It's the Algorithm, Stupid!

Like it or not, as everything goes digital, algorithms (software rules) will make more and more decisions for us. Every now and then, those rules will misfire. Note that this is totally different than good old bugs in the software; the issue is with the algorithm not being appropriate in a specific context or time period. Let’s take 2 very recent examples of this problem.

Remember that famous 1972 pic from the Vietnam War of a naked, 9-year-old girl fleeing napalm bombs? When a Norwegian newspaper, Aftenposten, posted that pic on Facebook (in the context of the horrors of war), Facebook’s algorithms removed the pic because it was (mis)identified as child porn. An honest mistake, obviously. But Aftenpost’s editor, Espen Egil Hansen, wrote an open letter to Mark Zuckerburg:

“Please try to envision a new war where children will be the victims of barrel bombs or nerve gas. Would you once again intercept the documentation of cruelties, just because a tiny minority might possibly be offended by images of naked children, or because a paedophile person somewhere might see the picture as pornography?”

Ok, Hansen, let’s roll back the rhetoric. After all, it’s not as if Mark Z personally pull down that pic on Facebook. And, as Hansen himself says later in his letter, to take or not to take offense varies across cultures, religions and regions. Citing the Mohammed cartoons example, Hansen rightly says:

“It was – and remains – different in Oslo and Karachi.”

Exactly. So how can Facebook possibly make a call that works for everyone?!

Personally, I think Facebook screwed this up. But I don’t see any solution: such things will happen now and then. If Facebook screws up too often, market forces will punish it: that’s the way the public and capitalism work. Europe repeatedly has trouble understanding this basic principle, tends to get on its moral high horse, and doesn’t mind if countries (as opposed to companies) censor:

“Under the proposed law, the "site manager" of Italian media… would be obliged to censor "mockery" based on "the personal and social condition" of the victim -- that is, anything the recipient felt was personally insulting… Truthfulness is not a defense in suits under this law -- the standard is personal insult, not falsehood.”

Apparently Hansen is OK with views being suppressed, just not when it’s done by Facebook.

The second example of algorithms creating bad press was after the recent explosion in New York. Surge pricing followed; and people accused Uber of profiting from a bomb explosion. Quickly, Uber deactivated the algorithm for that area. As a wag on Slashdot suggested, perhaps Uber could have tried putting a positive spin on the episode:

“Headline should read "Uber Increases Driver Pay to Help Meet Emergency Demand."”