Today, Meta (in the US) stops new political ads being created ahead of next Tuesday’s midterm elections. Ads will still run on Facebook and Instagram – lots of them – but they’ll have been created in the past, and won’t respond to new events that might happen before the election.
We think ad blackouts are generally a good thing for reasons we’ll make clear below, though we also think Meta’s ‘one week’ version is too long and insufficiently restrictive in that it isn’t a true blackout (campaigns can still run the ads they want, provided they are within Meta’s advertising rules, they just need to upload them early).
Meta’s blackout policy, which first ran in 2020, to us, stems from a combination of self-interest in managing internal ad approval processes and not being caught in further PR firestorms about electoral disinformation in advertising. It also results from the fact that the US can’t pass legislation of this sort, because the 1st Amendment would find it to be an unconstitutional governmental restraint on speech.
The outcome is that a potentially important piece of US campaign ‘law’ results from internal negotiations between a tech company’s policy and communications teams. As a result, many campaigns have complained about the legitimacy and nature of Facebook’s policy as well as inconsistent implementation of it.
The most important democratic goal of a pre-election advertising blackout should be to ensure people are able to hear the full range of candidates’ arguments in the days leading up to an election (when voters are most likely to be paying attention), while protecting them from information that avoids scrutiny and accountability (e.g. surprising stories, late-breaking scandals or false information about the voting process), when other accountability systems are unable to respond due to a lack of time.
Practically, that means blackouts should kick in around 48hrs before voting starts. This allows neutral sources of information (yes, you are right to have questions about whether these exist, but it’s a separate problem) to evaluate what the campaigns have been saying in those final few days and report on what they’ve seen – whether good or bad – while maximising the time available for new information and arguments to be made available to voters.
It’s also important that a blackout is a blackout. In the coming week, Meta’s policy will allow campaigns to run ads they’ve already uploaded. It’s a greyout at best.
In some countries, where they are part of electoral law, blackout periods are ’quiet periods’. In the final 48 hours or so of a campaign, the arguments are all made, and advertising should, at the very most, focus on candidates ’getting out the vote’ with simple messages to encourage those who will support them to go to the ballot box.
Some of the risks attributed to digital political advertising over the last few years have been overblown. Others remain theoretical risks that don’t seem to present themselves in practical reality. But in the hours before and during voting, we should take additional care to manage any risks, because the injustice felt about the ultimate result will be many times greater if its cause occurs in this final window.
Given the constraints of the US constitutional position, and the global nature of these problems, the solution probably isn’t a new law. But there is a quasi-legal alternative. The EU Code of Practice on Disinformation includes a clause that instigates future discussions about the use of election blackout periods to prevent the risks caused by disinformation in paid advertising. These are still to come, but when they do, the largest advertising services (including, ideally, the streaming services, whose day in the sun when it comes to political advertising policy is yet to come) will meet with civil society organisations with an interest in free expression and electoral integrity, and agree a common best practice position on pre-election advertising blackouts.
When they do, they’ll produce an outcome that’s better and more legitimate than the one that kicks in today.