The whole point of Bitcoin is that you cannot actually measure "majority", because of problems including censorship and sybil attacks. You can measure hashrate and PoW difficulty, which provides sybil resistance (at least, it can put a minimum cost on sybil attacks), but the Bitcoin protocol uses PoW as a dynamic membership multiparty signature which seems to add the requirement of resource requirement minimization otherwise you get fewer and fewer members in the membership set... Otherwise only high-resource (highly-coercible large-scale) entities can participate in the dynamic membership.

In the Bitcoin consensus protocol this is restricted to picking transaction ordering and some other details, but I am hoping to stress to you how the same problems apply even outside of transaction ordering and consensus over Bitcoin transaction history, even if PoW doesn't solve those other consensus problems.

The same problems (sybil attacks, etc.) apply to measuring "majority" and explain why "number of users" is not being used by Bitcoin Core developers to pick what each Bitcoin Core developer personally wish to work on. (Well, they tend to have their own individual reasons, some that are overlapping with this one, but I want to keep my explanation succinct.)

Bitcoin Core developers seem to have little desire to work on hard-fork (in)compatibility changes, although they might be okay with technically-meritorious-uncontroversial hard-forks. This is largely based on a belief of the importance of backwards-compatibility, resource requirement minimization (such as Bitcoin Core node bandwidth requirements), network effects especially in the context of financial asset adoption, decentralization, etc. Also this belief seems to be reinforced by how they don't actually control the Bitcoin network anyway, nor its rules, just one of the clients and their releases which users are free to voluntarily use (or choose not to use). Thankfully there are some alternative methods for increasing scalability which are getting deployed on testnet (or a new testnet?) today or tomorrow or something: https://bitcoin.org/en/bitcoin-core/capacity-increases-faq

Bitcoin users that disagree with Bitcoin consensus protocol rules should consider not using Bitcoin. There are many people at the moment who are not interested in what Bitcoin offers, and that's okay. Indeed some users were misled into believing Bitcoin had extremely high levels of anonymity, only to later find out that Bitcoin had a blockchain and a largely pseudonymous design.... While many Bitcoin developers wish this wasn't the case, users would be advised to stop using something that doesn't actually fit their requirements...... But "secretly" most Bitcoiners suspect that Bitcoin really does meet reasonable standards or requirements :-) so this is why many don't advocate disgruntled Bitcoin users to leave.

Some suspect that the Bitcoin industry could be coerced (such as by government) or influenced (such as by social media sentiment) to participate in a hard-fork, to specifically choose to become incompatible with the existing widely-deployed Bitcoin consensus protocol rules. Social media sentiment syncing is way cheaper than reading the Bitcoin source code, and anybody can use Bitcoin even without agreeing with the concepts behind the protocol or whatever. Anyway, a hard-fork split is not the end of Bitcoin, because even if Bitcoin-original ends up on a minority hashrate side, hashrate attacks (from the other sides of the hard-fork) can be mitigated by migrating to a different PoW function. Perhaps "adoption" will be reduced as a result (indeed perhaps for all sides of the hard-fork, esp. given wallet/merchant difficulties), but the opportunity for low-difficulty PoW mining will be overwhelmingly appealing to an extremely broad audience, plus the ability to sell UTXOs against exchanges participating on the other side of the hard-fork to fund development of -original, possibility of shaking out maladoption, and what I suspect will be an increasing unwillingness to accept payments from others using the social-media-sentiment hard-forked currency because of the difficulty of judging the technical reliability/stability/value of a currency operated by social-media-sentiment-influenced participants willing to increase incompatibility on the money they are about to pay you with.

Majority is not how Bitcoin works, not for transaction ordering, and not for what Bitcoin Core developers want to work on.

> At least to me, it sounds much more in line with the original conception of Bitcoin than a central foundation managing its direction.

Well, you're not wrong :-).