On Sun, 12 Apr 2020 at 18:20, James Cook via agora-discussion < agora-discussion@agoranomic.org> wrote:
> I'm more a fan of my own. I am not sure how others feel about it. I > think I needed to work out how it would interact with quorum or > something but don't remember anything else I wanted to iron out. > > It's been a while since I read your legal fictions proto, but I think > I had reservations in it based on what I wrote in the following Feb 18 > message: > > https://mailman.agoranomic.org/cgi-bin/mailman/private/agora-discussion/2020-February/056898.html > > (I phrased that message mostly about explaining differences between > our approaches. To explain why those differences give me reservations, > I'll just add that I prefer the rules to try to spell out what happens > as clearly and unambigously as possible.) > > James > Ah, I think I forgot to reply to that! I stand by what I said: I think we're both trying to limit the effects of the change to the minimal thing implied by the ratified document, without doing complex examinations of the rest of the gamestate. The main difference is that you set out explicitly what is attempting to be ratified, while my approach doesn't do that. But I don't think that your approach helps resolve contradictions. For instance, what if a self-ratifying report lists an invalid value for a switch, such as a non-player officeholder? It seems to me that, at best, this would create an inconsistency between the gamestate and the rules and cause the ratification to fail. You could try to make each switch flip fail independently, but it would be easy to concoct a situation where the valid values for two switches depended on one another, so that either could be split out and ratified individually, but they cannot both be ratified simultaneously. On the other hand, overriding contradictory facts prevents the need for complex analysis of the contradiction: the officeholder is simply set to a nonplayer value, then immediately corrected automatically. Even for interdependent switches, it would work fine: both are set to the values which turn out to be contradictory, then both are invalid, and both reset to default. If you got rid of the bit about inconsistency between the rules and gamestate, then I don't think there's a lot of substantial difference between the choices, on the axes you care about. The only thing I can think of is that mine is unclear whether creating a legal fiction also triggers state change effects (e.g. "when X happens") but that is easily fixed. The other differences are the explicit vs implicit specification of what exactly happens, but I still don't see that as a major difference, and the retroactive vs non-retroactive approach. At this point, I'm actually thinking that retroactivity might be better there; I think it may be easier to address time paradoxes explicitly. The pseudo-retroactive approach attempts to work around this by preventing interference with itself, but when you have complex interfering ratifications, I'm not entirely sure that works. I should probably attempt to put that into an actual paradox attempt to see if it works. -Alexis