On Thursday morning, President Trump retweeted a Twitter poll from @ProgressPolls, an anonymous account that you’re probably hearing of for the first time. The poll asked: “Who is a better President of the United States?” And the results said that Donald Trump was better that Barack Obama, by a margin of 61 percent to 39 percent.

Twitter polls, as you should know, are not scientific — just like any other online poll that opens voting to anyone with access to the Internet. They’re informal surveys of whoever sees, and votes on, the question contained in the tweet for several hours after it’s posted.

In other words, they’re not at all representative of the American population. But they are a good way to test the strength of your particular ideological faction on Twitter at any given moment.

Online polls, answerable by anyone, are entertainment at best. On Twitter, pro-and anti-Trump enclaves regularly flag and flood informal opinion polls run by their opponents to shift the results away from the pollster’s preferred outcome. And that’s basically the extent of their value: as entertainment, and a relatively harmless outlet for political trolling.

Given Trump’s seeming endorsement of its work, we got curious about where this particular Twitter poll account came from. It led us pretty quickly into a #MAGA Twitter rabbit hole.

@ProgressPolls, as a Twitter account, has a pretty neutral appearance. “ProgressPolls asks the tough questions that people need to be asked. Our polls focus on current events and politics,” the bio reads. Simple enough.

In a series of Twitter DMs (direct messages), one of the people behind ProgressPolls said that the account was actually run by four people “scattered throughout the United States.” The respondent said they’d been helping to run the account for about two weeks, or for the duration of the time that the account has been tweeting out polls like the one that caught Trump’s eye.

The person I spoke to found out that Trump had retweeted them with a text message this morning that simply read, “Check Twitter.” They added, “We’re glad the President acknowledged our tweet, we’ve seen a lot of responses and growth since then.” The account has gained a few thousand followers today, they guessed.

But despite its short Twitter history of running political polls on current events, the account already has 125,000 followers, which is quite a lot for a new account that does nothing but run Twitter polls, even with a bump from the president. That’s because @ProgressPolls is not a new account. It has changed hands — and, basically, entire Web identities — multiple times since it was started in the spring of 2016.

Take a look at what the account was tweeting during the presidential campaigns. It’s very different:

Judging by its Twitter archives, this account was a prolific, and popular, member of the pro-Trump Twitter community for much of the campaigns. The person answering my DMs confirmed that the account had gone through “several different management persons, so the account has changed as such.”

Before @ProgressPolls, its username was @Truth_Bombers, they confirmed, something I suspected after a bunch of digging into the account’s previous replies, and from a deep dive into the Internet Archive’s Wayback Machine. It switched to @ProgressPolls just weeks ago.

Here’s a sampling of some of the now-deleted tweets from Truth_Bombers, via the Internet Archive:

And before that, the account’s Twitter replies and Wayback Machine strongly suggest that the account had the handle @Always_Trump, a pretty popular Twitter account on the #MAGA Internet, with tens of thousands of followers this time last year. The current management of @ProgressPolls couldn’t confirm either way whether the account was once @Always_Trump, because, they said, it was before their time.

Shortly after we spoke, @ProgressPolls began deleting some of the account’s remaining pro-Trump Twitter history, from the time that it was possibly tweeting as @Always_Trump. In a DM, they said they hoped to clear it all out soon, and that the “current team is very diverse and young, we’re aiming to reach a wide audience here on Twitter to give them the chance to express their opinions freely and anonymously.”

In its current iteration, the questions asked by @ProgressPolls are of a drastically different tone to its #MAGA Internet past, but reflect the account’s conservative, pro-Trump base of followers — although the person I spoke to said they were personally a “registered Democrat.”

Not all of the responses are favorable to the Trump administration.

… Although there appears to be some evidence that a portion of their Twitter polls have been trolled by Trump opponents.

Other results are more closely aligned:

Over the course of our DM conversation, I returned multiple times to what value @ProgressPolls saw in conducting online polls like these, when they were clearly not representative of public opinion, yet often manipulated to sound as if they are.

“Any online poll or poll in general for that matter can be skewed and misinterpreted. We’re not aiming to be Rasmussen polls, we’re aiming to be an outlet for the average person to click their vote,” they wrote. They pushed back against the idea that their polls only reached small groups of like-minded Twitter users, because they used trending hashtags to prompt the question for their polls. So, the idea is, if something’s trending, they want a poll, carrying that hashtag, in the mix. But, they acknowledged, “it’s Twitter at the end of the day.”

In any case, the reason Trump may have retweeted a mysterious Twitter poll of little value as a representative survey should be obvious. Although it’s not asking the exact same question, legitimate polling on Trump’s approval rating gives him much less welcoming news: Two separate polls that came out Monday pegged his overall approval at 38 and 32 percent.

More reading: