Jillian C. York: We all know that hate speech is a huge prob­lem online but we don’t real­ly nec­es­sar­i­ly agree on what hate speech” is. Same goes for harass­ment, same goes for vio­lent extrem­ism. A lot of the top­ics that we’re try­ing to tack­le” or try­ing to deal with on the Internet, we’re not actu­al­ly defin­ing ahead of time. And so what we’ve end­ed up with is a sys­tem where­by both com­pa­nies, and gov­ern­ments alike, are work­ing some­times sep­a­rate­ly, some­times togeth­er, to rid the Internet of these top­ics, of these dis­cus­sions, with­out actu­al­ly delv­ing into what they are.

So, I work on a project called onlinecen​sor​ship​.org, and what we do is look at the ways in which social media com­pa­nies are restrict­ing speech. Now, you could argue that not all of this is cen­sor­ship, and I might agree. We look at every­thing from, as you may have seen last year, the cen­sor­ship of nudi­ty, which I firm­ly believe is cen­sor­ship, to take­downs around harass­ment, hate speech, and vio­lent extremism—some of which bor­der the line between incite­ment and poten­tial­ly law­ful speech in cer­tain jurisdictions.

So I looked at the dif­fer­ent def­i­n­i­tions of hate speech that the major social media plat­forms give. Twitter for exam­ple says that you may not pro­mote vio­lence against, or direct­ly attack­er or threat­en oth­er peo­ple on the basis of race, eth­nic­i­ty, etc. All of the dif­fer­ent cat­e­gories that you might imag­ine. It’s very sim­i­lar cat­e­gories to before.

So, we know that these com­pa­nies, these places where most of our speech takes place and spaces online…we know that they’re already com­mit­ted, or at least they say so, to tak­ing down or reg­u­lat­ing that kind of con­tent. But gov­ern­ments haven’t nec­es­sar­i­ly agreed, and I think Kirstin spoke about this a bit ear­li­er today. Governments have felt that what the com­pa­nies [record­ing skips] not enough. And so last year we saw the German gov­ern­ment at the end of last year form an agree­ment with the com­pa­nies to take down hate speech by get­ting them to remove it with­in twenty-four hours of being reported. 

So essen­tial­ly how this works is you already have these flag­ging mech­a­nisms that exist on Facebook. When I report con­tent, I can report it as…there’s a num­ber of dif­fer­ent, and it actu­al­ly is quite gran­u­lar. I just looked when it comes to hate speech. 

But then the German gov­ern­ment want­ed them to go a step fur­ther and make sure that they’re doing it with­in twenty-four hours. We’re talk­ing about 1.6 bil­lion users, con­tent mod­er­a­tors of…some num­ber. We don’t actu­al­ly know how many peo­ple are employed by these com­pa­nies to look at those num­bers, and hope­ful­ly we’ll have more trans­paren­cy about that soon—fingers crossed. 

But, essen­tial­ly they’re ask­ing for a twenty-four hour turn­around for all of this 1.6 bil­lion users. This is almost impos­si­ble. And it’s not just Germany. We saw the European Commission of course, which I think was dis­cussed ear­li­er, as well as more recent­ly a poten­tial agree­ment between the Israeli gov­ern­ment and Facebook to deal with incite­ment. Now, that one’s less clear. There’s some denial as to whether or not there is in fact an agree­ment. But it remains to be seen what kind of con­tent’s tak­en down. So far, how­ev­er, I would note that with­in the past two weeks since this went pub­lic, we’ve seem two edi­tors of pop­u­lar Palestinian pub­li­ca­tions cen­sored by Facebook for rea­sons unknown. Facebook apol­o­gized, said it was a mis­take, but nev­er­the­less, when­ev­er there’s that addi­tion­al scruti­ny put on a cer­tain cat­e­go­ry of peo­ple or a cer­tain geo­graph­ic loca­tion, you’re bound to see more erro­neous take­downs, as I like to call them.

And so then what do we do about this? Because if the gov­ern­ments feel that the com­pa­nies aren’t doing enough, and we as soci­ety have no input into that, then essen­tial­ly what we’re see­ing is this quick, reac­tionary attempt to you know, like I said before, get rid of all of the con­tent with­out actu­al­ly assess­ing what we’re look­ing at. What is it that we’re talk­ing about. And I think that that’s the first step, is that we haven’t agreed on a def­i­n­i­tion of hate speech. My vision of it might be dif­fer­ent from yours. As we’ve seen by dif­fer­ent gov­ern­ments, the vision is dif­fer­ent. And so if we want a free and open Internet—and I’m not say­ing that we should­n’t tack­le hate speech. We should absolute­ly tack­le hate­ful speech. But, if we want a free and open Internet where we all have equal access and equal oppor­tu­ni­ty, then this addi­tion­al frag­men­ta­tion that we’re see­ing through these dif­fer­ent pri­va­tized agree­ments is not the way forward.

