Thank you very much, Mr. Ouzzine.
Again, please allow me to answer in English. It isn't because I can't answer your question in French, but I think I'll be clearer in English.
I'm happy to take the first question with respect to what you were talking about—humans versus machines or humans versus algorithms. I think the honest truth on that is that we need both, because we have a huge amount of scale, obviously. There are over two billion people on the platform, so in order to get at some of the concerns that members here have raised, we do need to have automated systems that can proactively find some of these things.
I think to go back to Mr. Collins's first question, it is also equally important that we have humans that are part of this, because context is ultimately going to help inform whether or not this is malicious, so context is super important.
If I may say so, sir, on the human questions, I do think you are hitting on something very important, and I had mentioned it a bit earlier. There is this need, I think, for companies such as Facebook not to make all of these kinds of decisions. We understand that. I think people want more transparency and they want to have a degree of understanding as to why decisions were arrived at in the way they were in terms of what stays up and what goes down.
I can tell you that in the last few months, including in Canada, we have embarked on global consultation with experts around the world to get input on how to create an external appeals board at Facebook, which would be independent of Facebook and would make decisions on these very difficult content questions. We think there is—at least as our current thinking in terms of what we put out there—this question of whether they should be publicly binding on Facebook. That is sort of the way we have imagined it and we are receiving input and we will continue to consult with experts. Our commitment is to get this done by 2019.
Certainly, on our platform, we understand that this is challenging. We want a combination of humans and algorithms, if you will, but we also understand that people will have better confidence in the decisions if there is a final board of appeal, and we're going to build that by 2019.
Of course, we're all here today to discuss the broader question of regulatory frameworks that should apply to all services online. There, once again obviously, the human piece of it will be incredibly important. So thank you, sir, for raising that, because that's the nub, I think, of what we're trying to get at—the right balance and the right framework per platform but also across all services online.