Instagram’s Like Hiding Saga is a PR Stunt:

What Facebook’s Darling Hopes You’ll Forget About Social Media Metrics

By Ben Grosser

In the spring of 2019 Instagram announced to the world that it was going to test the hiding of visible “like” counts within its interface. In the words of Instagram Head Adam Mosseri, he hoped it would make the platform feel “much less pressurized” and less “like a competition.” This announcement came at a time when the social media companies were enduring significant scrutiny for their roles in the amplification and virality of disinformation, the erosion of democracy and civic debate, and the destruction of individual self-esteem that was so widely evident that the New York Times wrote: “that Instagram can feel ‘pressurized’ and trigger status anxiety is hard to dispute.”

So, when Instagram made its announcement that spring, not only did it make big headlines, but many publications took it as a given that the decision had been made: likes were going away. The tests to come were merely a formality, many presumed—most without asking what those tests might be testing for, or what different outcomes might mean for the future of visible like counts.

In other words, the media was duped. Despite their stories from the time heralding Instagram’s tests as evidence of the company’s newfound concern for user well-being, it was always inevitably going to lead to either no actual change, or, at best, an anemic one. This is because Instagram is a corporation whose profit depends on continued growth, fueled by the extraction of user data and the production of ever-rising platform engagement. Visible metrics have been, for its entire history, a key component of this production—I would argue they are the central mechanism responsible for Instagram’s success.

So, it should come as no surprise when, after two years of testing, Instagram’s Head reports that their “research” turned up no particular effects from hiding likes. One can hardly help but recall in response other moments from corporate history, such as when the tobacco companies said smoking wasn’t addictive, or when the energy sector says fracking isn’t bad for the environment. Apparently, if we’re to believe Instagram, it turns out that likes just don’t matter much. Nothing to see here!

My own research strongly contradicts Instagram’s findings. I originated the concept of social media “demetrication” in 2012 when I launched the artwork Facebook Demetricator, a free and open-source browser extension that hides all quantifications across the Facebook interface. In 2014 I published a peer-reviewed article about my findings. In 2017 I launched a Demetricator for Twitter, and in 2018 one for Instagram. After a decade of activity investigating, working to erase, and listening to users report about the effects of hiding likes (and other visible counts), it is abundantly clear that social media metrics have profound effects on users. When like/follower/share/etc. counts are hidden, users report feeling, for example, less anxious, less competitive, and less addicted to the platforms. They talk about feeling less compulsive in response to them, less manipulated by metrics to continually like, share, and post. And perhaps most importantly, when visible interface metrics are hidden, users learn and feel for themselves just how significantly their actions had been driven—almost automated—by the presence of the number.

Caption: The author’s original video from 2012 demonstrating and describing Facebook Demetricator, a browser extension that hides visible metrics across the Facebook interface

So what’s really happening with “like” counts? And why might Instagram’s findings be different from my own? Setting aside (for the moment) their vested interest in the perpetuation of platform metrics and their vague assertions without evidence or peer review, I would point to the company’s anemic implementation of metric erasure.

First, Instagram’s like hiding options are laborious to use. To hide others’ metrics takes 6 taps through menus to find the toggle for it, which is buried in “Privacy” settings. Burying that option behind so many steps discourages experimentation and individual testing, leaving the default option (showing likes) as the one most will stick with.

Second, hiding one’s own like counts is not only repeatedly laborious, but incomplete. If I want to hide my like counts on my own posts, I have to tap 3 more times to turn it on every time I post. I can’t just change that setting once and have it affect all posts in the future. More importantly, even when I turn off like counts for a specific post I’ve made, the interface continues to report that metric back to me in several ways. For example, it accumulates the counts into a red and white metric popup every time I load the app—and periodically thereafter (far left in the image below). Instagram also continues to show these counts whenever I look at my notifications tab (far right). In other words, one can’t really hide their own like counts.

Image: Visible like counts on my own posts after enabling Instagram’s option to hide them. On left, the counts as shown in the standard notifications popup that appears every time I load the app and periodically thereafter. In the middle is the count shown when I click “others” from the feed. On the right are the like counts as shown in the notifications tab.

Third is that Instagram has chosen to show all metrics by default. Interface defaults are powerful. They set the conditions upon which any adjustment is evaluated. And most users won’t ever change the defaults anyway. Mosseri reveals his hopes here when he suggests that even those who hide likes might “want to switch back” “after a couple weeks.”

