This is a fantastic question, not naive at all, and it's really at the heart of what the debate is all about. This is a really complicated issue, hence all of the vitriolic discussion. I'm going to try to take a crack at it and fail. I will mostly be reiterating ideas that are from this[1]. I'll try to be brief, and you can read more from [1]. I am also very opinionated on this issue, and I will try to be as impartial as I can.

Something that's really confusing is that the word "fork" actually means at least two different things in the block-size debate context. The first is forking the bitcoin software, and the second is forking of the bitcoin blockchain/network. Bitcoin XT, Hearn's project, was the former, a software fork, that would cause a "hard-fork" in the blockchain. Hard-forks in Bitcoin are very dangerous. There has never been an intentional hard-fork of the Bitcoin blockchain since its inception 7 years ago.

There's a very difficult question of just simply, "how do hard-fork?" A hard-fork would separate the p2p network into two different networks with incompatible rules. A hasty hard-fork could very easily destroy people's money and bitcoin all together. Many, including myself, strongly disagree with Bitcoin XT's hard-fork procedure.

It's also not clear that this particular software fork, Bitcoin XT, is better. I'm not going to go into that issue here as it's extremely complicated. We have an alternative solution, segregated witness, which is effectively equivalent to Bitcoin XT's short-term plan implemented as a blockchain soft-fork. Soft-forks are significantly less dangerous as they do not segregate the network.

[1] https://bitcoin.org/en/bitcoin-core/capacity-increases-faq