I reached the same conclusion after looking into the matter for a while. I've mentioned it several times on HN and elsewhere and have yet to find anybody offering a rebuttal. Worse, there are very little simulations being done on how it's going to work in practice (admittedly that was about 6 months ago, but the decision to implement segwit/lightning had already been taken at that point anyway).
We're talking about a $100+ billion market cap cryptocurrency and nobody knows how it's going to work. It's hype, speculation and memes all the way down. The power of greed.
The routing of transactions just won't work in a fully peer-to-peer network. Inevitably large "Tier 1" hubs, interconnected together, will surely be needed.
Djikstra or Bellman-Ford type algorithms won't scale to where they need to for anything else, likewise users won't maintain enough funds in channels to make potential routes usable.