The (maybe) presidential campaign of D. J. Trump issued a highly Trumpian news release in response to news that his announcement  Tuesday made a big splash on Facebook.

"Trump Smashes Social Records Following Presidential Announcement," it yelled. "3.4 million Facebook users in the US generated 6.4 million interactions regarding the launch of his campaign, the highest by far, among all 2016 GOP candidates."

Why? Because "Mr. Trump has a tremendous audience across the country," as the news release notes -- with probably more accuracy than it intended.

In general, how something does on social media is a bad way to gauge success. In politics? It's especially dumb, as we've noted before. We're comparing apples to cameras to a 1973 Deep Purple LP when we pit Donald Trump launching  Tuesday afternoon against Martin O'Malley launching on the weekend against Hillary Clinton launching a campaign that starts off 14 miles in front of the competition. And then we're using that comparison to try and predict tomorrow's lottery numbers. This is the first contested presidential campaign in which Facebook is so dominant. We have no baseline of comparison in politics.

What we do know is that online fame seems to help. The extent to which campaign launches generate interest from individual Facebook users and get "interactions" on the social platform is loosely correlated to how active the candidates already were on Facebook -- or at least how much interest they got. We compared user interest and interactions within 24 hours of the launch to how many followers each had on Facebook the day of the announcement, and voila. (Clinton is in italics because data from the day of her launch was not available. Instead, we used today's figure.)

As we noted before, the two people getting the most attention on Facebook last fall were Clinton and Ted Cruz -- numbers 1 and 3 around Trump on the users/interactions scale. (Notice, too, that Trump "smashed social records" -- in the sense that he did worse than Clinton.)

Clinton, Trump and Cruz are outliers. Or maybe Rand Paul and Mike Huckabee are; they didn't do as well as you might have expected given their high follower counts. It's hard to say, because we don't really know what this means. It is not the case that Trump did a lot of organizing around Facebook; it is pretty clear, too, that not all of those interactions were of the "hey, I'm voting for this guy" type. It's just ... a number. It's bigger than the number for the Nikon, but smaller than the number for "Who Do We Think We Are," (to continue the joke from above).

Here's one real thing we can look at: How those interactions compared to candidates' standing in the polls. We compare the Real Clear Politics average from seven days after the launch to the number of interactions in those first 24 hours. There's essentially no link at all.

(Notice, too, that we had to take out Hillary Clinton and the Democrats because they're on a different scale of support. Another way in which comparing Clinton to Trump is tricky!)

If you're wondering how these figures correlate to rate of change in poll numbers, we actually get some correlation -- stronger correlation than between followers at launch and the number of interactions. Cruz saw the biggest jump after 14 days and the most interactions within 24 hours. Rubio saw the second-biggest jump ... but fewer interactions than Rand Paul, who didn't see as big an increase in the polls.

But Paul was doing better anyway. And Cruz announced first. And all sorts of other things that come into play in polling. This is a classic correlation/causation conundrum.

We also have a smaller data set. We'll see what happens with Trump. Maybe he'll prove the validity of the last graph, jumping up in the polls. Perhaps he'll end up more like the graph above it, staying mired in the lower tiers of the Republican candidates.

The ferocity with which Trump pushed out those Facebook numbers, though, reminds us that we have these numbers because of marketing: Marketing from Facebook; marketing from Trump. And for that reason, if no other, it's very much worth defaulting to skepticism.

In short: We don't know what Facebook numbers mean, if anything. So we should stop pretending we do.