Facebook (FB) is in regulators’ crosshairs because of revelations that Russian operatives used the platform to sow discord among the American electorate and discredit Democratic candidate Hillary Clinton during last year’s presidential election.

Part of the Russian effort involved impostor accounts that seemed to represent ordinary Americans or American organizations but were in fact operated by foreign agents.

A Russian-run Facebook account called Blacktivist, for instance, attracted 500,000 followers with provocative posts highlighting racial injustice before Facebook discovered the subterfuge and shut it down. Facebook says about $100,000 in ad purchases by Russian entities since 2015 were linked with about 500 fake accounts.

Special prosecutor Robert Mueller is investigating Facebook’s and Twitter’s role in Russia’s 2016 disinformation campaign, with execs from both companies due to testify before Congress soon.

New laws governing what appears on social media sites seem likely, but first, policymakers must answer some confounding questions, including: How prevalent are fake accounts on Facebook in the first place?

Nobody really knows, because Facebook doesn’t release enough data for researchers to study the problem. In its latest securities filing, the social-media giant says that perhaps 1.5% of its 2.01 billion accounts worldwide are “undesirable.” That would be 30 million accounts.

But the company also says that estimate could be off because it’s based on a “limited sample.” Other estimates from a few years back put the number two or three times higher.