The miners are the ones securing the network and I think most of them can easily afford to buy more storage/bandwith. In this regard, less people running a full node isn't really a threat...
It would be still ideal for a significant portion of the users to have the capabilities to run a full node. Relying on someone else for validation isn't really a good idea, and yes, I did agree that the full node numbers would probably dwindle in the future.
new block size limit = previous block size limit x current avg fee / previous avg fee
We know that the avg fee depends on the amount of transactions that are waiting to be confirmed AND the block size limit, the demand for block space shouldn't change much in the short term.
Do you think we should cap the block size with your variable block size scheme? You do realize that an increase in block size or transactions per block will inevitably result in the blocks being propagated through the network slowly or a far slower validation. This makes it unsuitable for smaller miners to be mining and will have to be directly connected to at least half of the network to achieve a lower stale rate. There is a reason why the blocks are capped at 1MB, and that any scheme should strive for a reasonable block size that doesn't compromise the integrity of Bitcoin. An increase in the stale rates also results in a decrease in the perceived security, as there can be multiple conflicting blocks at the same height.
I don't disagree that we need a block size increase. What I'm thinking of is a sustainable block size increase that balances the tradeoffs to the benefits.