The article falsely dismisses L2 like LN because supposedly it can't be decentralized. I don't think that's the case. It already is decentralized. Most nodes by far are home and office nodes, not cloud nodes. There is no reason to assume it can't continue to grow in the same decentralized way. Therefore as LN and other L2 solutions grow the cost per transaction will continue to fall. Therefore the whole premise of the analysis is built on a false or at least currently unproven assumption.