By increasing the blocksize to an amount high enough to be able to deal with VISA's transactional volume (and we are looking at 25k transacions per second only at VISA, if Bitcoin has to be a global currency usable for everyday-tasks, we need much more) you would never be able to get around centralizing the network due the nodes being handled to datacenters, which are corporations, which are ultimately controlled easily by governments. You are handling control at layer 0.
With LN, everything happens at layer 1, layer 0 is still a decentralized network of nodes.
This is why BCash is dead and is not real competition when it comes to the idea of scaling Bitcoin.
Playing with block sizes or timing is not the answer and it comes down to having more and more data all being stuffed
into a database that is already 200gb in size and then trying to distribute this to 20,000 machines and keep them all in
sync.
Trouble is you cannot just jump to the top of block to calculate the value of a wallet and you have to go back down
the chain (Link list of block headers) and check all the history of each coin or part coin in the wallet to decide if the
wallet has the money to spend. Processing 200gb of data is bad and it will only get much worse.
Repeat this 20,000 times you see why it costs 90KWH of electric just to process 250 bytes of data
BLOCK-CHAIN DOES NOT SCALE and having a two tear system using lighting and hubs (Banks) giving out IOU's
is not the answer and it will create more trouble than it solves.
Read up on IOTA Tangle or better still HashGraph and do not support CPU wars, silly high fees or PoW because from
a programmers perspective it is a joke and the crypto guys have taken over the board-room and are trying to design
systems that they know nothing about and this is why BTC is slow.
Yes i am sure I could come up with a few radical fixes using coordinators and a tree type structure for the mass of
data (200gb just now) and split it up by node groups but it would be better to start again from scratch and just
learn from the mess Mr "s-m" has feed to the world.
Both IOTA and Hashgraph have been refuted already by creditable people in the field. Tangle is unsafe, and Hashgraph is closed source trash. Who would trust their money on that thing?
Bitcoin doesn't even need to scale if you want a decentralized gold, but lightning network can work good enough for smaller transactions.
None of the altcoins present convincing replacements for PoW. PoW is still king.