In the crypto space, we often approach protocol design with a focus on cryptoeconomics and game theory. This is primarily to prevent any party from gaining an unfair advantage at the expense of others. However, this can sometimes be deferred in favour of faster onboarding...
We see this with L2 sequencers. They're mostly centralized, as this makes it easier to deploy right now whilst the decentralized sequencer protocol design is worked on. The same could be said any protocol, app or game which starts out centralized and eventually becomes a DAO.
The same thinking can and should be applied higher up the stack in app and game design. E.g if my game has an in-game virtual currency, does that need to be an ERC-20? Couldn't it just be data in a db to start with? Players news to web3 won't initially care anyway.
Although this is primarily to speed up development and onboarding, it's entirely possible that such a "pseudo decentralized" architecture remains unchanged into the future as it turns out to be "good enough" for most use cases. We should embrace this more as a space.
A decentralized architecture in crypto is preferable, but not if the UX would be decidedly inferior. For some apps and protocols, we will have to wait until web3 wallets and tokens are ubiquitous amongst internet users such that UX has then been solved at a more root level.
In the meantime, let's work on best-effort architectures that integrate crypto in a beneficial manner so that we can innovate today and onboard more people into crypto. Better yet, design in way that allows decentralization to be progressively realized in the architecture.