And so first we need to find a def­i­n­i­tion that actu­al­ly works for us, before we even talk about what to do about the speech itself. And then of course vol­un­tary back­door agree­ments between gov­ern­ments that we’ve elect­ed demo­c­ra­t­i­cal­ly, in all of the exam­ples that I’ve giv­en so far. We’ve also seen some…less-demo­c­ra­t­ic gov­ern­ments try to strike deals with com­pa­nies, and that’s anoth­er prece­dent that we might be set­ting with these. But regard­less, they deny us input. And by us I mean cit­i­zens, I mean cit­i­zens of both our coun­tries and of the Internet, cit­i­zens” of these plat­forms inso­far as you could make that argument. 

But we have no input into this. We’re not part of these con­ver­sa­tions. Not only have civ­il soci­ety groups, NGOs, been exclud­ed from the actu­al table where these agree­ments are being decid­ed, but the aver­age user has no actu­al say in how these spaces are gov­erned. And so I’m not going to talk about nudi­ty this year—I know I’ve talked about it the past two years. But I will throw it in there as an exam­ple that I think it’s real­ly inter­est­ing that com­pa­nies have just decid­ed for us that this is an unac­cept­able thing. Now, their rea­sons might be valid. It might be real­ly dif­fi­cult for them to tack­le the dif­fer­ence between pornog­ra­phy and nudi­ty. There are all sorts of tech­ni­cal rea­sons why that might be a real­ly hard ques­tion, and I respect that. But. They’ve already made the deci­sion to go beyond the law there, how do we know they’re not doing that in this case, too.

And then I would also go a step fur­ther and say that cen­sor­ship alone does­n’t actu­al­ly solve the prob­lem of hate­ful speech. It does­n’t. And I’ll give you a cou­ple examples.

I was in Budapest a cou­ple years ago, just walk­ing around in the sum­mer. And this was in the mid­dle of the debate in the United States around the Confederate flag. And so for those who might not be as famil­iar, the Confederate flag was of course the flag of the sep­a­ratist South, and has now become large­ly known as a…well, at least where I’m from it’s known as a hate­ful, racist sym­bol. And in a lot of the south of the coun­try it’s a sym­bol of pride for the Confederacy. But nev­er­the­less it’s pret­ty known for what that is.

But when I see it inter­na­tion­al­ly in anoth­er con­text, my reac­tion is oh, maybe they’re just you know, try­ing for some Americana. And so I saw this mil­i­tary shop in Budapest and I saw the Confederate flag, and I thought oh, well maybe it’s just like an Army Navy store. So I post­ed it on Twitter and I asked some friends in the coun­try, and they were like, No no no, that’s a Nazi symbol.” 

I would­n’t have known that. Because what hap­pens when you cen­sor some sym­bols is that oth­er ones crop up in their place. And we’re start­ing to see that on Twitter and Facebook now, with secret codes to avoid cen­sor­ship. I’m not going to get into the actu­al def­i­n­i­tions, but if you look at this arti­cle, essen­tial­ly you’ve got real­ly far-right right wing com­mu­ni­ties online that are using inno­cent words like Google,” Skittle,” and Skype” as sub­sti­tu­tions for racist words. 

And so this hap­pens, and then we see this in China to get around cen­sor­ship there in more pos­i­tive con­texts. But it’s only a mat­ter of time if com­pa­nies are build­ing in algo­rith­mic meth­ods to fil­ter or cen­sor speech, it’s only a mat­ter of time before peo­ple just come up with new sub­sti­tutes. That’s how peo­ple have always got­ten around cen­sor­ship. I don’t see how that will not continue.

But then last­ly I would also say that cen­sor­ship isn’t the solu­tion to hate­ful speech. It might be a solu­tion; it may be a com­po­nent of the solu­tion. I don’t know. That’s some­thing for demo­c­ra­t­ic process­es to decide. But, it does­n’t solve the prob­lem. To solve the prob­lem we have to get at the root caus­es of it. And this is why I find this title for this talk real­ly chal­leng­ing. Because I’m not the expert on how we deal with hate­ful com­mu­ni­ties and hate­ful speech and all of the right wing groups that are crop­ping up in my coun­try and yours. 

But, I do know that we should be look­ing some­where else, and I think we’re ask­ing some of the wrong ques­tions as to the ori­gins of this. And I’ll just flip through these real quick to note all of these are peo­ple who are ver­i­fied on Twitter and who engage in hate­ful speech on a reg­u­lar basis. These are our lead­ers. These are the peo­ple that we should be ask­ing the ques­tion, how do we get rid of hate­ful speech? It’s not nec­es­sar­i­ly let’s just strike it from the record,” it’s let’s get to the root cause and then we can talk about what we do with it on our online com­mu­ni­ties. So thank you, and thank you for hav­ing me.