On the scalability issues of onboarding millions of LN mobile clients

The discussion in this email thread revolves around the potential implementation of BIP 157 in Bitcoin Core. The arguments against light clients are restated, including the burden on full nodes, potential lack of full nodes to serve them, unsustainability of relying on altruistic nodes, and the possibility of light clients having to pay for services in the future. However, it is argued that the choice isn't between using light clients or not, as people already use them. Rather, the choice is between offering a light client technology that is better or worse for privacy and scalability. The benefits of BIP 157 over existing light client technologies are summarized, including unique filters for each block which can easily be cached, the ability for clients to download from multiple sources, and the model allowing hybrid approaches where headers can be fetched from trusted/signed nodes. It is also noted that serving filters is an optional feature that must be manually activated by setting certain flags. There is disagreement about whether BIP 157 should be merged into Bitcoin Core, with some arguing that all efforts to improve the "full node-less" experience are harmful and should be avoided. Concerns are raised about potential DDoS attacks on the fullnode network if a large number of BIP157 clients arise. It is suggested that watchtowers could provide light-client services instead, though this is still seen as less than ideal.