3 ms·
It’s often overblown but Bayesian statistics shows you should ideally update your mental model with every single piece of new information. It’s unrealistic to
by Retric 4y ago
It’s often overblown but Bayesian statistics shows you should ideally update your mental model with every single piece of new information.
It’s unrealistic to actually carry that out in every case, but in cases where you have zero evidence updating your views based on the first bit of evidence you receive is absolutely the correct choice. The degree to which any single piece of evidence should change your mindset is of course minimal, but holding a view with zero supporting evidence in the face of multiple pieces of counter evidence is a sign of insanity.
- seti0Cha 4y agoWell I did minimally adjust my priors based on that evidence actually. The fact that a lot of people are making claims here also adjusts them very slightly (I already knew many people felt this way). But I've got heavy weighting on policy matters needing non-anecdotal data, so not much of a change really.
- Retric 4y agoThe statistics are available ex: In the United States between 2005 and 2020, of the 42 nonfederal police officers convicted following their arrest for murder due to an on-duty shooting, only five ended up being convicted of murder. By comparison: In 2018 70% of US murder cases resulted in conviction.
- ninkendo 4y agoThat would make sense though, right? We literally make officers carry guns and tell them “use this to defend yourself, and as a last resort against dangerous suspects when the situation warrants” as part of their job description. It stands to reason that the likelihood of a cop using their gun in a “legal” way is higher on average than the average person who fires a gun. Even in cases where an officer does get arrested on suspicion of murder, it’s less likely to be convict-able than average.
- Retric 4y agoExcept there are many steps before a trial where a cop legally shooting someone won’t end up on trial. Not every murder trial is going to be of a guilty person, but someone going on trial for murder suggests there is significant evidence they actually committed murder. Conviction rates of 12% vs 70% show a very strong bias.
- ninkendo 4y agoIf there’s a selection bias causing you to only ever see evidence pointing in one direction, you probably shouldn’t update your priors. Cops doing their job in a routine way doesn’t make the news. Cops taking steps to de-escalate things and avoid killing someone doesn’t make the news. Cops killing someone (justified or not) typically does. If every time you see a news report of a police officer killing someone, you come away thinking “police must just be killing people left and right”, you’re forming an inaccurate conclusion. If you get your news from anything that’s driven by attention/clicks (that is to say, approximately everywhere), it’s a doubly inaccurate conclusion.
- Retric 4y agoThis isn’t about the rates that cops are committing crimes, it’s about cops getting more leniency from the system than regular people. There aren’t that many documented cases of cops shooting people without justification, but in those cases you often find instances of cops covering up for other cops. That’s evidence not just for these cases but also otehr cases which where successfully suppressed.