November 17, 2017
This issue's contents Current issue

Can Free Speech Be Hacked?

By Jonathan Wallace

In today's dizzy information-sphere, tropes blow up in an instant: you hear a phrase for the first time on Tuesday and then a hundred more times by Friday. An amusing example during the 2016 campaign was "pivot"; there was constant discussion of whether Trump was going to pivot towards his base or moderate Republicans, or the Democrats were going to pivot towards the working class or Red States voters, etc.

A trope that blew up last week was whether the Russians have "hacked" our freedom of speech, or alternately, American liberty. This is a truly fascinating one. The idea is we have an open and free environment, in which anyone can say anything they want. The Russians show up, via covert social media teams, hackers and bots, and create web sites, Facebook pages and Twitter accounts, to participate in what should have been a sincere American conversation. By spreading lies and fomenting distrust, they pollute our discussion.

This sounds highly plausible on its face (at least if you suspend disbelief as to the quality of the pre-existing conversation), and "hacked free speech" is a catchy phrase. But, when I thought about it for a sixth minute, it occurred to me that what got hacked was not free speech, but capitalism.

Circa 2008, I was living temporarily in Florida, using as my main computer a used Macbook I had hurriedly purchased from a computer shop in my New York town as I was leaving. The computer had Microsoft Office installed and I assumed it was a legal copy. One day, a dialog window asked if I wanted to upgrade to a newer version. I clicked yes, and Microsoft paralyzed my installation of Office, with another dialog asking me to enter a key I did not have. I contacted the man who sold it to me and he said, "I can't help you", and offered me a new copy of Office for (I think) $300. I never did business with him again.

If I had been at work on anything critical, for example an amicus brief to the Supreme Court that needed to be filed that night, I would have been completely screwed. The incident got me thinking. For years, I had been haunted by the danger of computer viruses. A laptop I used for work, after I opened an email from a stranger, displayed the words on its screen "I am your new best friend" and became unusable. It seemed very bizarre to me that Microsoft could not build a version of Windows that was impregnable to viruses. I now understood that if Microsoft desired to leave back doors which would permit it to cripple my installation of Office, by definition it needed to leave holes which could be exploited by others as well.

Every ferocious monopoly on the modern Internet has the same consideration. There was a fascinating story a week or so ago that prostitutes with Facebook accounts in their real names are starting to get friend recommendations for johns who only know them by their street names. There is only one credible explanation for how this is done, which is that Facebook is buying cellphone proximity information, the fact that two cellphones were within ten feet of each other for an hour. Facebook swears it isn't doing this. But the technology which knows everything about you, what you searched for and what you purchased, must of necessity contain those same holes, tunnels and trap doors which the Russians allegedly exploited.

Let me come at this another way. I joined Compuserve in 1984. By 1986, we made a discovery that is made and forgotten by every generation (I have since read accounts of people who figured this out on even earlier online systems in the 1970's): unmoderated lists don't work. I joined an email list on Internet free speech maintained by a journalist who couldn't be bothered to read it hourly or to eject trolls. It probably had two or three hundred members. At inception, famous law professors and Internet activists engaged in intelligent conversations I found highly educational and engaging. Six months later, the trolls had chased all of them away. There is some number, probably not more than fifty users, above which you are guaranteed to have a troll and need a moderator. Facebook is an unmoderated list with billions of users.

This is a problem which cannot be solved. Of course some people will livestream murder, torture, suicide. Artificial intelligence is nowhere near at the level where it can detect all evil content (or avoid blacklisting innocent content). AI will probably never reach that level of competence, but if it does, there will be other highly scary implications of living in such a world. In the meantime, we are left with the question: how many human moderators would Facebook have to hire, train and supervise to moderate the entire service? Would 100 million be enough? Human organizations probably can't scale to the size necessary, and even of they could, capitalists have no incentive to do it, and pocket far less profit, as there are no consequences of creating huge, diseased systems. You can't run a system the size of Facebook on automation without making it a cheesy framework easily hackable by Russians.

Craigslist tried to solve this problem by inviting the users to act as moderators. This resulted in the trolls stepping up to take over--the lunatics in charge of the asylum--which is why, at least when I explored this issue a few years ago, the quasi-official user guide contained actual insults directed at people who complained about their ads being inexplicably deleted. Craigslist, Facebook, Google, Youtube, are true capitalist monstrosities, entities that should not exist without human supervision, which have grown too large to be supervised, and are making too many billions of dollars ever to be reigned in. Google's hypocritical slogan is "Don't Be Evil", but children searching for "Peppa Pig" on Youtube can, according to a disturbing article I read this week, easily find troll-created knockoffs in which Peppa Pig drinks bleach or eats her father.

Bear with me on a third pathway into the subject matter: a critical problem in software development has always been authentication and trust: how does the system, and the other users, know you are who you say you are, when you order a product, transfer money, or direct a train to move? Compuserve had a much better idea in 1984 of my true name and identity than Facebook has today. To Facebook we are each a consumer: it knows what we buy, not whether we can be trusted. In creating a machine to make billions of dollars, the entrepreneurs have lost any interest in trust, which is an alien concept. This is made startlingly evident at every moment when you use the Internet these days, for example, by the ubiquity of clickbait ads on what should have been reputable sites: articles followed by links implying that a celebrity is dead, which when clicked take you to nonsequitur content and ultimately to an ad for, say, vitamin supplements.

The statement that the Russians "hacked" free speech leaves out a whole lot of information. A nascent free speech infrastructure, the Internet and online services circa 1984, about which we were very hopeful, was first hijacked by Microsoft, Apple, Google, Facebook and Twitter, and converted into a huge, cheesy, insecure platform for trolls and lies. The Russians' behavior--promulgating gross lies and conspiracy theories--was already the norm in American political campaigning; witness the "Swift Boating" of John Kerry. Our billionaires built the machines which made it possible; the Russians enthusiastically joined and used these systems, as there were no mechanisms of authenticity and trust, and no human guardians, to keep them out.