After each of the three major presidential announcements we've seen over the last month, the public relations people at Facebook and Twitter have hustled out data on the size of the splash that resulted from each plunge. And the media, chastened after 2012 into showing respect for numbers, have dutifully tweeted and Facebook-ed those numbers right back out.

Welcome to the era of the social straw poll.

Hillary Clinton's launch video, we know now, was seen 1.8 million times in 12 hours. Her tweet was seen three million times, in whatever sense counts as having a tweet seen. Across the aisle, Ted Cruz "destroyed" Rand Paul on social media, in the evocative wording of the National Journal. Cruz had 5.7 million "interactions" to Paul's 1.9 million.

So can Paul recover? Can anyone stop the juggernaut that is Hillary Clinton?

We can start to answer that by considering another headline: "#Hillary #TedCruz rule." That's Politico, last December, noting that Clinton and Cruz were "already dominating" on social media. "Of the 27 million Facebook posts, comments and content likes related to the potential White House candidates between Aug. 22 and Nov. 22," Hadas Gold reported, "Clinton and Cruz each were mentioned in 20 percent of the posts, according to Facebook’s data scientists." So in other words, the biggest splashes we've seen for 2016 so far ... have come from people that were already popular on social networks.

What's more, these bits of data lack both clarity and context. What on Earth is an "interaction"? Is it positive? Negative? A comment? A like? Both? Are those equivalent? And as social networks continue to grow and flex and evolve, is there any way of drawing a meaningful comparison to other interests -- much less to other candidates?

All of which skips over one of the most important points: Voters don't vote until early next year. There have been no real campaign ads, save Ted Cruz's Easter weekend made-for-media buy. There have been no debates, no campaign. There have been some studies showing a correlation between tweets before an election and election results, but those are both very much disputed and -- even if true -- only dealt with the last few days of the campaign. There's been no study of how early support on social media affects a presidential race, in part because there have only been two presidential races in the era of social media.

Not to mention, the size of the numbers -- millions, when political watchers are used to tens of thousands -- seem nearly incomprehensible.

It's not impossible, of course, that these surges could predict something important about what's going to happen in 2016. To some extent, I mean that in the sense that it's not impossible that a psychic could guess your grandfather's name. But it's also possible that the Cruz burst involved a lot of new people, who've now committed to supporting him, come Hell or high water.

There's no reason to think that this is the case. The political press learned in 2012, though, that there was value in considering metrics besides the perceived acid levels in their guts. In the great war between Nate Silver and the pundit class, Silver won handily. Overlay that with the obsession that anything to do with technology will necessarily shift the end result of the election, and these Facebook/Twitter/YouTube/everyone else numbers are irresistible catnip.

They're like the early numbers on election night. With 1 percent of the vote in, Candidate X leads by 5 points. A useless tidbit, but in the absence of any other concrete bit of data, it's something to chew on. Same here.

Except that on election night, those are actually votes. This is indeed more like a straw poll; it can be gamed (to some extent), and it details nothing about how actual voters actually plan to cast their actual votes. Nineteen months before the presidential election, we might just as well be talking about the number of times each candidate is blinking.