In a normal democratic election, candidates who win the most votes prevail. In American presidential elections, it’s vastly more complicated, and candidates who receive fewer votes can — and occasionally do – take office.

It’s against this backdrop that Joe Biden isn’t just focused on winning the U.S. popular vote, which is likely, but also winning the popular vote by a large enough margin to actually prevail over Donald Trump.

And what kind of margin will the former vice president need? FiveThirtyEight’s Nate Silver crunched the numbers and determined Biden’s chances of winning the electoral college vote based on the size of his popular-vote victory.

• 0-1 points: just 6%!
• 1-2 points: 22%
• 2-3 points: 46%
• 3-4 points: 74%
• 4-5 points: 89%
• 5-6 points: 98%
• 6-7 points: 99%

Nate added that a 2020 victory isn’t really “safe” for Biden unless he wins the popular vote by 5 or more percentage points. (The Economist’s G. Elliott Morris published a related set of numbers, which were slightly more favorable to the Democratic ticket, but overall pointed in a similar direction.)

In other words, if Biden narrowly receives more votes than Trump, there’s a 94% chance the Republican will win the election. If Biden defeats Trump by 2 or 3 percentage points, the former vice president would still likely lose, despite the will of the American electorate.

The Washington Post‘s Paul Waldman added yesterday, “Turnout projections are running at around 150 million this year (137 million voted in 2016), which would mean that if Silver is right, Biden could win by 3 million to 4.5 million votes and still have less than a 50 percent chance of becoming president. If Biden won by 4 percent to 5 percent, or 6 million to 7.5 million votes, Trump would still have a one-in-ten shot of prevailing.”

I bring this up for a couple of reasons. First, it helps provide a lens through which to see polling: the closer the 2020 surveys, the more likely it is Trump will win the election, even if he receives fewer votes.

Second, it’s worth occasionally pausing to emphasize just how indefensible the current system is.

Many Americans probably think electoral-college/popular-vote splits are incredibly rare and not worth worrying about. After all, most of the time, the candidate backed by the most voters actually becomes president.

But if Trump wins in 2020 after losing the popular vote, it will be the sixth time in 58 presidential elections in U.S. history. That’s a failure rate of nearly 10%, which isn’t that rare.

More to the point, however, if Trump wins after losing the popular vote again, it will mean three of the six failures will have occurred since 2000. Or put another way, there will have been six elections in the 21st century, and the candidate who failed to garner the most public support will have taken office in half of those elections.

I remember one prominent political figure declaring not too long ago, “The electoral college is a disaster for a democracy,” which is hardly an unreasonable point.

That prominent political figure was Donald Trump.