On Sun, 2019-12-29 at 03:32 +0100, Ørjan Johansen via agora-discussion wrote:
> Rule 1551 states:
> 
>                                                the gamestate is modified
>         to what it would be if, at the time the ratified document was
>         published, the gamestate had been minimally modified to make the
>         ratified document as true and accurate as possible; however, if the
>         document explicitly specifies a different past time as being the
>         time the document was true, the specified time is used to determine
>         the minimal modifications.
> 
> Back when we introduced the "different past time" possibility, my 
> reasoning was essentially that the "minimally modified" specification for 
> ratification is only sensibly calculable if the intuitive time for 
> "retroactively" changing the game state is the same as or very close to 
> the time for which the ratification is calculated - in particular, there 
> should be no in-between follow-on effects, since it might be _more 
> minimal_ for the ratification to ignore these rather than include them.
> 
> I claim that both the below ratification attempts, as well as the one 
> Murphy has proposed later, fail horribly in this respect, as there are a 
> plethora of possible follow-on effects between the time of the large 
> number of possibly failed emails and the time of the ratified document. As 
> a result, the true "minimal modification" may differ greatly from the 
> intuitive result we're trying to achieve.
> 
> The simplest way I can see to fix this is to pair each dubious email with 
> its own ratifying document, specifying the date stamp of the message as 
> the time it was true.

What about ratifying a-b and a-o as not having been public fora? (To
BAK, obviously.) That seems to make all the potential knock-on effects
clear in an easily understandable way.

-- 
ais523

Reply via email to