The big problem with that Gallup poll that shows Donald Trump's re-election numbers equal to Barack Obama's

Wait, WHAT? The narrative of Trump's first 15 months in office has been that he is at historic polling lows. Then this.

What gives? In an attempt to suss that out, my friend -- and resident CNN big brain -- Harry Enten and I exchanged a bunch of emails trying to figure out the Gallup result and what, exactly, it tells us about Trump, Republicans' chances in 2018 and Trump's own fortune heading into his 2020 reelection race.

I've reproduced that email conversation below.

Cillizza: I was struck by this new Gallup poll -- or at least the headline: "Trump's Re-Elect Figures Similar to Those of Obama, Clinton."

Trump is at the same place as Obama and Clinton! That's remarkable, given that he has had the worst 15-month polling start of any president in the modern era.

Then I read the fine print. Yes, 37% say that Trump deserves re-election which is the exact same number who said that about Bill Clinton in 1994 and Barack Obama in 2010.

BUT, the numbers for Obama and Clinton are from October 2010 and 1994, respectively. That's after a full year of Republicans spending millions of dollars to campaign against them. We are only in April for Trump -- and the fall campaign is at its earliest stages. Virtually no ads attacking him have even run.

So, apples and oranges, right? Or wrong? And what else did you see in the poll?

Enten: Thanks for the email. I think you're on the right track with thinking this poll result isn't all it's cracked up to be.

Gallup itself released numbers closer to this point (late April of the midterm year) for both Obama and Clinton. Obama was at 46% in late March and 48% in late May of 2010. Far above Trump. Clinton was at 46% in late March and 40% in late April of 1994.

So, I don't know why Gallup did what it did. Of course, one could argue that because Trump's approval rating has been so steady that it will be in a similar position later this year, but I don't like to make assumptions like that. Who knows what all the ads Democrats will run this summer and fall will do to his ratings?

Does that make sense? And I also think there are larger problems with this question, if you're interested in hearing about them?

Cillizza: OK. We agree on that one: The Gallup poll is making an apples-to-oranges comparison.

And, yes, tell me what else you don't love in the poll.

One thing I don't want to lose sight of amid all of this: In 1994 and 2010, the president's party took an absolute beating at the polls -- losing control of the House in both elections. So, even if Trump winds up at 37% in October -- like Clinton and Obama -- that's not great news for Republicans.

Enten: Re: your last point, I think you're 100% correct. The reelect figure is really a stand-in for presidential approval. In the latest Gallup poll, Trump's at a 38% approval rating. Based upon history, that, on average, points to a net seat loss in the high 30s. Democrats only need a net gain of 23 for control.

So perhaps biggest problem I have with this question is it doesn't make it a choice between a Democrat and a Republican. There's a good chunk of voters who belong to the same party as the President who are thinking to themselves that they want a different person representing their party, but will ultimately vote for their party's nominee.

If you're truly interested in gauging the presidential matchup at this point, I think it's better to ask a question which forces the respondent think about a Democrat and a Republican. At the very least, it gets us beyond simply the incumbent. As it stands, this question is really just another way of asking about approval and disapproval. So what's the point?

When we ask about a Democrat vs. a Republican, it's when we see how unique Trump's standing in the polls really is...

Thoughts?

Cillizza: If by "unique," you mean "historically bad," then I am with you ;)

I think the country was significantly polarized before the 2016 election. But that election coupled with the fact that Trump won and immediately began governing in ways to please his base (and not a lot of other people) took that polarization to new heights.

The impact of that rush to your partisan extreme on polling is that almost EVERY question you can ask essentially becomes a proxy for this one: "Do you like President Trump or not?" Which makes nuance in polling -- and how it informs opinions about Trump -- very hard to divine.

Enten: I think there's something to that. If you're a Republican, you like Trump. As I've written about, the Republican Party is Trump's now. Heck, take a look at where Trump's approval rating is with Republicans. It's in the mid 80s (very close to the re-elect question). It basically makes it impossible for there to be a primary challenger who has any shot of taking Trump down.

But to your larger point, I think in order to fully grasp the historicity of Trump's bad ratings, we need to look at how he's doing against potential Democratic nominees. Then we need to compare that to how other presidential incumbents were doing at this point.

That is, we need to make an apples-to-apples comparison that takes into account the polarized era in which we live.

In a CNN poll conducted earlier this year, Democrat Joe Biden and Democrat Bernie Sanders both led Trump by double digits among registered voters. I cannot begin to explain how bad of a position that is.

There were six different times since the 1948 election with polls at this point in the cycle in which the eventual matchup was surveyed.

1948: President Harry Truman was tied with Republican challenger Thomas Dewey 1956: President Dwight Eisenhower held a 8-point advantage over Democrat Adlai Stevenson 1980: President Jimmy Carter held a 7-point lead over Republican Ronald Reagan 1984: President Ronald Reagan was tied with Democrat Walter Mondale 1996: President Bill Clinton was ahead of Republican Bob Dole by 9 points 2012: President Barack Obama was up 8 points over Republican Mitt Romney

Note that in none of these polls was the incumbent in anywhere near as bad shape as Trump is right now. I'll note there were other years where the eventual matchup wasn't polled at this time, but in all of those the incumbent was either tied or well ahead of the leading challenger.

Trump is in historically bad shape. He could recover and do better than expected, but at least right now it's not looking great.

Cillizza: RIGHT.

This is the broader point I keep reminding people about. Sure, Trump's approval numbers have improved marginally over the past six weeks or so. But, he is STILL in a position that, if history is any guide, suggests his party is in very bad shape in the 2018 midterms and Trump himself faces an uphill fight to win a second term in 2020.

The retort to that point is something like: "2020 is a long time from now." And it is! 924 days to be exact!

But Trump's approval numbers just seem to have a very hard floor and a very hard ceiling -- almost certainly because of how polarized we are as a country and how much the policies Trump has pursued have pushed that polarization.

So while, sure, anything can happen, the likelihood of a double-digit (or even high single-digit) improvement or decline in Trump's approval ratings seems very, very low to me.

Right? Wrong? Tie a bow on all of this for me.

Enten: I think these are all valid points. I think as I've pointed out before Trump is scoring pretty well on the economy, and he's still stuck in the high 30s or low 40s. That's quite unusual.

There just hasn't been a ton of movement in Trump's numbers. I mean he couldn't even hit 50% approval in the usual post-inauguration bump.

How in the world is his approval rating going to go higher than the low 40s? Maybe, he gets a bump if special counsel Robert Mueller says "Trump is good." Maybe he gets a bump if we start seeing an even better economy.

I think the better question that I'm asking in a piece that hopefully will see the light of day is whether you believe the normal rules of politics apply to Trump. You're going to see a lot of fancy models out there about Trump and low approval ratings...

But let's be clear, presidents who have approval ratings of 40% usually don't get reelected. That's especially the case when their favorability rating (a slightly different measure) is also in the same area.

Unless the normal rules don't apply to Trump, the only way he wins in 2020 if his approval rating is 40% is for the Democrats to nominate an equally unpopular candidate.