Though in fairness, they put all the polls together and then crunch the numbers
This time though, it appears the numbers crunched 538

That was the difference. Not 538's fault. Again Hillary wins big with Democrats and Bernie wins with all the other non Democrats that vote in the Dem primary.

Nate's time is done. He is finished.

Got the order correct, and the % of the top 3 nearly exact.

They overestimated Rubio a bit, but had him last anyway.

garbage in, garbage out. The polls were wrong. Why is the question. But 538 is polls-based projections. If the polls are bad, the projections will be bad. This was a historic miss by the polls. Why is a totally different question. And I think understanding why will help us mitigate what happens in Ohio.

Their projections weren't at all inconsistent with the polls, it was the polls that were consistently wrong!

So I'm annoyed with the pollsters, but can't fault 538. They don't poll, they just put together the projections. It's not their fault the data was bad.

Can't fault Nate. He's just taking the data and calculating probabilities. The polls were wrong therefore 538 was wrong. He will now have to correct his models based on huge errors in the polls.

Fluke. Congrats to Sanders. On to the 15th.

Now if you never believe in another poll, I could kind of understand that. But primaries are very hard to poll - especially open primaries and sometimes polls are just wrong. It happens, but not often. I would much rather my candidate be ahead in polls than behind

unless they are unscientific

are at least a little art as well. They not only ask the questions, they estimate what the normal numbers of voters for each demographic will be.

Get that wrong and everything changes. We'll need to wait for the forensics, but it seems likely that they missed the cross section of voters and went with an inaccurate model.

It happens, and they'll correct for it in the future--if it wasn't a fluke. That's why we call it science, and flukes do happen when dealing with human beings.

Like weather forecasts. They take the best info they can find, compare to historical data and then make an educated guess about future events.

It's just math. In this case, the raw data was wrong. It might be that 538's system needs tweaking too. Primaries are always harder.

Do you think weather forecasts are unscientific when they are wrong?

They call 100 people on Sunday and Monday. 60 of them say they are Hillary supporters and are going to vote. On Tuesday 15 of them think, "Eh... they don't need me. The polls give her such a big edge. I'm going to jog instead, or.... or.... or....

I'll still look in on them--they do tend to learn from their mistakes.

And big picture, it's not like it is a rout. He gets 65 delegates, she gets 58.

She still won the evening as she got all the MS delegates. That's what, 36? And that's not counting the super delegates.

If, after all the money he hemorrhaged all over Michigan in terms of ads and staff, he didn't do well, I'd be quite shocked. He at least got more bang for his buck than he did in SC.

So...we slog on!

I f all the data is off, they'll be off. Not really their fault.