Now, there’s no reason in principle why a candidate who wins “only” 12 or 13 million votes in the primaries (or even just five or six or seven million) could not win a general election. That depends entirely on the candidate and his campaign. But the point here is that even a terrific primary performance offers zero evidence that a candidate actually can win a general election. As all of the head-to-head polling illustrates, it isn’t even a sign that that said candidate would perform better in a general election than other candidates who got fewer votes in the primaries.

This is especially true in Trump’s case. His hard-core supporters fail to comprehend just how deeply unpopular he is with everybody else outside their relatively small group. According to the last eight polls taken on the question, Trump has an unfavorable rating of between 60 and 70 percent among the general population that will vote in the 2016 election. He is not that much more popular than the ebola virus. (Although no virus has ever tried to run for president, so we cannot be sure.)

One can quibble with a poll here or there, but to deny that Trump would be the most unpopular person ever nominated for president requires the belief that all current polling is wrong — and not just a bit wrong (as some polls were in 2012 in 2014) but completely, uniformly and entirely wrong in a way it never has been in any modern presidential election. Yet in reality, the polls from April and even March of 2004, 2008, and 2012 were, on aggregate, reliable indicators of the eventual winner in those years.