On “Free Speech” and the internet

We’ve recently seen Cloudflare decide to terminate service for “The Daily Stormer”, because the stormer is seen as a Neo-Nazi site. There is an ongoing internet debate which tries to untangle if this is a valid or invalid decision. Both are right. Let me explain:

The solution I would propose is the same as one by Dan Geer (see [0]). Any internet company has a choice between two options with respect to their net neutrality on the internet:

  • Either, you are a carrier and enjoy common carrier protections. You move data around for other people. As a common carrier, you can not be held liable for what is carried in “your network” or over “your service”. But — as a common carrier you are not allowed to inspect the data you are carrying.
  • Or, you are allowed to do (deep) inspection of the data you carry, charge differently for different types of data, reject some data over other data, or otherwise interfere with the normal operation. But you can thus be held liable for the data you are eventually carrying, since there is no way you can claim that “you did not know about the contents.”

The key observation is: you can’t get both at the same time.


Cloudflare must decide if they happen to be a carrier in the above scheme, or if they want to be held liable for the traffic they carry. If not, it would open them up to DMCAs, for instance, since they aren’t allowed to just protect themselves under the carrier label. If they are a carrier, they also have ample protection against people who disagree with the Daily Stormer: “we are just a carrier, your complaints should go to someone else if you think the Stormers operation is illegal.” In short, Cloudflare is protected against the political pressure by individuals, lobby groups and so on.

Likewise, an ISP who wants to inspect TCP/IP streams of their customers, build profiles, and sell the profiles to a 3rd party can be held liable for the eventual illegal data transfer of their customers. They inspected the data, so they should know. The fact that it was a machine doing the job is a detail[1].

Google, Facebook, et.al, takes data produced by society and abuses it for machine learning purposes. Under the choice-law, they are now liable, since they inspected the data for another purpose. Of course, they can pay people for access to the data if they want. And they are free to set any price they like on the data. If Google thinks it is worth $3 per year for them, I doubt many people would accept. But make that $3000 or more, and I think a lot of people would do it. The current problem here is the lack of transparency in the market.


The current situation, in which you can pivot between being a carrier or not, is not efficient, nor very productive. In particular, it doesn’t give individuals basic rights, irrespective of their grouping (be it political, ideological or racial, etc). And it doesn’t give companies basic rights either, so they can give in to ideological pressure, one way or the other. There is a fine line between being a (public) utility and a company who can make choices in this regard.

You might argue, from the heart or mind, that a site like “Daily Stormer” shouldn’t be allowed to have any kind of representation at all, but I think this goes against some first principle of free speech. It doesn’t violate any laws, but one has to acknowledge laws are lowest-common-denominators of what can work in practice. They are not morally or ethically superior by definition, because they have to be clear-cut and arbiters of what is right and wrong. You can easily have examples where a tyranny of the majority, or a specific political/ideological leaning, can restrict speech for some out-groups. It happens in the small, for instance by silencing neuro-atypical human beings. It happens in the large as well, for instance by claiming there are certain areas which we as human beings shouldn’t even research (Examples: Global Warming, Race/IQ, Gender Differences, …).

But, rather rarely, the most outrageous speech is some times the truth we don’t want to hear. I think we can recognize that by giving our political or ideological opponents the liberty of expression, we can hope to gain the courtesy of getting it back. At least we have the moral upper hand in demanding that they do so, if you grant me the naïvety of such a proposal at first. On the other hand, cancellation and termination does nothing but breed animosity. And animosity eventually becomes violence in the fullblown scale.


The underlying problem, of course, is that law has not caught up with the internet yet. Politicians still believe they can treat the internet as any other entity, and since most of them don’t really get what it is all about, they create some rather indecent proposals. They often lean entirely to one side, that of government convenience, without giving the citizens an equal amount of rights in the process. Such laws are likely to be opposed.

We need internet laws, and tech laws. Dan Geer’s work is an excellent place from which to spawn the discussion.

[0] Security as Realpolitik: http://geer.tinho.net/geer.blackhat.6viii14.txt

[1] We should also have a law which forces an ISP to convey how much they earn on selling the data, so the cost of your internet becomes transparent.