That's exactly what happened during the 2000 crash -- much of the "revenue" being booked by internet companies was imaginary (in the form of contract swaps), and much of that was driven by intra-industry advertising. Everyone was paying everyone else to advertise, and it all fell apart more quickly than you'd expect given top-line revenue numbers.

I suspect something similar will happen this time around -- there will be a slowdown in the real economy for some unpredictable reason, and the imaginary, startup-servicing-startups economy will implode.

This is an important observation. There are absolutely angel and venture-backed startups with "real" revenue (fewer with actual earnings), but if you look closely, it appears that a non-negligible percentage of them derive a not insignificant portion of their revenues from sales to other angel and venture-backed startups.

More broadly, an even larger number of startups are also indirectly dependent on the current landscape, as they operate in regional markets where consumption is being driven in large part by the influx of capital into tech.

There's nothing wrong with building a business that takes advantage of these types of situations, but my sense is that a lot of the entrepreneurs running these types of companies aren't consciously aware that they're the beneficiaries of an environment that will eventually change. As such, they're not at all focused on trying to diversify their revenue base and to be frank, it many cases there really isn't an opportunity for them to do so anyway.

To YC's credit, it's incredibly hard to sell anything to the YC ecosystem from outside of it, because they already have chosen options from inside of it. Yet another reason to give radical differentiation a shot if you can stomach it.