Why would anyone NOT prefer a federated solution over a custodial one?

This whole critique paints Fedimints in most dis-favorable light, yet tries to compare it with a solution that is effectively worse on every single metric (HC), justifying it by Fedimints not being "perfect".

Yes Federations are not some golden bullet, and federated custody is a compromise comparing to a zero-custody solution, but in exchange the users gain: entirely anonymous and private off-chain solution, great scalability, great on chain anonymity set, possibility of amazing UX, and possibility of integrating of all sorts extra technologies and solutions.

I have nothing against HCs, and I find them useful and promising, and wish well open Standard Sats in all applications where they work well, but I think in this post and in some comments, not only authors have very narrow vision of possible use cases, but also commit serious logical errors.

I am one of Fedimint devs. I see plenty of possibilities for them, so I quit my previous job to work on them full time in August, and I honestly think they are the most promising and needed Bitcoin project right now. I think we have great documentation that is really honest about the tradeoffs. If you still have any questions, feel free to hit me up on Twitter (@dpc_pw) or join the Telegram group and talk it out. We're very busy building the project, so please be patient but we love to engage and educate with interested people.