Illustration: Jim Cooke

We are, right now, in the midst of a digital media upheaval. What was previously conventional wisdom—that a media company with hopes of turning a profit needs, above all, to achieve scale—is being proven false. The new conventional wisdom is that video will be digital media’s savior, but it is only a matter of time before this is proven false too.

The Digital Content NewFronts are being currently being held in New York, and your favorite media and media-adjacent companies are pitching their upcoming projects to advertisers, practically begging for sponsorship or mutually beneficial branded content opportunities or something or other. It’s all about video: “To qualify as a NewFronts presenting company, presenters must create original content in video format that is available online.”

BuzzFeed—the digital media company which has invested the most in video—presented last Monday, and of course talked about their wildly successful rubber band watermelon explosion, watched by 807,000 people at its peak on Facebook Live. CEO Jonah Peretti said the video’s popularity brought BuzzFeed into the realm of television:

For context, last year CNN averaged 712,000 viewers in primetime. Brands, according to the argument BuzzFeed is implicitly making, should therefore pay something like the same to message and tell stories against BuzzFeed’s next smash hit as they currently pay for commercials during the evening on CNN. Right? Well, no.

The number in the corner of the Facebook Live window that you saw going up while BuzzFeeders strapped rubber bands onto a watermelon—or while the Try Guys spent 83 minutes failing to pop a giant balloon last week—represented the concurrent number of people watching the video (or, more specifically, who had the video playing in a browser window) at any given time. At its peak 807,000 people watched that damn watermelon explode, which is actually orders of magnitude fewer than the average of 712,000 people watching primetime CNN last year.

Television ratings are measured by the privately-held Nielsen Corporation, which pays a small amount of money to around 25,000 of the 116.4 million “TV homes” in America to install People Meters. Using special remote controls, these People Meters measure what is being watched, by whom, and for how long. Nielsen then uses statistical techniques to extrapolate this sample and calculate the ratings—how many people are watching—of each TV show.

The TV ratings Nielsen reports aren’t concurrent viewers, but rather “average minute audience,” which is exactly what it sounds like. It measures the average audience watching across each minute of the show.

If BuzzFeed’s watermelon video had been measured the way a TV show is, its viewership would’ve been closer to zero than the 807,000 it trumpeted to advertisers. Viewership started off low and took 45 minutes to build to that 807,000, and few people watched the entire video; many tuned in for five or 10 minute blocks at the end. Facebook’s metrics also wildly inflate the number of people watching a given video, as they count somebody as a viewer once they have been watching for just three seconds, and by default Facebook videos autoplay as you scroll to them in your feed.

The conflation of digital and traditional viewership metrics has gotten under the skin of TV people, and for good reason. If advertisers can be hoodwinked into believing that a sizable number of people are actually watching things on Facebook Live, they will direct their money online, where the ad rates are much, much lower than they are on TV. The thing here is that the TV people are right—even serious online video hits deliver numbers that would barely register if measured the same way TV programming is.

Here, for instance, is Nielsen explaining how digital numbers lie:

In our second example, the 2014 World Cup on ESPN had an average-minute TV audience of 4.6 million persons, and received 115.5 million digital views. But 4.6 million for TV and 115.5 million for digital is the wrong comparison—if we translate digital viewership into a TV metric, the average-minute digital audience of the World Cup on ESPN was 307,000, representing just 7% lift of the TV audience.

Since it was broadcast live, the watermelon explosion has been watched by 10.7 million people, per Facebook’s count. If those people were as engaged as online World Cup viewers—and I’d venture that, on average, they were less engaged than people watching the most popular sporting event on the planet—those 10.7 million digital views would translate into an average-minute TV audience of 28,563 persons. If Peretti brought that number to advertisers at the NewsFronts as evidence of wild success, they would’ve laughed in his face.

FX’s research chief got testy making just this point:

Specifically, Piepenkotter spoke about last week’s Facebook Live video that featured two Buzzfeed employees wrapping rubber bands around a watermelon until it exploded. Buzzfeed bragged afterward that the video had 880,000 viewers at its highest point.

“That’s like the two-second view that we see in digital currency right now, which is creating an extraordinarily false narrative and a meaningless narrative,” Piepenkotter said. Applying the same metric to her network’s offering, “The episodes of ‘O.J. Simpson’ to date would have been 143.9 million hours viewed and 259 billion views.”

Leaving aside digital video’s low viewership when measured like TV, advertisers don’t even look at TV ratings the way everybody else does. They buy ads based mainly upon C7 ratings, which measure how many viewers watched commercials as they aired and up to seven days after. And more precisely, they care about the number of a specific type of people watching the commercials—the key demographic. Most generally, the key demographic is people 18-49 (or 25-54), as they’re the ones with disposable income to spend on whatever advertisers are selling.

That’s broad; things get more finely-tuned than this. Nielsen collects demographic data on the people in its approximately 25,000 homes with a People Meter, and can tell advertisers how many people who fit various criteria related to age, gender, race and so on are watching. This data is necessary for advertisers: A cosmetics company wants to know specifically how many women will be watching, for instance.

