Political advertising is both the greatest celebration of democracy and, presently, one of democracy’s most strained relationships. The courts scrutinize government regulation of political advertising more tightly than they do commercial advertising; yet the companies hosting media and social media platforms are themselves under no obligation to allow any kind of advertising at all. They can’t discriminate about it, but they are not obliged to run categories of ads.

In response to months of criticism for its role in the 2016 and 2018 elections, Facebook banned political ads earlier this year. The outright ban had a lot to do with delays in election results, a byproduct of COVID-19 and “unprecedented rates of mail-in and absentee voting.” Facebook’s theory was that the very effective, often micro-targeted political ads on its platform could cause “confusion or abuse” as ballots were counted and results were contested as groups might have had incentives to run ads not to influence the vote but to influence the political pressures around counting the vote.

The presidential election has been called and certified, but the question remains: how will Facebook treat political advertising in a post-Trump electoral world? The first opportunities we have to investigate this question are the immanent run-off races for the two Georgia Senate seats. In these contests, Facebook is not only lifting its ban on ads — it’s exclusively lifting its ban on ads, just for those races. In other words, it is still prohibiting all other political ads, whether they are very early Trump 2024 campaign ads (or those for any other aspiring candidates for any office anywhere, except the two races in Georgia).

The decision feels a little arbitrary and capricious. Yes, the Georgia runoffs are important, monumental, and historic. The stakes are high for both Republicans, who could maintain control of the Senate and be able to block most of President-elect Joe Biden’s agenda, and Democrats, who could gain that control and implement that agenda.

But if the criteria for allowing or disallowing political advertising is the level of importance of particular races, that remains a subjective calculation. A race for city council may be just as important to candidates and residents of a city as the Georgia race is for national politics. And the importance of political races is, in part, determined by the discourse that develops around the race, including through advertising.

In a blog post written about the lifting of the ban, Facebook promised to “continue to prohibit any ad that includes content debunked by third-party fact-checkers or delegitimizes the Georgia runoff elections.” That part is refreshing. One of the biggest impacts of Facebook’s earlier, anarchic philosophy of content regulation was the running of ads designed to discourage people from voting altogether. Those ads are particularly insidious because they don’t need to stand for anything, call for anything, take a normative position on anything. Instead, they simply need to convince people that neither candidate is good — and, depending on how and where these people are targeted, this could bias one candidate over the other.

I think it’s accurate to see Facebook’s Georgia position as a kind of “test run” for allowing political advertising in the future — a great source of revenue for Facebook. However, banning an entire category of advertising is quite strange coming from advocates of free speech.

There are deeper problems ahead for social media, from deep fakes to quantum-level microtargeting. But this relatively case-study illustrates that even simple advertising — and even a simple, blanket, self-imposed ban on it — can raise complex questions in a free-but-corporatized society.