> The absurdity lies in any discussion re: iO. The above description of unidirectional channels doesn't sound problematic, but actually instantiating the KES does.

kayabanerve: ok, cool. I just wanted to make sure there wasn't another point you were making somewhere.

<kayabanerve:matrix.org> No worries, apologies for any ambiguities on my end which enabled taking the above in ways I didn't mean.

<kayabanerve:matrix.org> Professionally, I don't Grease is absurd, but I don't think it's on good footing and I don't think an agreeable KES is probable. I unfortunately wouldn't support more time/effort on it at this time.

Development is paused anyway due to lack of funds. But if someone were interested in funding some further work with eyes wide open as to the risks, I would say this: The best KES approach _right now_ is the TEE approach via Secrt with the appropriate warning labels slapped on the side. Someone compared this approach to ecash -- I'm guessing they were referring to a fedimint type

approach. That type of federation afaict (I'm not super familiar with it) has some advantages, but i) it looks at least as complicated as Grease and ii) if the federation rugs, all the funds in the entire system is gone. A KES literally can't do anything on its own. They have to collude with a channel party and then only the funds in the channels that party is party to are

vulnerable, so the blast radius is much smaller. I might have to restrict use cases where channels are relatively short-lived (<30 days) and relatively low value to reduce the blast radious even further. Even then, you could argue that there are many applications that fit this criteria such that the UX benefits outweighing the disadvantages from the risk assumptions. And since the

KES and channel designs are almost completely decoupled, improved KES designs are by and large drop-in replacements.

Thanks for your input everyone.

I'll see you all at MK6 I presume

<kayabanerve:matrix.org> The ability to be replaced with an improvement isn't meaningful unless an improvement exists :/ TEEs and weird multisig structures don't seem to pass the bar.