The Problem with Algorithms
Social media can be stupid, but using algorithms just makes it far worse.
I do this somewhat sheepishly, since the motivating complaint is kind of trivial and petty, but the underlying issue is real.
So I got this today:
This is fantastically stupid for at least two reasons. Any idiot who understands American English would never think that “knock yourself out” is any indication of an impulse towards self harm. It’s a flaming joke. Algorithms do not get humor. I guess I missed the opportunity to have them review it, but in my experience, that does absolutely nothing. This is idiotic.
Second, I’m no expert, but why is responding to a person who is considering self harm by cutting them off from a communication medium they are using a good idea at all? I should think that this response would be as likely to push a person over the edge. Apparently it has not happened, at least not that anyone has yet documented, or we would have heard about it.
This article is from 2013, so maybe their current, idiotic approach is an overreaction to what they used to do, but it definitely is not an improvement. Twitter, of course, it’s “aw shucks” homey approach notwithstanding, is as much a large, soulless corporation as Facebook, Exxon, or the Catholic Church. It’s not at all clear why running such corporations turns people into amoral automatons, but that has been common since the emergence of large corporations in the 19th century.
I was pleased to see that my web search for “racism algorithms” produced a number of responsive articles, including this important piece. That algorithms reproduce the racism in our culture could only surprise the sort of smart, geeky math types who produce them. Anyone who knows anything about history or culture would expect that outcome. We know, I know, brain researchers know, that our most important, motivating thoughts are subconscious. We can become significantly more aware of our subconscious thoughts with Buddhist meditation, but few people do that, and anyone who grew up in the United States did so with a constant barrage of racist images and meanings that are impossible to resist or avoid.
Even people who rightly see themselves as adamantly opposed to racism, who want to eradicate it from the earth, as all thinking people do, still have racist images and tropes in their minds. It’s inevitable.
Stupidly zapping people from Twitter is not racist, but it is an indication of how reliance on algorithms can produce unintended outcomes and why no one should rely on them uncritically.
The more specific, bigger problem, is the racism built into facial recognition technology. I have a serious problem with using it at all, apart from any racist outcomes. The racism only compounds the problem. It’s a very deep point that the Founders never thought of and did not state explicitly, but it is definitely a fair reading of the Constitution that it assumes anyone engaged in criminal prosecution will have some skin in the game as a way to reduce the propensity of prosecutors and police to persecute people. I think criminal defense lawyers should contest any case that relies on facial recognition technology on due process grounds, as well as equal protection where they can show invidious discrimination.
We should push back on the use of algorithms in every instance, from trivial, stupid Twitter decisions to racist criminal prosecutions.