We fully agree with you on the centralization risks, but considering our architecture we might go as far as say that unless we choose to exit-scam ourselves it's pretty much impossible to exploit Whirlwind or take it down.
It's enough that ''someone'' continue DDOS attacking your website and take them down or find your real location.
Clearnet is still down because it wasn't a priority until now, we wanted to finish up all changes on the actual product before dealing with this. It's on the list for tomorrow, and once we solve it we don't expect any more downtime. It's also worth noting that the Tor service wasn't offline at all as it's completely unrelated to the Clearnet server.
I prefer using onion version myself, and we previously saw scammers hijacking clearnet websites and/or replacing addresses for other mixers.
Open-source everything and implement something similar to a Proof of Stake mechanism making the process to run a signer(validator) trustless
That would be nice, but making this open source can be a double edge sword sometimes.
I am not criticizing this idea at all, just stating my opinion and potential dangers of moving in that direction.
Open question for everyone: We could fund some wallets ourselves and offer users an extra option to receive the private key of an address that was funded before they made their deposit instead of withdrawing from the multi-sig after the deposit was confirmed. In your opinion is this something worth pursuing?
That means you and receiver would both have private keys for those addresses?
If that is the case, than my answer is No.