Posts by BenWilson
Last ←Newer Page 1 2 3 4 5 Older→ First
-
Speaker: Confessions of an Uber Driver…, in reply to BenWilson,
Furthermore, it is only a matter of time before this happens in NZ. Then both Uber and our government will be in all sorts of shit. They have sat on their hands far, far too long.
-
Speaker: Confessions of an Uber Driver…, in reply to Ian Dalziel,
I have a feeling that this will lead to huge lawsuits in the US. There is no way Uber can distance themselves from this, when they have not followed any due process to ensure the drivers comply to the state screening processes.
-
Ah, so you mean they're showing up as closed in Google Maps?
-
Speaker: Confessions of an Uber Driver…, in reply to goforit,
Really? I wonder where they are now? Can you send me a screenshot? benedictwilson@hotmail.com
I notice that referral bonuses for new drivers have now hit $750. If they go any higher, they might as well just be paying for the compliance in the first place.
Suggests that it’s becoming difficult to get new drivers. Here’s a thought, Uber. Put the prices back up, like what all drivers have been asking for since April.
Edited to Add: Or you could post a screenshot here. You have to make a post, then submit it, then edit it. There will be a button for adding attachments.
-
On how people are seeing Trump, I have to say I really don’t rate my chances of doing it without numbers. I wouldn’t vote for Trump because he’s a clown, and I tend to agree with what Obama said last year, that the presidency is a serious job, and should be treated that way. I can’t say anything about Clinton particularly appealed to me, and under those circumstances it’s a bit hard to articulate why one should vote for her, other than the purely negative reason of keeping Trump out of power.
But how I feel about either of those points doesn’t help me understand where the visceral dislike of Clinton comes from, nor what the attraction of Trump is. He’s simply a mystery to me, the way that my enjoyment of watching MMA probably is to others. It’s just not my bag to watch his act, which makes a farcical show of the serious business of running a business, in the way that professional wrestling does of actual competitive fighting, and in the way his entire campaign has of the very serious business of running the most powerful country in the world. Presumably this is how he will run a war, or his foreign policy, like he’s in the professional wrestling as Mr Boss Man or something. As happened during the Apprentice, his post-production staff will have to work out how to spin why he fired someone so that it sounds rational, and it won’t wash with me in any way.
But clearly he’s built some kind of narrative that’s appealing to a large audience. He’s not even the first actor in the White House doing it, the cowboy actor did it first. Which direction it will now take is anyone’s guess, just like his show, on which he would fire the better person on a whim sometimes, just to show that he could.
What is that narrative? Ultimately the clearest thing, about the only thing you could say for sure, is that there’s going to be change. More importantly, I think Clinton was cast very clearly as the opposite, the status quo. She certainly never convinced me that anything would be different on her watch. Hell, it’s even the same actual person who was in the White House when I was younger. And her husband too. To me, it was fucking madness to have chosen someone who had already been there. How can you possibly convince people who lived through your last tenure, and then kicked your crowd out in favour of Dubya, that you’re going to be anything more than business as usual?
And clearly, business as usual is not what the Trump supporters wanted. Whether they will actually get it anyway remains to be seen. I don’t see professional wrestling having gone through any important structural revolutions in all the years I’ve seen it, despite the actors changing. Trump’s play acting is not going to be strong enough to make the enormously powerful institutions in the USA change how they operate, that will be down to the Republican machine, and is there any real reason to think they’re any different than they’ve ever been?
-
I can't really make up my mind about the ramifications of a Trump Presidency yet. But I'm nowhere near as surprised by it as I think most others are who really wanted Trump to lose. This is not because of some deep analysis of the underlying drivers that caused Americans to vote for Trump. It's purely because I've been watching the numbers closely, particularly on 538, and at no point have they indicated a shoo-in in the last few months. And the fundamental thing you have to understand about probability is that sometimes things don't follow expectation. Particularly a weak expectation, which is all we had by yesterday morning.
I chose not to look at the polls themselves at all. They're incomprehensible in such large numbers, and you get into all the silly arguments about which ones to trust, who is biased, etc. To get a better picture, you've got to aggregate them. You've got to explore the extent to which they are correlated in different states, and you have to model how they translate into electoral outcomes in a statistically robust way. I can't and won't do that, particularly not since 538 already does it very well.
And one thing that struck me all along was how close the race continued to be.
When the sexism scandals took off, those probabilities went steadily down for a while, but then tapered off at around the 10% mark. If the election were held right then I'd have felt much more confident. But interestingly, as time kept marching, the numbers didn't continue to drop, which I'd expect if a robust level of a 6 point margin was something that it stopped out at. Underneath, clearly, the support was rising. Then it began to steeply rise on the FBI interference. By the time the probability was sitting on 35% again, my feeling was that it was anyone's game. That's what I'd be feeling if it was the last few minutes of an All Black test with them one point ahead and the French suddenly get the ball.
Silver did a good job of explaining why his aggregations gave higher chances to Trump than the other ones out there. Every bit of that made sense to me. Particularly the correlation between the errors in the states.
It's a big mistake to think that the polling in the states are statistically independent events, and thus errors should even out across the large number of states, and thus a 3 point lead is unassailable in aggregate, despite being quite uncertain if you're only taking a single state into account with such a high number of undecideds. When a poll is getting it wrong, it's also known that the other polls are more likely to be as well.
So I don't see a huge methodology fail, except in so far as people can't learn to trust that when you're doing these aggregations, you really need to start trusting what the simulations are telling you, you can't possibly work this stuff out analytically, by hand. I had no interest in all this calculation of "pathways to victory" counts. Monte Carlo simulations work that shit out for themselves.
Yes, in the end, the polls were still wrong in their predictions. But the variance was so high in that predictions, that it's not really that surprising that the unexpected outcome happened. This is purely looking at the numbers themselves without any notions of what should have happened or how people should be seeing Trump.
-
Hard News: The Long, Strange Trip, in reply to Rob Stowell,
Goodbye TPP
Thank God there's a silver lining.
-
Hard News: The Long, Strange Trip, in reply to Carol Stewart,
Yeah, I saw that and had to chuckle. The phrase "mathematically impossible" is usually reserved for things like proving 1+1=3. Not making spurious statistical claims, in massive disagreement with the opinions of actual statisticians.
-
Hard News: The Long, Strange Trip, in reply to simon g,
There’s much more involved here than margin of error.
Can't say I'm convinced. The event was unlikely, but not outrageously so.
-
Hard News: The Long, Strange Trip, in reply to Tom Semmens,
First brexit now this – polling is completely broken in the post MSM age.
Nah, it's a couple of unlucky chances, and forget about all the times things went exactly as the polls said. If anyone can think of a better method, go for it. Hindsight will be 20:20, as usual. But that's not much use for predicting the future. Betting markets weren't doing a better job.