I don't think we really have different design goals in mind here. I agree that there's no point to Bitcoin without censorship-resistance.
Some points:
Blocksize vs. miner rewards. I don't see the link here as strongly as you do. It seems to me it's demand that will drive miner fees, not supply, whatever the blocksize. With an infinite maximum blocksize, anyone mining for profit would simply set their fee requirement to the whatever level they see fit. The differences between full blocks enforced by design and blocks with room to spare seem to me to be:
- resource usage, with the accompanying repercussions. To me, Blockchain bloat is more of an issue than the resources it takes to transmit a single block.
-transaction rate, obviously. Directly related to maximum potential userbase for direct on-chain transactions. Full blocks sets an arbitrary cap on this, and 1 mb was chosen pretty much at random.
-full blocks push out miners not motivated by direct profit. Potentially, those with large stakes in BTC success might see fit to mine even at a loss to secure the network, allowing for a larger amount of transactions. We won't know if there's no space for those transactions.
Total miner fees absent subsidies = (average TX fee) * (# of TX), right?
Full blocks caps # of TX in hopes the fee will rise. Non-full blocks hopes number of #TX will rise so fee doesn't have to.
I don't think you can say offhand that one of these clearly cannot happen.
Incidentally, and this is a bit of a digression but it worries me:
Absent block subsidies, Total miner fees = cost to secure network.
Cost to secure network / x = cost to profitably attack network.
What is the optimal "x" for network security? How much should we be spending on securing the network? Currently the cost is pretty high thanks to the block subsidy. I'll admit to illiteracy here, an estimate for optimal network maintenance cost as % of market cap must have been hashed out at some point. I don't think aiming for current levels is desirable. Actually, if it turns out the security cost can't be lowered I think I'll have to take a much dimmer view of Bitcoin's prospects.
As for why I think mass adoption is necessary, I should probably qualify that a bit. I think it's important that the network not consolidate into the hands of a few large players very early on. I'd prefer it never did but that's likely a pipe dream. Currently, the hypothetical big players of crypto in the future are still small, because they require crypto to build the businesses that will make them big. Big players right now, in turn, are not likely to do anything very interesting at all with crypto, because they're already successful even without it, and are too entrenched in the financial system to risk upsetting it. So in a nutshell, my worry is that capping userbase too low too early will lead to a situation where the only parties with the clout to use the system are those with no interest in running it.
Finally, 2nd layer transactions - in principle I have no problem with these, it's the accompanying reduction in trust that worries me. Not trustworthiness when it comes to single transactions, I worry about cascading failures. The risk would depend on the particular off-chain solution used, but overall I worry that for convenience's sake, pushin most users off the blockchain will result in them choosing convenient, centralized options. I worry we'll end up with most users of crypto actually transacting in IOUs or other nontransparent assets. I worry about linchpins in such systems - what if mt gox had been the Paypal of BTC?
With on-chain transactions, you can prove holdings. It's not done enough but at least it can be done. IMO we need to increase the real trustworthiness of payments in BTC, not decrease them. Otherwise we risk cascading defaults, with people accepting as payment, and passing on again, assets whose backing they're not able to assess.
Solve that for off-chain transactions, I'm all for them. However then I'd ask what we need Bitcoin for if this problem can be solved for a lower resource cost.
Since you believe "there's no point to Bitcoin without censorship-resistance" I must apologize for misidentifying you as a Gavinista.
But you're still a fellow traveler until proven otherwise!
Propagation for 1MB blocks requires >2Mb upstream, so transmission is currently more of an issue than storage. This issue is exacerbated by the superlinear validation times of larger blocks (and Galvin's 100k max tx size KwikFix turned out to be a non-starter).
Tx prices are of course a function of supply and demand. Let's not try to predict or make assumptions about the elasticity of their respective future (especially medium term) curves, because Peter R has single-handedly done enough such crystal-ball-wanking for the entire community.
You've missed the huge issue of powerful/well-connected miners/pools making larger blocks to leverage the advantages inherent to their position at the expense of weaker/less-connected ones.
That's a much larger problem than your (original?) concern about charity miners' voices being silenced.
And as larger blocks cannot be propagated/validated quickly enough, SPV mining (and flaky hacks on the concept) ensue proportionally.
The result is self-defeating; empty blocks increase in frequency as average blocks get larger, thus failing to achive the goal of higher tps despite the cost in orphans/reorgs.
Which brings us to where our design goals differ...
You are stuck in the 'payment network' paradigm, which is at this point inconveniently bound by the lack of Layer 2 and intrinsic limitations/trade-offs of Layer 1.
But in my design goal, payments are secondary (or tertiary) to wealth storage and high-value settlement.
For wealth storage (including fiat hedging and investment purposes), I can "use Bitcoin" simply by keeping my coins safe. TPS limitations are fairly irrelevant for those applications.
In high-value settlements, paying a reasonable (ie competitive) tx fee is not an issue either. Who cares about a $1000 fee when you need to settle out a Lightning (or Storm) channel worth millions?
MtGox (Magic the Gathering Online Exchange) was a toy; of course it broke and ended in tears. You need to start thinking about real Black Swans, such as the ones which will occur if Circle/Coinbase become the Paypal/Visa of Bitcoin, only to succumb to their requisite counterparty/institutional/jurisdictional/geographical/firm risks (probably along with the rest of the Petro-Fiat Empire).
Your objections to 1MB being "random" and "arbitrary" sound like Gavinista whining and social engineering. The ossification is a feature; you should accept the NSA conjecture and stop worrying about what you cannot change. If you want to test a theoretical change (like a putative perfect maxblocksize number or adjustment algorithm), start an altcoin or a sidechain, because IT AIN'T GONNA HAPPEN in Bitcoin.
Worrying about "cascading failures" is a good thing to do. I expect you will apply that motivation to our fight against features, because the more code (and girth) Bitcoin accumulates the more likely a successful attack vector will emerge.