Facebook can’t tell you this; more precisely, they can, and someday will, but currently won’t. Were those 807,000 people concurrently watching BuzzFeed’s watermelon explosion young? Old? Men? Women? Facebook won’t say.

None of this should be read as supportive of Nielsen, a private, for-profit company whose ratings are absolute garbage. Its statistical methodology is opaque; its sample of homes isn’t anywhere close to truly random; its ratings rely to some degree on self-inputted data; its ratings don’t take into account communal television viewing at places like bars and airports; and Nielsen struggles to capture the entire universe of “television” so far as it takes in DVRs, streaming services, mobile devices, and so on.

One day—maybe soon—video viewership will be measured much more accurately than it is now. Facebook doesn’t have to rely on a sample; it can measure every single person watching a video. And, in an advertiser’s wet dream and a privacy activist’s worst nightmare, it can provide much better data about viewers than Nielsen. Rather than advertising based upon how many women are expected to watch a video, advertisers will be able to advertise based upon how many women aged 18-29 who live in the Dallas area, are single, and say Shawshank Redemption is their favorite movie, are expected to watch a video.

It is worth considering, though, whether the availability of more precise metrics is beneficial to media companies. In some ways, the huge disparity between ad rates for print and digital is because online we can better measure exactly how ineffective advertising actually is, while in the past it was enough to know a Vogue reader might’ve looked at an ad. For years TV has relied upon suspect data—20 years ago Nielsen ratings were even less accurate—and it seems that the more precise the metrics are, the less everything is “worth.”

None of this should be read as a criticism of BuzzFeed, either. However poorly their viewership compares to routine CNN programming, getting 807,000 people to watch something on Facebook Live at the same time is legitimately impressive. If Gawker’s Facebook Live interview with a small child about politics had garnered 800,000 concurrents instead of a few thousand, you can bet Gawker Media would be trumpeting it to advertisers.

But if digital media companies want to command anywhere near the same lucrative ad rates as TV networks, they’ll have to prove their videos consistently get the same audience as television networks. Which they can’t, because they don’t.

For all that, Facebook Live and its ilk are (as of now) bullshit, which is why the only thing anybody will talk about is the best case scenario for online video. The other week, Facebook decided to put a new metric front and center. It’s called Peak Concurrent Viewers, and it might be the greatest vanity metric in the history of vanity metrics. It doesn’t measure anything useful about the viewership insofar as being able to sell it to advertisers, and only works if it can trick them into believing that a significantly greater number of people are watching the video than actually are.

Video is the future of online media. People are spending an increasing amount of times on their phones, and phones are becoming terrifyingly powerful, able to flawlessly run the largest video—though data plans and cell networks remain a bottleneck. Some online media companies can ignore it, just as some print media companies have remained successful mostly ignoring the internet. The vast majority cannot. As the newspaper industry painfully learned a decade ago, evolution is necessary.

But just as many newspapers have died and even the most powerful are still struggling to evolve—hey, look, in the past three months the New York Times added digital subscribers but still lost a bunch of money—the emphasis on video is going to produce way more losers than winners. The first losers are digital journalists paid to write, not make video (hi!), but soon many digital media companies that have bet heavily on video will be losers too.

While the media is more broadly retrenching, and digital media growing only slowly, demand for video journalists is outstripping supply. Everybody this side of Harper’s is building out a video team, without any idea how they will pay for it in the long term. In their pivot to video, for instance, the VC-funded Mashable went from eking out modest profit to burning through cash.

Just as publishers rushed to achieve scale under the naive assumption that scale would somehow free them from fundamental economic precepts, today they are rushing into video. It has the makings of a panic, with companies more focused on not being left out than an actual strategy for making money. This is a gold rush, with little evidence that there is anything more to rush for than a few flakes. To put it more pithily, video is the future, but just because you make video doesn’t mean you will survive.

Facebook is nobody’s friend, and especially not journalists’. They want to make live video look as attractive as possible to suck news organizations into providing high-quality content that users will engage with more than Grandpa Edwin’s latest rant about Benghazi. They are giving live video priority in your feed and push notifications about videos are sent to followers, all of which serve to get more eyeballs onto videos.

Facebook can pull the rug out from publishers at any moment. One of the reasons your favorite media companies are suddenly embracing Facebook Live is because Facebook is paying them to do so (Disclosure: Gawker Media is now part of this program), but this obviously won’t last. They can unilaterally change the rules of the game—say, charging to send a push notification to followers—and we already know that Facebook does not mind if their changes devastate publishers.

I sure as hell don’t know how to fund modern journalism on a broad scale, but I have been around just long enough to understand not only that live video isn’t a silver bullet, but that there is no silver bullet. A few publishers will figure out how live video can improve their reporting and make them money, but most won’t. It is worth noting that the most successful Facebook Live videos so far have essentially been gimmicks; the type of things more palatably referred to as “content” than “journalism.”

And that, right there, is likely to be the future of digital media, and increasingly its present too. Multimedia production shops that very much resemble TV studios—and make no pretense at having a wall between reporting and advertising—thriving, while more traditional journalism shops attempt to eke out a hardscrabble existence.