Shouldn’t that be “smoking pop gun” ?
My guess is the voters will care less about the promise than the vote – and ACA is actually pretty popular now that it’s out there.
As to the polls only vs added fundamentals discussion - the results in Scotland argue some for an appreciation of the limits of the polls only method. Of course which historic fundamental to have added is only evident in retrospect (“undecideds tend to swing to the status quo”; the “shy Tory” effect; etc.). The polls only approach predicted the correct answer but the margin was way outside of what the +/- bars would have stated was remotely likely.
Wang addresses the specifics here. He says “In fairness, Silver has reasonably called out my pre-2012 writings, in which I mistakenly interpreted the sharpness of the Election Eve snapshot as meaning that sigma_systematic<<1%. His factual error is in supposing that in 2012 and 2014, I am making the same assumption. I am not.”
That is all to the good, and I’m glad he addressed the problem. The decay over time in the predictive power of statistically grounded fundamentals still remains an open issue.
IMO, what we need to do is persuade one of the experts on forecast evaluation to set up a website and assess the performance of these various modelers, then devise a system of bets moving forward. Also, clearly articulate the lower power of the devised tests, given low effective sample sizes (by year) and within year cross correlation (by race).
I don’t think Wang had a Gauss problem. He isn’t claiming black swans in his quote upthread: he’s claiming mis-modeling. Again, all to the good: this is how science progresses.
You might want to think again regarding weak sauce. If Nate makes one hundred 5 to 1 picks, you will expect 20 to come up, “Wrong”. If it’s much fewer than that, then his modeling is off. There are a number of ways of bringing the odds closer to 50%: it’s possible that this is all his fundamentals are doing. It will take a while to tease these issues out. Geeks like myself enjoy the process.
Chance of Republicans taking over Senate:
Sam Wang: Still holding at 30%.
538: 54.8%
Upshot: 56%
Note that upshot now exceeds Nate’s forecast by an insubstantial margin. 538 is now at the median.
MfM labs: 47%.
Past forecasts: 59%, 45%.
We’ve had this conversation before … I can’t find the exact thread I am thinking of but this portion of this thread captures some of it. In fact, as I recall, that was pretty much exactly the circumstance with Nate’s Presidential results … he was inside his error bars much more than he should have been if the model wasn’t, as you opine, “off.” It was discussed more here. But we had discussed it more elsewhere too that I not finding right now.
The interesting thing to me is the rating using the Brier score. Nate did poorly on that because that model punishes excessive hedging, discounts predictions with error bars that prove much wider than they turned out to be. Wang did well on that one but unless I misunderstand that scorecard punishes too narrow of error bars (Wang’s apparent flaw) much less. Explanation of the Brier scoring system here.
It will be interesting to judge the two this time not by who holds control, but by how well each of their error ranges holds up. Is Wang only outside his 1 sigma bar 32% (or thereabouts) of the time? Is Nate inside his not far in excess of 68% of the time? If neither (i.e. if Wang’s bars are too narrow and Nate’s too wide) which is the worse error for a modeller to make?
Oh, MtM - if you add in WaPo’s 65% then 538 is now barely above the mean. 
One of the differences in the Senate control numbers at this poiint does seem to be based on how an Orman win is dealt with. Wang scores that as not GOP control and assumes he will caucus with the Democrats while Silver puts that at 75%. Quite a few of the scenarios end up with Orman shifting the balance and how that result is handled does impact the numbers quite a bit.
How can they do more nothing that the nothing they are already doing?
Apparently HuffPo (currently running 56% odds of GOP control) places him as only 50/50 odds of Democratic caucusing in the case his decision tips the balance, as does DailyKos. NYT’s The Upshot puts it at 100% Democratic like Wang does.
Silver also put discounts the polls in Kansas (which give Orman a lead) by “just” 15% based on fundamentals like “Kansas is a very red state, and this is a somewhat Republican-leaning year.” Which may be favored based on his “subjective feeling is that the race is still more like a true tossup” - sorry but at this point that “fundamentals” correction factor IS punditry. OTOH giving 100% probabality to Orman caucusing with the Democrats in the case his decision alters the balance of power seem a bit too high. On the other other hand (where is a prehensile tail when you need one?) any number is one made up without any actual data to support it and Wang’s presentation is both clear on how he handles it and allows those interested to figure out how many scenarios would be changed based on that assumption. (His “nowcast” - snapshot view - puts about 34% of all outcomes giving Democratic control based on the assumption of Orman caucusing with the Democrats if that decides control. In the metamargin view putting Orman at 50/50 changes the metamargin from D+1.8 to D+1.4 and would have had it in GOP territory for most of August.
Orman said that he would expect whoever he caucused with to get rid of their current leader. Who is more willing to chuck their leader, the Democrats or the Republicans? How attached are Dems to Harry Reid? McConnell strikes me as like Boehner, the Tea Party would be glad to chuck him if they could.
Somehow I don’t think the Tea Party kicking out leadership less extreme than they are will convince him to caucus on that side. The only reason he has a chance in Red Kansas is that the TP has fractured the party in the state and driven many moderates to look for an alternative short of voting for the Democrat. He’ll ask for Reid to step down as his price but accept something less and hold his jumping ship at anytime over the next two years as his assurance of follow through. The GOP can win him over only by taking concrete steps to stop pandering to the TP. And you know, if that happens then I can live happily with GOP Senate control.
Nitpick: 5 to 1 odds = one chance in 6. So you’d expect 16 or 17 to come through.
For those that are interested, Sam Wang has a fairly easy to follow post explaining how he calculates race probabilities.
No special sauce or hallucinogens.
There is a threat to the system. Lets say that there is a grandfathered policy that doesn’t cover jack shit and because of that grandfathered policy, that insured is not subject to penalties for failure to carry insurance, then they get into a car accident. The system soaks up that cost unless they can get the insured to pay.
Not as much of a problem as you think. Most insurance covers catastrophic costs, except for those mini plans they give college students and some fast food workers. The reason most plans don’t qualify is that they don’t cover a lot of small stuff that the government wants covered.
It’s more akin to making everyone in a condo pay for cable so that it can be cheaper for those who want it. If everyone buys maternity coverage even though they’ll never need it, it makes maternity coverage cheaper for those who will. But the promises made were that people could keep their insurance. He did not tell 50 somethings that they would have to lose their coverage if it didn’t include maternity coverage.
All I’m asking is that we keep the promises we made to every American: if you are satisfied with your health insurance, you can keep it. It makes no damn difference if Barack Obama is satisfied with your coverage.
Then why have you been so loudly hoping, for so long, for the repeal of the entire thing? 
Is it that you now recognize the reality that it works and is popular and hoping to get rid of it is a bad idea politically too? Is that the last ditch you’ve backed into now - hoping for the government to force the insurance companies to do something they’ve never been required to do and have never committed to do?
Time to accept a little responsibility for your statements here. High time.
Dems are now claiming an edge in the Iowa Senate race, based on a 2-1 advantage in absentee-ballot requests. (The polls are still neck-and-neck, though.)
Democrats are less motivated, I would expect they’d have more people who don’t want to actually leave the house to vote.
That race is as tied as they get. Which means I favor Ernst, because Republicans are more reliable voters.
When can we change the title of the OP?