Everybody already knows this, but in case some random noob reads the post, Bitcoin is not designed to facilitate micro-transactions. Yea, you can do it right now, but not in a mature network. The design of Bitcoin is most coherently designed to be used as a checkbook and/or savings account in the modern world. To get into micro-transactions, you would to fork the protocol to something like Bitcoin NG, which would have it's own pros and cons, or use a second tier network that integrates with Bitcoin, such as Lightning Network, but there will probably be others.
In terms of designing network variables for what we currently have, the maximum block size variable is either the least understood, or most propagandized with variable to create confusion and gridlock on purpose, cointelpro-style, take your pick. People love to try and make the claim that huge block sizes will destroy the fee market and reduce mining incentives to nothing. This is probably the biggest lie ever told in Bitcoin.
It doesn't matter if the block size is infinite, miners are only passing on transactions with which they can keep themselves afloat with and profit. The fee doesn't magically decrease to nothing just because the network theoretically can pass on far more transactions. The only purpose of the maximum block size variable is to prevent monopoly formation by other miners not being able to keep up with infrastructure demands to pass on blocks.
If you accept as fact that Bitcoin is not designed for micro-transactions, and agree on the above, main, functional purpose of the max block size variable, it's much easier to determine ballpark figures for what this should be. It is insanely clear that Bitcoin will likely have a large glass ceiling, just like gold does, unless on-chain transaction capability (as opposed to coinbase IOUs) is high enough to prevent fractional reserve while being used for it's intended purpose - a checking account for high dollar transactions.
In my quote below, you can see why gold has a glass ceiling, why Bitcoin has a glass ceiling, and how to fix the Bitcoin glass ceiling:
Gold is also money, yet it's inherent high friction in use and lack of granularity means that nobody actually uses it for anything. It's just stuffed into bank vaults as a "reserve". The market deemed gold was an invalid form of currency and needed something with less friction and higher granularity so gold IOUs were born, thus leading to fractional reserve. This puts a glass ceiling on the value of gold no matter who is trying to manipulate it up or down. Gold creates and spawns layers of trust based systems due to these inherent weakness discussed.
For Bitcoin to pass whatever glass ceilings it has, it probably needs at least enough TPS to process all of the world's large, high value transactions such as houses, cars, yachts, etc.
At 7 TPS, you can do 604,800 transactions per day (probably closer to 500k in reality but whatever).
There are 14,000 houses and 43,000 cars sold per day in the US. 1 meg blocks are probably too small, or dangerously close to maxing out if handling all of the world's house and car transactions. We don't need to scale to gigantic blocks, but something like 8-10 meg blocks is probably what it needs to scale to for being a major player in the world, without a glass ceiling, where most things of several thousands of dollars in value are done on-chain. This is an achievable goal without things like the Lightning network existing.
In the following post by DeathandTaxes, he's probably done the best at explaining why, even with the Lightning Network, block size would need to be increased over 1 MB still for the Lightning Network to function and not have most people on earth priced out of Bitcoin, thus hurting market cap:
Sidechains, payment channels, and cross-chain atomic transactions are decentralized system that can move some of the transaction volume off the primary chain. In essence like centralized solutions they can act as a multiplier allowing a higher number of total transactions than the number of direct on-chain transactions. It is important to realize they still rely on the primary chain having sufficient transaction capacity or they aren't trustless. As an example a payment channel could allow hundreds or even thousands of off chain transactions but it requires periodic on-chain transactions to create, adjust, and takedown. If the individual user loses direct access to the primary chain they also lose trust free solutions like payment channels. If direct access becomes prohibitively expensive then only alternative which provides sufficient scale is using trusted third parties.
Summary and conclusion:Bitcoin probably can't function without a glass ceiling that would have a considerable, negative impact on price while utilizing 1MB blocks, even with Lighting network. The good news is, it probably can function without a glass ceiling somewhere around 10MB blocks, and isn't really even required to scale significantly larger than that for maximizing market cap with it's current intended design of checkbook/savings account. It is my opinion that Satoshi made a huge error capping at 1MB, and he should have done so at a number closer to 10MB. The Gavin, continuously increasing, BIP101 proposal doesn't make sense and the only thing that should happen is a 10MB change with a wait and see approach after that.