On the scalability issues of onboarding millions of LN mobile clients

The concern with consensus capture by any subset of users whose interests diverge from the overall consensus is equally damaging as consensus capture by miners. If more light clients outpace full nodes, the costs of security are being externalized from the light clients onto the full nodes, making full nodes harder to run. This could price out some marginal set of full node operators, causing a net new increase in light clients and a redistribution of load onto a smaller surface area, which is a naturally unstable process. As node counts drop, the set of node operators will increasingly represent economic actors with extreme weight, whose interests may diverge from the population at large and can be coerced into behavior they otherwise wouldn't. It is easier to find agents who carry lots of economic weight. We should be just as wary of consensus capture by exchanges or HNWI's as we are about miners.Antoine Riard argued that the attack scenario has a lot of assumptions on topology and deployment and for such an attack to succeed, miners' nodes need to be connected to clients to feed directly the invalid headers. He suggested implementing fork anomalies detection and having fallback connections to a trusted server authoritative on the chain view. However, he agreed that a wide-enough, sane backbone network is needed to build on top, and efforts to improve the "full node-less" experience are harmful and should be actively avoided.Luke Dashjr opposed merging support for BIP 157 in Core as all efforts to improve the "full node-less" experience are harmful, and without serious improvements to the full node ratio, Bitcoin is likely to fail. For Bitcoin to remain secure, a supermajority of the economy must verify their incoming transactions using their own full node. If Bitcoin can't do it, then Bitcoin can't solve any problem.