Merged All things Trump + Russia

[TofuFighter]

Then even saying that Trump had 0.1% chances of getting elected would have made then not-wrong

Correct. Now you're getting it. See how easy maths is?
If I say Palermo has a 10% chance of beating Juventus, and they do, you wouldn't argue that i was wrong.

Or maybe you would. Your sig suggests that you are the arbiter of who is right and who is wrong. There's a 10% chance that you're right about that.

 

[uke2se]

Correct. Now you're getting it. See how easy maths is?
If I say Palermo has a 10% chance of beating Juventus, and they do, you wouldn't argue that i was wrong.

Or maybe you would. Your sig suggests that you are the arbiter of who is right and who is wrong. There's a 10% chance that you're right about that.

I think you are seriously overestimating the odds. But then again, 30% apparently means "no chance", so 10% must be less than no chance.

 

[Dragon]

Then even saying that Trump had 0.1% chances of getting elected would have made then not-wrong

Good for you

Still not-getting it. Argumemnon addressed what might be said about giving Trump a small chance of victory. Nearly 30% is not small.

 

[McHrozni]

538?
The site which was giving Trump as almost surely probably losing at the elections?

The site that posted this four days before the election:

Trump Is Just A Normal Polling Error Behind Clinton

https://fivethirtyeight.com/features/trump-is-just-a-normal-polling-error-behind-clinton/

It turns out the article was right on the money. They did it again with the French election.

McHrozni

 

[NoahFence]

The site that posted this four days before the election:

Trump Is Just A Normal Polling Error Behind Clinton

https://fivethirtyeight.com/features/trump-is-just-a-normal-polling-error-behind-clinton/

It turns out the article was right on the money. They did it again with the French election.

McHrozni

I wonder what it was before comey vomited out his information about Clinton's emails?

 

[phiwum]

28.6%
Which is much less than 50%
Trump won
They were wrong

Are you new to probabilities?

An event may have a very small probability and yet occur. This is a simple fact.

From which it follows that if I says that an event E has probability 0 < P < 1, then E's occurrence or non-occurrence cannot show that I was wrong. In fact, no finite amount of replays of the same situation could prove that I was wrong (though it could serve as evidence that I was probably wrong).

 

[phiwum]

Then even saying that Trump had 0.1% chances of getting elected would have made then not-wrong

Good for you

Yes, that's right. If 538 had said Trump has 0.1% chance of winning, then we cannot definitively say that 538 was wrong. We can only say that either they were wrong or the improbable occurred.

 

[Tony Stark]

Saying that 538 is wrong because Trump won is like saying that I'm wrong when I calculate that there is a 25% chance of me getting heads both times if I flip a coin twice and then I flip a coin twice and it is heads both times.

 

[BobTheCoward]

Correct. Now you're getting it. See how easy maths is?
If I say Palermo has a 10% chance of beating Juventus, and they do, you wouldn't argue that i was wrong.

Or maybe you would. Your sig suggests that you are the arbiter of who is right and who is wrong. There's a 10% chance that you're right about that.

That Is only correct if you are being willfully ignorant of how models work. 538 was not right because they had the highest number. They were right because their conclusions from the model were consistent with reality. They will be the first to tell you the final number a model spits out is far less informative than the information it provides about uncertainty.

I watch MMA. A prediction that gets the winner wrong bit predicts the nature of a fight is more useful than a prediction that gets the winner right, but completely gets wrong how the fight plays out.

 

[TofuFighter]

That Is only correct if you are being willfully ignorant of how models work. 538 was not right because they had the highest number. They were right because their conclusions from the model were consistent with reality. They will be the first to tell you the final number a model spits out is far less informative than the information it provides about uncertainty.

I watch MMA. A prediction that gets the winner wrong bit predicts the nature of a fight is more useful than a prediction that gets the winner right, but completely gets wrong how the fight plays out.

My only suggestion is not to watch MMA.

 

[BobTheCoward]

My only suggestion is not to watch MMA.

Why?

 

[The Great Zaganza]

Russian Cyber Hacks on U.S. Electoral System Far Wider Than Previously Known
https://www.bloomberg.com/politics/...h-of-39-states-threatens-future-u-s-elections

According to this, Obama used the Red Phone to tell Kremlin to stop meddling in the election. Putin keeps denying.

At this point I am more scared that the Russian hacking was done freelance then if it was state-directed

 

[NoahFence]

Russian Cyber Hacks on U.S. Electoral System Far Wider Than Previously Known
https://www.bloomberg.com/politics/...h-of-39-states-threatens-future-u-s-elections

According to this, Obama used the Red Phone to tell Kremlin to stop meddling in the election. Putin keeps denying.

At this point I am more scared that the Russian hacking was done freelance then if it was state-directed

Well if that's the case why not have Putin help find the culprits and put this all to rest?

 

[pgwenthold]

Saying that 538 is wrong because Trump won is like saying that I'm wrong when I calculate that there is a 25% chance of me getting heads both times if I flip a coin twice and then I flip a coin twice and it is heads both times.

I do a lot with probabilities, and my usual approach is that I don't consider anything over 15% to be surprising. The analogy is if I had a 6 sided die, marked with one 1 and the rest 2s. Would I be surprised if I rolled a 1? Not really.

Otoh, if it is a 20 sided die marked with one 1, I would be surprised if the 1 came up. It's not impossible,of course, but so unlikely they you are surprised when it happens in a single roll (of course, if you roll it a lot, it will come up; it would not be surprising to see it once if you roll 4 times (20% chance)

 

[Belz...]

Otoh, if it is a 20 sided die marked with one 1, I would be surprised if the 1 came up.

Really? I wouldn't be surprised if a 1000-sided die came up with 1.

 

[Skeptic Ginger]

Add probabilities to the things nobody knew could be so complicated.

Yeah, about that, a trip to the horse races might go a long way toward teaching someone how probabilities work, that is if you can teach them how odds determine payout.

 

[The Great Zaganza]

Well if that's the case why not have Putin help find the culprits and put this all to rest?

What's obviously worse for Putin: looking like a supervillain or looking like he's unable to control his minions?

 

[quadraginta]

Saying that 538 is wrong because Trump won is like saying that I'm wrong when I calculate that there is a 25% chance of me getting heads both times if I flip a coin twice and then I flip a coin twice and it is heads both times.

And 538 had Trump's chances of winning higher than the chances of getting heads twice in a row.

"No chance", indeed.

 

[quadraginta]

Yeah, about that, a trip to the horse races might go a long way toward teaching someone how probabilities work, that is if you can teach them how odds determine payout.

If that was true then the bookies wouldn't make anywhere nearly as much money.

Or casinos, for that matter.

Repeat customers is where they make the big bucks. Not much learning there.

 

[Skeptic Ginger]

If that was true then the bookies wouldn't make anywhere nearly as much money.

Or casinos, for that matter.

Repeat customers is where they make the big bucks. Not much learning there.

I didn't say the house wouldn't always win. That's what payout odds ensure.

The probability of horse 1 winning is how the payout is calculated if horse 1 wins. Highly likely to win, payout is lower; highly unlikely to win, the payout is bigger.