One of the most common arguments against any proposal to increase Bitcoin's block size limit is that it would inherently reduce the decentralisation of the system. This is of course true, but (when you back away from the tribal politics that have recently emerged where your position on one technical issue is used to signal your entire philosophical/political alignment towards cryptocurrency) pointing this out seems pretty pointless if not disingenuous once you realise that there are no incentives, formulas, or processes in the Bitcoin system at all that are aligned toward increasing or maintaining any amount of decentralisation (not that any particular level of decentralisation that would be desirable has been proposed). It's quite the opposite in fact. From mining to data storage, centralisation is persistently economically rewarded by the Bitcoin network. This is well reflected in the fact that Bitcoin is already much more centralised than many would like, as core developer maaku points out here.

Decentralisation is obviously a very important value for many members of the community (myself included), but from the point of view of Bitcoin's currently implementation, it does not have any role in dictating the allocation of resources and therefore essentially does not exist. The Bitcoin network is a monetary system that plays by economic rules. It is politically neutral, which also means that there is nothing about it that is fundamentally opposed to the corrupting influence of humans that most of its proponents seek to eliminate. If you really care about decentralisation, then shouldn't you worry about fixing this broader issue instead of scapegoating one particular aspect of the system (the increasing demand for larger blocks) that is not in any way inconsistent with its broader context? You can't enforce a principle like decentralisation with piecemeal, band-aid solutions, because people will always find a way to get around the intended consequences of a system unless they are made explicit.

That raises one of the other fundamental problems with with the decentralisation debate: it has not been defined in a useful sense. What is decentralisation, empirically speaking? How do you measure the amount of decentralisation in the Bitcoin network? If you can't measure or objectively define something, then bringing it up in a debate that is fundamentally about technology is simply signalling ideological affiliation, not proposing a solution. If decentralisation advocates want to incorporate the issue into Bitcoin policy making, then they must come up with a method to measure the results of those policies. Simple metrics like how many full nodes are running aren't the answer (and don't capture the full picture, as we know from the large number of Chinese nodes that benefit from the smaller block sizes and yet simply by being run on hardware in China are basically under the de facto control of the Chinese government) since it is trivial for one entity to masquerade as many full nodes. Any proper measurement of decentralisation must account for and work around such Sybil attacks. This is part of the problem that Bitcoin was designed to solve in the first place so it won't be easy. Authentically measuring the "independence" of any given node/miner might be impossible, requiring the formulation or discovery of a clever proxy measurement (like proof of blockchain fair sharing to give an example of that kind of thinking) that correlates strongly with overall decentralisation by its very nature.

Then once you've established how to measure decentralisation, you will have to change Bitcoin's core formulas to incentivise its increase. Instead of "proof of work", you might have a "proof of independent work" value that incorporates some degree of weighted incentives for independent mining. If you were going to incentivise running full nodes as has been proposed, then you would want to reward independent nodes more heavily than a large number of nodes run by one entity. Yet you would also have to make sure that you aren't sacrificing overall mining security or the speed of the network in pursuit of these goals. Of course all of this in itself would be a dramatic change to Bitcoin that could very well be just as controversial and opposed as XT.

My main point is that there seems to be a fundamental conflict underlying the block size issue that is not being addressed. A large portion of the Bitcoin community is strongly in favor of decentralisation, but nothing about the basics of Bitcoin's design encourages it. Its P2P topology certainly allows for it, but ultimately does nothing to counter the economy of scale that does not favor it. I certainly don't have the solution to this very complicated problem, but I don't think that anybody benefits from ignoring the issue. As it stands, Bitcoin will inevitably continue to centralise simply as a result of basic economic laws. If decentralisation in Bitcoin is truly a desired goal, then it must be explicitly baked into the system or it won't happen. If decentralisation advocates are serious, then it's time to start addressing that fact. Block size is only one small part of the picture.

As far as the raising the block size limit goes, the only other argument that I can see against it other than the decentralisation aspect is the fear that blocks that are too large might literally break the network and make it impossible to process transactions. But obviously nobody is just going to allow Bitcoin to stay broken. The whole issue would be fixed relatively quickly just as all other network-breaking bugs have been. It would be an inconvenience, but hardly the existential crisis that many see a centralised future for Bitcoin as. So if you really are in favor of decentralisation yet only focus those concerns on block size issues then you are still being myopic. Block size is a red herring.