The last we heard from actor and Bernie Sanders mega-supporter Tim Robbins, he was disparaging Hillary Clinton's big wins in the South by pointing out that those states would vote for the Republican in November. (We pointed out that Sanders won a lot of red states, too, and that Clinton had also won more purple states. Oh, and that winning states in the primary isn't linked to winning them in the general.)

Anyway, Robbins is getting his thoughts out there again, this time on Twitter. For example, he tweeted this thing.

This is a very bad tweet.

Last week, we explained how exit polls work. In short, the pollster (Edison Media Research) conducts interviews at polling locations, collecting a ton of demographic information about voters. Since they aren't talking to every voter, they use statistical analysis of turnout, past and present, to estimate overall patterns -- including the likely winner.

Think of it like waiting outside a movie theater to determine what sorts of people watched which movies. You get data from a theater, and 60 percent of the people you talked to went to see Star Wars. Most of them were under 30 years old. Now, you know that this theater has more young people that go to it (since you're near a college campus, maybe), so you know that probably less than 60 percent of everyone nationally went to see Star Wars. Then you get actual box office figures, and learn that only about 45 percent of tickets sold were for Star Wars -- so you adjust your demographic numbers accordingly. Probably fewer young people than you may have estimated went out to the movies. But if you did this survey at 40 theaters in a hand-picked, diverse set of communities and most people were seeing Star Wars -- you could assume it would top the box office.

That's sort of how exit polls work. As actual voting comes in, the exit pollsters refine their estimates, since they now have a better idea of who actually came out to vote. That means that the numbers the minute that polls close are necessarily less accurate than those an hour later, once they can include actual vote tallies in their data.

And that's why the numbers in New York, for example, moved so much between 9 p.m. (when polls closed) and the end of the night. Lots of young people completed the exit poll, which made it look like Sanders was doing better than he did. That margin of error is why networks use exit polls in their decision to call a race -- but don't rely only on exit polls.

And that's why Robbins's tweet is bad. Or, really, that's one reason his tweet is bad.

It's also bad because thinking exit poll results trump actual voting results ("in reality, Bernie has won four more states") is idiotic. And it's bad because thinking that "the machines are rigged" in 19 states to Clinton's advantage (including a variety of different types of "machines") without anyone knowing suggests a conspiracy of truly impressive scale.

I mean, voting machines are not dropped off by DNC chair Debbie Wasserman Schultz from the back of a U-Haul. They're operated by cities and counties all over the place. The suggestion is, I guess, that there's a common software system that was somehow edited? That someone snuck into the warehouses where all of these things were stored and tweaked something? Or is it that boards of elections in a dozen states agreed to lie about the results?

The irony of theories like Robbins' is that it's actually far more naive to assume a giant conspiracy that goes undetected than it is to understand that these are complex systems that simply by virtue of scale resist the sort of tampering he alleges. (Voting machine fraud is the Democratic version of the voter ID fraud lamented by Republicans: Responsible for all problems but lacking in any widespread examples.)

People pushed back on Robbins, so he tweeted at a ton of them -- 88 of them, over the course of an hour, in fact.

Everyone who disagrees with him is an operative of the pro-Clinton Correct the Record. Okay.

Tim. I get it, man. You're mad. You want Sanders to win and you don't get why he isn't. It's hard for you to believe that people would vote for Clinton over Sanders, so you assume that there was somehow widespread fraud, and you're willing to assume a huge conspiracy and misunderstand how polling works to reinforce that belief.

It's simpler than that. Sanders has run a good campaign and done a staggeringly impressive job of getting his message out, but, thanks largely to big margins of support from older and black voters, Clinton has beaten in him in a lot of big contests, building an almost-certainly-insurmountable lead.

Not the script you wanted. But it's the movie that's playing.