You seem to trying to explain why it hasn't happened to journals before. The answer to which seems immediately obvious to me. So I'm probably misunderstanding you.
I'd not addressed why journals haven't already been disrupted. I did suggest how they might become disrupted, borrowing on the model of MySpace/Facebook. See also: Newspapers/radio, cinema/television, and radio/television. It's not that the old networks fail, per se, but a new network (that is, monopoly), which delivers on the same need (or want), and does so in ways which bypass the original.
But since you raise the question of why journals haven't been disrupted, part of the point is that they'd emerged as a blocking / gatekeeping mechanism only fairly recently. Peer-review is largely a post-1970 phenomenon, and really largely since 1980.
I'm not entirely sure when journal subscription costs became an issue, though I do recall hearing buzzes about it in the 1980s. The emergence of the Web as an alternate distribution mechanism was already suggesting academic journals were obsolete by the late 1990s, and I was part of early Open Access discussions at the time. I basically see that solution still developing, which suggests there simply hasn't been sufficient time.
Sci-Hub is doing a great deal on the time-sufficiency department.
I'm not sure it's convincing, but then I'm not settled on what is so generally. This is a question I've been thinking through heavily.
One model I'm working on is that monopolies are synonymous with economic rents, and derive from network structures, and/or control systems. These might be physical or logical, and would include political control systems, one of the typically sugested forms of "artificial monopolies". I'm not sure that particular distinction is meaningful. It also doesn't seem to be the one you're making.
In the case of extant structures vs. novel ones, which does seem to match your case, there's the question of why an apparrently better or more robust novel system doesn't displace an incumbent. Much of that is seen as a critical mass problem, though viewing it as a critical cohort problem might be more useful: Facebook overtook MySpace by starting and building from a high-appeal, high-value core (Harvard undergraduates). This overcame network effects by seeking individual nodes with a far greater value function both in social and advertiser appeal.
That pattern of overtaking seems repeated among other networks as well. Private jet charters vs. commercial 1st class, even Concorde-speed service, might be another.
For academic publishing, defection of fields, researchers, journals, or specific universities might be a similar mechanism.
Is this in any way along the lines of what you are thinking?