The price of native Bitcoin transactions will be artificially fixed above their free market price because there is a production quota.
I think of it as a bad thing that blocksizes grow too fast if it leads to centralisation. And also a bad thing if people are blocked from getting access to full bitcoin features (eg because of the effects you mention). Being offchain generally degrades your features sometimes to no better than legacy "trust us" banking.
I think there's a middle ground which can grow the blocksize a bit for now if it has to. And some technical hope that the problem can be solved so that we have decentralisation and more scale. Sidechains have a security trade off but do give you full (and potentially more) bitcoin ethos functions - eg someone can do zerocash, or network enforced limits etc. Other things also snarks can do magical things (full node validation and low bandwidth), sharding like Rusty Russel is exploring with pettycoin (its not an alt, its an approach to sharding bitcoin). Federated pegs. And Open Transactions offers some interesting protection via its trust but verify approach though maybe not quite full bitcoin features, it can offer many features and maybe some different ones also (eg blinding though that as I understand it hampers audit, and I didnt see a way to repair that either yet).
What I want is for transaction processing to be a competitive open market. That means let the market decide how many transactions to put on the blockchain instead of smugly asserting that most of humanity "doesn't need" that level of security and therefore will be forbidden from having it.
One problem we face is it seems blockchains are inherently more expensive (bandwidth, latency, convergence time) at the limit that server clusters like Voting Pools for OpenTransactions and other related ideas. That might encourage people to not take up full bitcoin features if those systems turn out not to be able to match the features quite due to less decentralisation.
The correct response to that kind of problem, however, is not more central planning - it's more price discovery so that all costs are reflected in the price of using the network.
sounds good to me. I think there is sort of assumed to be some price discovery via user preference and miner policy but as the blocks havent filled up so far we've never seen much supply shortage other than at 0 fees.
Adam