US Politics Mega-thread - Page 5889

On November 03 2016 23:29 Acrofales wrote:


If you don't see that there is a difference between "if you don't do X I will shoot you in the head" and "if you don't do X, there is a 0.00000001% chance I will shoot you in the head" I don't know how to continue the discussion further.

Yes, your moral compass is "forcing" you to vote for someone you don't like. No, it is not coercion in any meaningful sense of the word.

On November 03 2016 23:32 KwarK wrote:
Igne, your freedom to choose isn't in any way limited by external consequences of those choices. Those are what you are choosing between. If it were then no freedom would exist because any choice will by definition have external consequences. At this point you might as well argue "How am I free to choose between a burger and chicken nuggets if choosing the burger means I won't get chicken nuggets?!?! If that's freedom I'd rather choose death!". The consequences of the choice are what the choice entails, not only do they not make it coercive, they need to exist for the choice to even exist.

On November 03 2016 23:38 IgnE wrote:


so are you saying votes are worthless? or are you saying that russian roulette isnt suicide?

On November 03 2016 23:44 Stratos_speAr wrote:


This is simply not true.

There is a difference between the consequences of executing a choice (what your actions actually do) and the coercion forcing you to choose one in the first place or die.

There are countless precedents where people were coerced into doing illegal things and then exonerated from doing them because their only other choice was to refuse and then die. There are very few philosophers (or legal experts) that would argue with a straight face that choosing between A and B, but being killed by someone if you choose B, is an ethically meaningful choice.

On November 03 2016 23:50 KwarK wrote:

What you're describing here is the difference between meaningful choice and choice. Obviously with a gun to your head while you may still have choice a reasonable person would not say that you have a meaningful choice and therefore you are being coerced. The problem is that Igne is attempting to exploit the fact that there is no absolute defining point at which it stops being a meaningful choice and starts being a coerced choice and conclude from that that all choices are coerced and that humans have no choice and therefore people shouldn't vote. That's the issue here and that's why my counterpoint works.

Igne is arguing the very existence of consequences, for example oppression of trans people in the event of a Trump win, makes voting for Hillary a coerced choice. The problem with that is that it doesn't meet the reasonable person standard. And Igne knows damn well it doesn't so he's ignoring that and presenting it instead as "all consequences make choice coercive" instead of the commonly understood "in some cases a reasonable person might think the consequences make the choice coercive". I illustrated the absurdity of his point with my chicken nuggets example.

Obviously I believe in coercive choices. That does not mean I accept that all choices are coercive, which is the argument Igne is making.

I guess there's a few questions I still have. If you have two different types of probabilities, one of which is sort of summarized as "if an event has an X % probability, it means that out of 100 trials we would expect that event to occur X times," and the other summarized as "if a proposition has X % probability, it means we can say that proposition is true with X % certainty," that makes some intuitive sense. I guess the latter is still a little unclear to me in exactly what it means; if we say in the Bayesian sense that we're 75% sure Hillary Clinton will win the election, my only intuition for explaining what that number means is to imagine 10,000 parallel universes, and then in 7500 of them, Clinton becomes president. But that's the frequentist approach; It might be that the true result of the election has very little variance, but we're just fairly uncertain what the result will be. So due to our lack of information we estimate a 75% certainty, but if we checked in on our 100,000 possible universes, all 100,000 would go to Clinton. I imagine I'll have to read through some of the philosophy of probability wikis you linked to figure out why that 75% isn't a relatively arbitrary number, then, if you take the frequentist definition away.

The other question I have, though, is how to judge what possibilities are factoring into a probability like that. In 538's analysis they've done plenty of talk basically explaining why the model says what it says, and they'll often reference things like "it's accounting for the possibility of a large polling error" or "it's accounting for the possibility that the race will tighten in the week or two before the election." Where exactly do you model factors like that in a Monte Carlo simulation? It would seem very difficult for a model which gets a lot of its work done by averaging polls to account for the likelihood that those polls will change by x amount in one direction or the other by such and such date, and it would seem near impossible for a model with an input consisting almost entirely of polls to calculate the probability that those polls are wrong by x amount in one direction or the other.

I mean, put it this way: it would be pretty easy to come up with averages and t-distributions of polls in individual states (easy enough that I could probably do it). You could then randomly generate results based on these distributions that would, in a frequentist sort of way, mimic the "10,000 parallel universes" idea to generate probabilities. If you did this with all the states, randomly generating an election result for each one based on your distribution, and then adding up the electoral votes of all of them, and then repeating that 10,000 more times, you'd get an estimate of the likelihood taht each candidate would win the race. But this would be kind of like saying "well Trump only has a 65% chance in FL, 72% chance in OH, 33% chance in Pennsylvania... and he needs to win all of those, so I'll just multiply .65 * .72 * .33 ...and see? he has a tiny chance of winning the election!" But as 538 has often pointed out, you can't do that because their errors are correlated, so in the scenarios where he wins Ohio, he's more likely to win those other states, too; not to mention this doesn't account for the possibility of new events moving the polling average this way or that before the election. I don't know how you could just take poll numbers as input and use that to calculate the probability of those polls systematically underestimating one candidate; surely it's impossible to discern that from those data.

I guess this is all to say that the 538 forecast boasts the ability to state an absolute probability of the election coming out one way or the other. And in an absolute sense, it's at least true that when the forecast is above 50% that candidate is more likely than not to win. It's also true that if one candidate was at 75% last week and they moved to 85% this week that means their position has improved.

What doesn't seem clear to me is that if you were a bookie, and you gave people odds based on the 538 forecast for the next 100 elections, that you would come out even most of the time. If you look at 2012, for instance, and see that 99% of the time the candidate he said was winning went on to win, then unless he was giving those candidates a 99% chance of winning, isn't that actually a failure of his model? If most of those winning candidates were only getting 60-80% from the 538 model, and then 98% of them won, then surely he was actually underestimating those candidates quite a bit – or am I thinking of all this wrong?

On November 03 2016 13:46 ticklishmusic wrote:


i skimmed the article and i have no idea how economist is getting the "this is a mess" - maybe because of the use turmoil? i mean, the author is right on all counts but this is all stuff people know and conclusions that have already been reached.

i've said before, any big new federal program has some hiccups. we didn't get social security, medicare, etc. right on the first try, and these programs need tune ups over time. however, the difference is that back in the day both parties would get together for the common good and say "hey you know we need to make some adjustments to make this work". instead, we have an intransigent GOP that's tried to repeal the ACA god knows how many times and has offered nothing remotely reasonable as an alternative, let alone an improvement.

On November 03 2016 23:54 oneofthem wrote:
your use of the word coerce is not proper. it is not a coerced vote but a limited choice set, or limited outcome tree.

that's it.

On November 03 2016 23:54 IgnE wrote:


thats not the fucking argument i was making. theres no such thing as a "reasonable person standard" inherent to the english usage of the word. you are importing that made up notion.

On November 04 2016 00:01 IgnE wrote:


what is the difference then, judge? if trump said (and had the power to enforce) a choice between voting for him or making sure i never get a mortgage loan for the rest of my life (pretend hes a monopoly banker) would that be coerced? what if he threatened preventing my sister from getting a loan?

now imagine you hate hillary and trump and would rather vote for neither but hillary says vote for me or a fascist who will deport all muslims will take power. thats not coercion?