Correct, I'm baffled by your article. What's the point? I feel like I'm missing some context, which is why I mentioned not knowing anything about Asset Universes.
The fact that other people _can_ have virtual channels only seems useful if they _want_ to have virtual channels. So why would they want to advertise support for relaying payments denominated in a CSV token that they don't own (because if they did own it, they wouldn't need a virtual channel)?
The point isn't that people will advertise channels based on CSV tokens that they don't actually own. The point is that clients don't have to care whether or not the channels are real: public routing works fine whether or not the channels are real, so validation is unnecessary.
Thus, the amount of bandwidth you need to successfully route, say, a USD token implemented on a CSV scheme like RGB is much less than you'd expect, as you don't need to fully validate every channel. (Remember that validating a single CSV token txout might require megabytes of proof data)
I see. Maybe another way of thinking about that is that trampoline routing is also an option, as it is on BTC-LN. E.g., Alice and Zed don't care about the specific hops between them: they only care that the money arrives, that the total forwarding fee is less than x, and (for some users) that no third party can easily learn the identity of both the spender and the receiver.
Trampoline routing is not a comparable option as it has much worst privacy.
The remarkable thing here is that you can do fully self sovereign routing, with full privacy , over a lightning routing network without validating any channels other than your own.
The only requirement is having an anti-DoS mechanism for channel advertisements. There's lots of obvious ways to do that, such as generic proof-of-sacrifice.
