Welcome to democracy; Take II

Or: Predictably unpredictable.

Source: The Guardian
It was about five months ago that I posted some post-Brexit-vote comments about the reaction, and while I'm even more annoyed by the recent US election results than Brexit, now in the post-Election period it seemed worthwhile to revisit.

First, going over the old stuff:

STILL what not to do when a population votes opposite to how you'd like:
Yes, it happened again. Technically it's asking for the electoral college voters to not match what turned out on election night, which isn't quite as bad as just asking for another vote, but I still don't see why 6 million votes in a petition should count over the 120(ish) million votes during the election.

There's a lot of problems with the electoral college system: all-or-nothing states, no instant-runoff, unusual weighting system, ... but that was known before the vote started. Note: if not calling for changes to past elections, it can still be useful to call for changes to future ones. But if you're only asking to change because your preferred candidate lost, that doesn't help much. Please understand that any change may cause preferred candidates to lose in the future, and instead advocate for change because it leads to a better representational system (which proportionality and IRV do, I believe). Plus advocate for it continually, not just after an election.

This point is almost verbatim from the previous one, except if anything, there's way more examples of Trump supporters being called things, so I won't bother linking them. You won't convince people by calling them stupid, and they're less likely to discuss politics with you later too.

Source: The Mary Sue

-) Don't make predictions about the economy... 
Ok, this one needs a bit of a change. I mean, it's still valid (there are a lot of predictions out there, both about how it'll be great and terrible), but it's better to change this to:
I feel like Brexit already made this warning, but I guess not. Note in particular the Huffington Post prediction of 98.2% chance of winning. If you recall the Bayesian reasoning I mentioned in the previous post here, it's possible that Huffington Post were right and this was a 1-in-50 event. More likely though is that Huffington Post are bad at predicting.

One thing worth adding here: for anyone sad at being misled by polls, perhaps you can more understand the sentiments of Michael Gove regarding having had enough of experts. If groups of stats nerds experts like the Princeton Election Consortion (Princeton!) predicted Clinton had a 93% chance of winning, how much can you trust them on other stuff? Especially given that elections have waaaay more polling and sampling than the average question like "is this tax good?". [answer: you can trust them on average. but once-off decisions are not on average, and you are more likely to remember the ones that the experts got wrong]. Prediction is my thing (e.g. this post), so I'll stop myself before I go off on a tangent, but for those interested in expert vs layman prediction, SSC had a good thread on this recently that's worth reading.

By the way, maybe there's a positive here: If your prediction that Trump would lose the Election (and/or GOP primaries) is based off the same facts/axioms/reasoning/intuition/... as your prediction that life under a president Trump will be terrible, perhaps the latter is just as incorrect? Probably not, but it's nice to have hope to cling to :)

Source: The Daily Blog
Note that the things-to-do still apply here too, so instead I'll use this space to make two more observations that were even clearer after the US result:

-) The filter bubble not only applies to who you interact with, but also what they talk about.
Over 45% of American voters picked Trump. If 45% of your social network feed of choice aren't people who'd vote Trump, then it's not representative. There's one possibility that's sometimes missed: X% of your network being Trump supporters doesn't mean that X% of the political content you see will be pro-Trump. Maybe they just share less in general? Maybe they don't want to burden their opinions on others? Maybe they're worried that claiming to support Trump will end up in ridicule / ex-communication from their friends? This last one is a problematic thought, but has been hypothesized as to why polls were so wrong: people only vote Trump in the anonymity of a polling booth.
Note that this applies to real-world networks, too. Were 45% of the news articles you read pro-Trump? Were 45% of the guests on the shows you watch pro-Trump? The actors you watch, or writers you read, if they're not 45% pro-Trump perhaps they're not the most balanced source of opinion?

-) Don't try to simplify the issue so you can blame it on one thing.
Back on the online social-network thing, apparently Facebook is why trump won. Or maybe it was the FBI. Or white people. Especially if they're working class. Or....rich people? Or Maslow's hierarchy (lol). Or DNC arrogance. Or political correctness. ... I could go on, but thankfully it looks like CNN have already cataloged 24 different reasons. Sure, they're easy to consume, present a simple message and result in a comforting conclusion that it's not your fault, and we know how to prevent future Trumps. Except...not. It's a complex mix of all of the above, and more. And progress for most of these will take a long time. At least it's an ok starting point for deciding what to do though - go through the potential problems, figure out the probable benefit:effort ratio, then spend effort in whichever is the best?

Aaaaanyway, enough of politics. Exam season is approaching at UBC, which means lots of final projects, but also snow season. Unfortunately it started with a week-long flu (I know I just got older, but age shouldn't hit me that fast), but now that's improving, exam study period does also mean more free time, hopefully more blog posts, and an Xmas trip back down-under :)

Comments

Post a Comment

Popular posts from this blog

Perceiving Frequencies

Project cutting floor

Sounds good - part 1