There are several entities which would benefit from a LN node even if their income from LN fees is tiny. The most obvious examples are exchanges and other custodial wallet providers. But there are others, like arbitrageurs (we discussed that some pages ago), but also online merchants (from a certain size upwards).

This would lead, if we take the graphics above, to a scenario closer to the middle one from Fyookball's blog post, but I would argue that these "indirectly profitting nodes" would try to be more well connected than in this example where each one has only 2 connections to nearby nodes. So we would have, at the end, a network between the "decentralized" and the "distributed" model, with clearly identifiable hubs, but which are much better connected than in the example.

This however depends again from popularity of LN itself (only a few of these entities do not depend from incentives which are related to the size/network effect of LN), which brings us back to the chicken-and-egg problem we discussed in the pages before. Clearly, exchanges could be the "boosters" of LN if they bring the arbitrage business "on board", and this already could mean a significant growth. But for now afaik Bitfinex is the only larger exchange which supports LN fully, so arbitrage via LN still doesn't make sense.

Fyookball isn't an "independent researcher", they are clearly a big blocker and probably a Bitcoin Cash supporter. That's ok for me but their postings (on their blog and here) aren't neutral in any sense.