Well larger blocks are never going to be faster or as fast as smaller blocks. The goal is to reduce the latency time per kB. The faster a block can be broadcast the lower the "orphan cost" per tx. Still larger blocks will always have a higher orphan rate but they also have higher gross revenue.
...snip...
This looks promising.
As the adoption rises, the number of tx processed by the network per day will need to be increased greatly.
I'm coming back here with the idea of dynamic minimum block size... Couldn't we just index that size (or min tx count/block) requirement on the current difficulty?
Miners could then reject small blocks that don't meet the requirement (so basically, miners trying to send small blocks to have less latency would get their blocks orphaned, which looks like a good way to reduce orphan cost?).
Combined with your proposition to reduce block size dramatically, and keeping in mind that the network will also become faster and faster, that would allow us to
ensure have a better chance that the tx throughput remains consistent with the global need, progressively?