ais523 via agora-discussion [2023-05-19 04:51]: > Here's an idea I had as a way to a) shake things up in a way that's > likely to lead to lots of interesting CFJs for the next few months (I > came up with it after reading the CFJ archives for cases that looked > interesting), and b) let us experiment with mechanisms for awarding > Radiance that don't need a whole proposal cycle to go through.
I like this A LOT. It seems the right way to implement what I was going for with Golems. Actually, I would like to suggest the following: * Rebrand Raybots into Golems (aesthetic personal preference) * Make Golems into another kind of entity. The Agoran consent mechanic seems a nice way to throttle golem creation, and the promise-granting is a terrific way to make golems act-on-behalfable. As to the “other kind of entity”, I think it'd be better to define these golems as just something else other than people. We just grant them precisely the habilities we need them to have (acting on behalf, for example). But I'll elaborate. I notice a quirk in this proposal. R2152 “Mother, May I?” defines POSSIBLE as such: > CAN, POSSIBLE: Attempts to perform the described action are successful. R478 “Fora” read: > Where the rules define an action that a person CAN perform "by > announcement", that person performs that action by, in a single public > message, specifying the action and setting forth intent to perform that > action by sending that message, doing both clearly and unambiguously. And R2466 “Acting on Behalf” (needed for cashing in promises) reads: > When a rule allows one person (the agent) to act on behalf of another > (the principal) to perform an action, that agent CAN perform the action > if it is POSSIBLE for the principal to do so, taking into account any > prerequisites for the action. So, a person cashing-in a raybot's promise CAN do so if it is POSSIBLE for the raybot to do so. R478 possibly overwrites R2152's definition of CAN, which gives two possibilities: 1. It does overwrite. Then, that a person CAN perform something by annoucement means that they perform it by sending a message, etc, etc. But then, I'd argue it is not POSSIBLE for a Raybot to do it, as it can't (if it even exists) send a message. 2. It doesn't overwrite. Then, it is just explaining what “by announcement” means when under a CAN. In that case, it *is* POSSIBLE for a Raybot to do that action because the rules specify, by using CAN, that sending a message to perform certain actions *is* successful, when done. And Raybots can NEVER send messages. So “attemps to perform” such action must be successful: they don't exist, so can't possibly fail. This might be interesting. But anyway, I think better to just define Raybots/Golems as a new kind of entity and just duplicate the few capabilities they need. What do you think? -- juan
