Some say satoshi expected consensus to be eventually reached in order to raise the blocksize, which seems pretty naive in retrospect, since it is clear now that it's almost impossible to reach said consensus, and most likely we are stuck with 1MB.
Other's say that this was completely expected by satoshi, and he knew perfectly well that the 1MB limit was set in stone, and this is just part of the Bitcoin game theory, so it can be immutable, while remaining open source and so on.
We will never know what his true intentions were, what do we know now is, no increases of the block size are going to happen anytime soon.
Instead I support block time decrease because of its proportional desired impact on decentralization.
Anyway, it is very important to understand, scalability is not addressable by either approaches because of nonlinearities involved, both should be considered like some complementary improvement.
Assuming block weight limit is still 4.000.000 weight, block time decrease also could lead to centralization on different way. Without proper research, block propagation could be slower than block time and increase block orphan rate.
Core campus argument was that scaling bitcoin by increasing block size would continue to reach to insane orders of magnitude eventually and eliminate more and more full nodes gradually.
They could simply disagree with such a linear continuous increase, instead they preferred to reject any increase proposal. Why? Nobody knows!
They could consider up to 10 times increase without affecting anything in the network in terms of centralization, but they made a taboo out of it: You are in favor of on-chain scaling? You are a bitcoin cash puppet!
Mining centralization is a real fact, I've extensively discussed and analysed it and proved that there is no 'small miner' in the network right now (other than few hobbyists I guess), it is all about very large mining farms and pools with millions of dollars worth of investments. I got mathematical proof for this, undisputable.
So, handling say 10 times more transactions in a period of time either by increasing block size or reducing block time is not a big deal for mining nodes, they are wealthy enough to use more powerful nodes. Actually, they already use fault tolerant server grade systems for this.
And guess what? When it comes to my suggestion about block time decrease, it has direct, complementary positive effect on decentralization: More distributed rewards, more chance for smaller pools to compete and remain competitive.
Finally in a Core dominated atmosphere somebody realized a non-Core related proposal got one useful feature, see? We are progressing