Fourth is that Instagram leaves all other non-like metrics in place. So even if a user hides others’ like counts and (partially) hides their own, they’re still faced with an interface full of metrics. Comment counts, view counts, follower counts, notification metrics, etc. All of these influence the user, and will serve as a ready substitute for metric evaluation when navigating the feed (e.g., it’s easy enough to focus on comment counts if like counts are hidden).

In other words, Instagram’s like hiding test: 1) made it hard to toggle like count hiding on and off, 2) made it impossible to truly hide one’s own like counts, 3) split like metrics into different categories controlled from different parts of the interface, 4) set the default as showing like counts, and 5) left in place all other interface metrics. If a social media company wanted to create a user interface test designed to conclude that hiding like counts doesn’t change much, this would be it. And lo and behold, the outcome from their findings will be continued platform growth—at the continued expense of human anxiety, compulsion, addiction, and diminished well-being.

Ironically, Mosseri confirms some of these effects with his recent statements. For example, he said (as quoted by the BBC):

‘“The spirit of this is to give people a choice,” using the example of going through a break-up in a relationship or switching schools.’

So, Instagram found no particular effect on user well-being, but Mosseri uses moments of extreme life stress as the example for why one might want to hide likes?

Another example Mosseri gave was:

“Maybe you want to be a little bit less worried about how many likes everyone’s getting for a couple weeks or a couple of months, and then maybe you want to switch back.”

So, if you want less worry, you turn off likes? Sounds as if like counts do in fact affect user well-being.

I appreciate Instagram’s decision to enable the hiding of others’ likes. This change will help users blunt the competitive feelings those metrics produce. But the anemic half-implementation of hiding one’s own likes reveals they don’t really want the idea to catch on. Instagram has spent more than a decade conditioning users to focus on the numbers. Any transition away from metrics was thus going to require substantial rethinking of what the platform is and how it works. Tests and experiments would need care and rigor; instead, Instagram came back with small clunky tweaks. A real test would make possible complete erasure of all visible metrics: no like, comment, view, or follower counts anywhere in the interface. This would be accompanied by a one-tap toggle so that users already dependent on the numbers could feel comfortable experimenting with hiding/showing the metrics at any time.

In a statement, Instagram said they consulted with experts during the testing period. Experts in what?, I would ask. Though I’ve worked on this topic for ten years—and released Instagram Demetricator a year before Mosseri started talking about their idea of hiding likes—Instagram never reached out to me for any discussion. Yet, tellingly, I did hear from the company during this period when their legal arm acted to force Instagram Demetricator off the Chrome web store in 2020. Unsurprisingly, this mirrors the actions of their parent company, Facebook, who did the same thing against Facebook Demetricator in 2016. Thankfully, the Electronic Frontier Foundation worked pro bono on my behalf to get the Facebook version reinstated. Given the company’s now repeated attempts to knock my Demetricators off the web, I haven’t worked too hard to reverse this latest move.

This whole saga is a public relations stunt. Instagram announced to the world in 2019 that they were testing the hiding of likes. They gained tremendous positive press from this move, with many lauding how much Instagram cares about user well-being. They then proceeded not to hide likes for everyone but instead to test the feature for two years—an eternity given their resources and capacity—only to come back later and proclaim that hiding likes doesn’t matter much? Not only does this assertion contradict my own research and the experience of countless users, but Instagram has a vested interest in this finding.

Visible metrics are key to the production of user engagement. Engagement is essential for user growth and profit generation. Their hiding tests were incomplete, leaving a user’s own like counts visible in multiple places. They didn’t reach out to some (all?) of us with a long research history on the topic. And along the way they acted to block users from fully hiding metrics via my projects, and even added new metrics to their interface with the addition of Reels. I find their conclusions and statements difficult to trust and would encourage others to be skeptical as well. Always remember: Instagram is a Mark Zuckerberg property. When Mosseri says something, it should be treated with the same level of trust that Zuckerberg has earned.

As a coda, one final comment on timing. Why now? After two years of testing and all the positive press, why come back now and say they’re done? I would point to the strongly negative reaction to the recently floated idea of Instagram for Kids. Many of the concerns expressed thus far have centered on fears around what a platform like Instagram, with all of its negative effects on user well-being, would do to children. What better antidote than to come out in response and say hey, it turns out our research shows that like counts don’t have much effect on anyone, so don’t worry about it! When companies release PR disguised as research, the media should hold them accountable for it.