Yeah, only up to a certain point because if people use custodians and lightning, miners receive very little revenue from the occasional channel opening and closing. 

That's why the miners supported big blocks in the block size wars. 

Decentralisation is maximised by minimising the total cost of using Bitcoin in a permissionless way. 

Total cost = capital cost of a node + transaction cost x number of on-chain transactions. 

Currently you can run a node with a $100 second hand computer. At a fee cost of $1 per transaction you can do one transaction a month for 4 years before the cost of the transactions exceeds the cost of the node. 

If the block size was increased to reduce transactions to 10 cents but the node now costs $1000 it would cost more overall. 

But if fees go to $10 per transaction, you can only do about 2 transactions per year before the cost of transactions exceeds the cost of the old computer you are running the node with. This could be managed with just opening one channel to a fedmint and splicing once a year, but that comes with some tradeoffs. 

Where to go once fees continue to rise, say to a theoretical $100 per transaction? 

 You answered your own question.  Lighting, custodians, fedminits, stuff not yet invented is what will happen when it’s $100 equivalent per transaction. 

Normal people already hold checking, savings, credit cards, brokerages, HSA, Venmo, etc accounts. In the future they could mix a custodian lighting wallet for checking, a custodian L1 / Fedimint / whatever for short term savings, and cold storage self custody for retirement. 1 solution should not meet all of these needs as they are different. 

Necessity breeds invention. Bitcoin already transacts as much as Fedwire and its open nature means it should be easier to verify your funds with a custodian than the current black box of fractional reserve banking. Not saying it shouldn’t ever change, but the barriers being so high is a good thing. It means there needs to be a real problem to solve instead of hypotheticals in the future. 

 I'm not trying to argue, just verifying the trade-offs wth a back of the envelope calculation. 

The goal is to maximise participation in running nodes, but there is not much point running a node if you can't afford a transaction. 

The block size means there are a maximum of 4000 transactions every 10 min. That's 210,240,000 transactions a year with full blocks. 

Even if you saved up to add to or withdraw from your retirement savings once every 10 years it only allows 2 billion people to use Bitcoin onchain. What are the other 5 billion meant to do? 

I'm not arguing for larger blocks. It's an inefficient scaling solution and one that is not beneficial for a long time. We should develop the scaling solutions cautiously first, because these could become more different to implement as time goes on.