These are not atypical cases. They're the majority of the larger service providers I mentioned - in the case LN is popular and successful. I should have written "there are several categories of entities" instead -  I am sure you also did understand it that way (but my non-native English may have given you the opportunity for this "answer") And no, we don't need 90% to be "profitable" in the sense that they make money charging fees. It is enough if the money they save in comparison to on-chain fees, all costs (opening/closing channels etc.) included, is > 0.

His calculations may be correct but he then tries to "use" them to prove his point (which he has sustained during years here in this forum) that LN is not viable. And this exceeds the field of mathematics but enters the field of politics.

If you had read my post completely, you may have seen that I even agree with him that a completely distributed LN where most participants are well-connected, without hubs, is probably very difficult to achieve (and would also not be necessary, imo). But Fyookball didn't address the point that there are many entities who are incentived to run well-connected mini-hub nodes without having to profit directly from them. Their profits come not primarily from the fees they earn, but from the fees they save.

(I will not continue this discussion cluttering the thread, but leave it again for more technical FAQs.)