Social media outrage is driven by something like a base rates fallacy
Consider the following two statements:
Statement 1: “80% of incidents of X are done by bastards.”
Statement 2: “80% of people who have done X at some point are bastards”
If you want an example of X, try “getting into conflict with service staff.
The two statements do not imply each other, because it could be, for example, that while bastards fight with service staff numerous times and thus commit a solid majority of overall offenses, many non-bastards, perhaps even a majority, have got into an argument with service staff at some point in their long lives. The point that people get mixed up between these things is well-established in the cognitive science and behavioral economics literatures.
But let’s say you know statement 1 to be true. If you see someone fighting with service staff in naturalistic conditions, and without any distorting factors or selection bias, that would be, on this evidence, a reasonable indication that the perpetrator is a bastard (in fact, an 80% chance).
Suppose though you had an fightingwithservicestaff-o-scope and you could point it at someone, and if they’d ever fought with service staff, they would come up glowing red.
The internet, both in what it stores and in the collective memory it networks, is sort of a generalized bad-deeds-O-Scope.
People assume that learning something bad about a person by the internet is just as damning as learning it about that person in real life, because, hey, why wouldn’t it be? It’s the same information, just a different channel. Actually, the channel through which you learn information should shape how it alters your evaluation of their character, even if there is zero doubt as to the accuracy of the information.
We all have moments in life that don’t speak highly of us. I’d say for most of us, the worst of them is probably far worse than “fought with service staff”. Getting a sense of someone’s character can be seen as a sampling problem- what proportion of their moments are bad moments? It’s a huge problem if your sample is biased because there are very few people who wouldn’t look bad if the only thing you knew about them was their or even one of their worst moments.
If you run into bad moments in another person’s life “by accident”- say, by watching them do that-, that’s an indication they probably have many more such dark moments- what are the odds that you just happened to be looking at them during their worst moment? On the other hand, if you run into a dark moment of theirs on the dark moment aggregator, that gives us little evidence of anything [unless it’s unambiguously bad: rape, murder, torture, that sort of thing.]
So why don’t we cotton onto this and stop making this mistake?
The main reason is that cognitive biases make us not notice base rates etc.
However, I also think there’s a secondary factor: People genuinely do have darker secrets than we used to believe before the internet and it is taking us time as a culture to adjust to that.
