An interesting report by Antonio García Martínez at Wired on the Facebook ad machinery in play during the 2016 election highlighted an issue that I hadn’t considered before: that Facebook’s ad marketplace gave a huge pricing advantage to one candidate over the other. It seems like a no-brainer in retrospect that these pricing differences should at the very least be made public in the case of political contests, and arguably should be limited in the way political ads are elsewhere.
The pricing differences are not detailed in the article, exactly, but in chatter on Twitter afterwards, two parties one would ordinarily not expect to agree did just that.
Brad Parscale, who directed Trump’s digital media campaign (and was just tapped to do so again), said that compared with the Clinton campaign, he was paying pennies on the dollar for ad space. “I bet we were 100x to 200x her,” he wrote in a tweet, referring to the number of impressions they’d receive for a given sum. “We had CPMs that were pennies in some cases.” (CPM is cost per 1,000 impressions.)
Unexpectedly, Jennifer Palmieri chimed in; the former communications director for the Clinton campaign merely replied to Parscale: “Agreed.”
In other words, although we don’t have any hard data here, people with certain knowledge of this and no reason to collude agree that the Clinton campaign paid orders of magnitude more money for its ads on the social network.
Now, as the Wired article points out, and as anyone in the ad business could tell you, differences in costs are commonplace, always have been. Targeting wealthy folks in cities is generally more expensive than targeting less-wealthy folks in rural areas, for instance, which already does some work to explain the cost difference in this case.
There’s also play in the pricing depending on how much Facebook thinks your ad will boost traffic and engage users. If its algorithms think an ad will cause users to spend more time on the site, that ad will be offered at a discounted rate compared with one that doesn’t seem to push that button in people. (Kind of flies in the face of the site’s supposed pursuit of “time well spent,” but that’s a discussion for another time.)
As Facebook’s former ads VP Andrew Bosworth (AKA Boz) explained:
“Lots of confusion about ads pricing today. Remember, price for any given impression is determined by an auction so by definition this pricing cannot be discriminatory. 1/4”
So it’s perfectly reasonable in a way that the Trump campaign’s savvy positioning, targeting and provocative content resulted in Facebook giving them an excellent rate on ads. And that wouldn’t be a problem if this was Coke versus Pepsi — it would be a simple case of smart ad spending.
This is certainly a case of smart ad spending — there’s no use denying Trump’s campaign got far more mileage out of its money than Clinton’s — but there’s nothing simple about it.
Political spending is, of course, a massively complex issue and I don’t intend to dive deeply into it. And in the end there are plenty of loopholes through which any candidate or PAC could slip.
But just that fundamental fact, that one political candidate was being charged a hundred times more than the other for similar access to a platform’s users, should give anyone pause. That’s fundamentally wrong, because a Presidential election is fundamentally different from choosing between Coke and Pepsi. (I hear you cynics out there saying it isn’t, but bear with me here.)
Practically speaking, it’s a dangerous precedent to allow to continue. It happens that in this case the one losing out could spare the cash — this was a major election with hundreds of millions being spent. But that price difference could be crippling for a smaller race with less money — the difference between voters seeing one ad even a little versus seeing the privileged one all the time.
This kind of media exposure makes or breaks campaigns, which is why it’s regulated with things like the Equal Time rule, and why it’s repeatedly proposed that limits be enforced on campaign spending.
There is the counter-argument that if the Clinton rates were the “going” rate, then in fact the other “discounted” rate could in fact encourage challengers and empower those with less money to spend. But this isn’t very convincing, since Facebook seems to set the rates entirely arbitrarily depending on how much good it will do for itself, not based on some real, fixed cost (as opposed to, say, a full-page newspaper ad).
If Facebook’s intention is to encourage lesser-heard viewpoints by subsidizing their rates, the rates should reflect a judgment of that scarcity and the validity of the message it intends to amplify. Obviously that’s not the case, or if it is Facebook has failed to make that clear.
I don’t mean to propose any regulations here — I wouldn’t even know where to start. But in terms of basic transparency and civic duty, I think we can expect more from Facebook.
After all, it was Facebook’s algorithms that created this massive delta in pricing between two rival candidates. And of course it is at least partially because Facebook’s algorithms favor things that increase engagement and time on site, which often means the likes of clickbait, misleading and provocative content, divisive content that generates argument and so on.
So Facebook has two things to fix, if it is really serious about being more than a race to the bottom for perpetuating worthless “engagement” nurtured solely to multiply ad eyeballs.
First, it should work out how to voluntarily disclose and transparently explain the details of its ad sells in cases of genuine public interest. Start with elections. The bare fact that one major Presidential candidate was being charged a hundred times more than another should be considered a major failure of Facebook’s so-called commitment to objectivity, promoting multiple viewpoints and so on.
We have learned to expect nothing better: the company has proven remarkably consistent in failing to predict the abuses and shortcomings of its platform and priorities. (I suppose their livelihood depends on it.)
Second, it should reconsider how its algorithms are applied in these cases. Engagement is not a suitable metric for determining how political ads should be distributed and charged for. The idea that voters were being shown political messages based on how much attention those messages would generate should cause Facebook great embarrassment.
Basically Facebook needs to recognize that the advertising process it has developed is incompatible with ethical and transparent use in situations like elections, and it’s an insult to the democratic process (something in which the company’s founder purports to be interested) to act otherwise.
It would be much better to do this now, ahead of the regulations some would say are inevitable. Facebook is already thought poorly of by so many, in so many ways. This could be a way to get in front of that sentiment and demonstrate with actions — not more empty pledges and assurances that “we take this very seriously” — that the company has the best interests of its users at heart.