in the same way that altcoins don't dilute the value of bitcoin
They obviously do, because you cannot have both, at the same time, on the same key, moving simultaneously in the same transactions (because many people create obstacles called "replay protection", while many networks would benefit, if every transaction would be 100% replayed). So, it creates an unnecessary competition, where in theory, you have 21 million coins limit, but in practice, as long as you can flood the market with a lot of new monetary systems, people will stop using the original network. And the supply of new altcoins is unlimited: you can create as many new altcoins, as you want, abusing the maximum supply.
I believe that's the single arbitrary convention about rare sats.
That's the problem: if things would be signature-based, then you would know exactly, what is signed, and which order is signed. But if signatures are ignored, and things just fly in a way, as they are specified in a given transaction, then they can be re-ordered in many cases.
For example: some wallets sort transaction outputs in ascending/descending order, or shuffle them specifically, for example by sorting them by address types. Then, you can trace someone, if you know, which client is used by that person. And then, if sats are flying in unsigned way, you are forced to give up your privacy, to put a given satoshi in a specific place.
but that doesn't make the ones that make sense weaker imo
Of course it does. See:
https://en.wikipedia.org/wiki/Interesting_number_paradoxThe interesting number paradox was a good read, but I don't think it brings any obstacles in practice. I argue that there's a spectrum of interestingness. Sats from the first bitcoin are higher up the ladder than a sat that has a palindromic id. Ultimately, the market decides which are the most interesting ones, and that's a good thing because value, especially with collectibles, is subjective.
I'd also argue that without altcoins, there would be far fewer people onboarded in crypto nowadays. We shouldn't ignore the real attention market that we're living in. Altcoins, and especially altcoins that have good price actions serve as a very good way to onboard people to Bitcoin. I suspect neither of us knows if Bitcoin would've been better off without altcoins.
I was first very fascinated by Ethereum when I got into crypto. It seemed to me that it's a bit more potent than Bitcoin because of its programmability. It took around 1y or so for me to figure out the real difference between the two. I could say that Ethereum onboarded me to crypto, and then I turned to Bitcoin. I'm pretty sure there are many people like me out there. Even if altcoins are diluting Bitcoin, they cater to a broader market, which in turn has much higher chances of turning to Bitcoin.
You have a point of failure with the arbitrary way in which the sats flow through fees, but I believe that the broader market already agreed on the established consensus. This is also in the ord indexer code which is run by many entities/platforms, and this strengthens the consensus. If Casey wants to change this particular consensus and try to get the market to agree that the sats flow first to the miner fee, the market will reject this.