On the web, by contrast, the vast majority of ads are not native. Instead, they’re intrusive, annoying, unpleasant, and — in most cases — completely ignored. We’ve now been consuming content on the internet for 15 years; we all know how to do it, and we know what we like, and publishers, including BuzzFeed, have become very good at delivering exactly what we want.

In stark contrast to the increasing sophistication of web publishing, however, the overwhelming majority of web advertising is still based on standard IAB ad units which were introduced in 1996 and haven’t changed much since. We’ve all learned how to tune such things out, either mentally or technologically, with ad-blocker software. Banner ads are never engrossing, they’re never shareable, and insofar as they attract your attention they do so in an evil way, by animating or blinking or otherwise distracting you from whatever it is you are trying to read.

When someone reads a BuzzFeed ad from Virgin Mobile or Geico or GE, they might “only” have a 20% or 30% chance of sharing it. But that’s not really the point. The point is that they read it, and they liked reading it. The “social uplift” is an indication that the ad is connecting with consumers — it’s like clickthrough rates, but real. Native advertising (as well as content marketing, insofar as there’s a distinction) is a way of communicating with web readers in a language they’re receptive to. And it turns out that when you do that, they actually listen.

In terms of disruptive force, then, native has a huge advantage over banners in that it is much more effective in connecting with consumers. And there’s another way that it’s disruptive, too: it utterly upends the standard ad-agency business model. This is the real reason that ad agencies are less than enthusiastic about native — they can’t make money at it. Banner ads are a lovely income stream for agencies, and ad-sales networks, and the whole crazy ecosystem of display-advertising companies. Every time there’s an impression, lots of intermediaries are sure to take their cut.

Native, by contrast, works on a very different model: you spend a certain amount of money putting it together, and then it lives online forever, generating marginal views at zero marginal cost. The agencies can still charge for their creative work, but they can’t charge for media buying any more — which is where the real money is.

As a result, most native campaigns tend to be worked out between publishers and brands directly, with ad agencies helping out but not driving the decision-making. It’s the beginning of the disintermediation of the agencies, and so it’s hardly surprising that they’re unenthusiastic about the trend. This is real digital disruption: native shops like BuzzFeed or Barbarian Group will never be as profitable as the huge ad agencies, but they can still cause those agencies to suffer very large drops in their digital revenues.

The big unanswered question, then, is not whether native has disruptive potential — it clearly does. Rather, it’s whether native will ever be able to truly scale. Native is growth-constrained on two fronts, and that means that if you’re betting on industry-changing disruption, you’re making a risky bet. The first constraint is creative. Native is hard work. Rice talks about how Virgin Mobile has to come up with “several posts a week” when its running a BuzzFeed campaign, and his article is illustrated with a photo of a “creative strategy meeting” where I count 19 people in frame, plus untold others out of it. The amount of human time and effort that goes into a native campaign is enormous, continuous, and it doesn’t decrease much once the campaign is up and running. You can’t just run the same banner a billion times: the marginal daily cost of native campaigns is vastly greater than the marginal daily cost of buying banners.

Felix Salmon is an Audit contributor. He's also the finance blogger for Reuters; this post can also be found at Reuters.com.