, it would seem that Mininova should have trained it to implement their filtering policy for them, as that way they wouldn’t have needed moderators.

Another classic piece of courtroom comedy, perhaps rivalling the King Kong Defence from the Pirate Bay trial, was when Mininova claimed that the “Disney” category must refer to fanart of public domain Disney characters. Of course, Disney would like to claim that there is no such thing as public domain Disney characters, but at least some jurisdictions do not have infinite copyright terms, and some Disney characters have probably made enough money to have rewarded their original creator (and his children, and grandchildren). Perhaps Mininova’s brave attempt here should be called the Mickey Mouse Defence, which is only fitting in a world with a Mickey Mouse Copyright Act.

The last pieces of evidence from the prosecution were quotes of mods thanking users for uploads and giving passwords to encrypted files, and that Mininova made money from adverts. Based on the conspiracy theory, the thanking and passwords were used to show that Mininova was “encouraging” the illegal downloads, as explained in subsection 4.35:

THE FOREGOING should lead to the conclusion that Mininova making copyright infringement by its users, and promotes itself occasionally makes copyright infringement.

and the fact that Mininova made money off their adverts was used to show that Mininova had a financial incentive to not block copyright material. Admittedly if 90% of the torrents would have to be taken down for Mininova to help prevent copyright infringement, and Mininova’s profit were linearly related to the number of torrents, then that would make them reluctant to do this. It is sort of begging the question, though, of whether it was Mininova’s duty to stop other people infringing copyright, even assuming that Mininova were aware that this was going on.

The defence

Apart from the defences already mentioned, such as the EU safe-harbour defence, and the non-incriminating nature of the categories on their site, Mininova did make several strong arguments in favour of their innocence. One of these defences was that a download through the site is not the same as a lost sale, therefore BREIN cannot prove they are harmed by the site. Apparently Mininova even produced research to back up their claim, but the court decided to not believe it. The court did, however, present a fairly reasonable objection to this defence (subsection 4.48), which was that even if the research was true and only a small percentage of downloaders were actually downloading instead of buying, that small percentage has to be applied to 2 billion (the number of downloads through the site up until June 2007), equating to a significant amount of financial loss for the companies being represented by BREIN.

The next defence was one of “freedom of expression” (subsection 4.52), with the argument being that Mininova wasn’t in a position to determine which files were legitimate and which were not, and would thus be forced to block all files to be on the safe side. This did not sway the court, however, as BREIN was willing to provide a list of titles of infringing works and Mininova, it seems, only needed to block the torrents which matched. Mininova did respond to this, saying they were worried about false positives, but the court said that there is only a small risk because the list was based on titles, and that Mininova could set up a reverse takedown procedure so files could get re-added. Later on, in subsection 4.71, Mininova raised the problem of how closely the torrent titles had to match the ones given in the list if they were to implement such a system, and the ruling states:

Mininova’s objection to the fact that the penalty also covers advanced “corresponding titles,” the court shall not. As Brain occasion of the call explained, sees this phrase on small variations in titles with technical means can be recognized as equal to the titles in the list.

Subsections 4.55 and 4.56 quickly address the issue of proportionality, and it is calculated that it would take 4 full time employees to filter the 1 million files listed on the site. The court somehow decides based on this number that it should be Mininova’s responsibility, but it’s unclear what logic they used. Then Mininova’s final attempted defence (in subsection 4.57) was that the Notice and Takedown (NTD) procedure should be sufficient to achieve the ends BREIN desired. The court disagreed, saying that this procedure added a delay before the files were removed, and once removed the files were quickly re-added. If the takedown procedure included a “memory” of the files taken down and prevented them from being re-added, the court, it seems, might have looked on it more favourably, but I guess we’ll only find that out if another site with a different NTD procedure gets sued.

What did Mininova do wrong?

That is the real question, not “Was Mininova breaking the law?” or “What were Mininova found guilty of?” but “What did they do wrong?” Someone else can give their opinion of whether the law reflects some form of morality, or the democratic will of the people, but I fill focus instead on how they got caught. I have already mentioned some of the evidence presented against them, and will reiterate my worry that their guilt or innocence could be decided by comments they made in a forum. I’m not saying that if someone overheard a murder suspect saying “I could just kill $future_victim” then that shouldn’t be admissible as evidence, but I am saying that the rule of law would become very murky if people knew they could get away with murder by simply not talking about their intended crime. I would like to think that the legal system could catch criminals who lie or keep secrets, and not just the minority who slip up or, even more unfairly, the minority who don’t realise that what they’re doing is a crime.

To imagine the implications of this, suppose that Mininova were not hosted in the conventional sense, with an individual or company paying a regular fee to a hosting provider, but instead the site took the form of a wiki, edited by anonymous accounts. The admins and mods could run an invite-only (OTR) encrypted back channel to discuss internal policy, and leave it up to the users to guess what the policy was. A technologically literate prosecution team could still probe this policy by uploading various types of files and seeing which got taken down, so perhaps the solution really would be to have no policy, or to democratise and crowd-source the policy by letting users themselves vote on which files should be taken down, or at least which files should be ignored by other users.

It really does seem that this court case has provided a blueprint for how to make a legitimate torrent indexing site (in the Netherlands at least). The admins should let the users vote on everything from categorising to quality assurance of torrents, obviating any need for the admins to ever contribute to (or even read) any of the forums. Under this architecture, implementing a NTD process might actually make the admins more liable, as they could lose their safe-harbour immunity, depending on how responsive they were to NTD requests and how close a match they required against a file to take it down. Adding a “memory” to the NTD process so that files couldn’t be re-added after being taken down might also help, and it appears that having a counter-NTD process wouldn’t harm their liability. I would be interested to see what would happen if there were a “community NTD” process, where NTD requests were made visible to the whole community, who could activate the takedown of a torrent by giving the request a sufficient number of up-votes.

Is this the line to be drawn between Google and The Pirate Bay, then? Would online copyright infringement be reduced if every service stayed the right side of this line?