Index ← 3924 CFJ 3925 3926 → text
===============================  CFJ 3925  ===============================

      If a Mad Scientist announces a random selection for a rule for a
      specified future week, and the ruleset doesn't change, e both can,
      and needs to, use that pick to fulfill eir weekly duties.


Caller:                        G.

Judge:                         ais523
Judgement:                     FALSE



Called by G.:                                     31 Aug 2021 14:43:34
Assigned to ais523:                               01 Sep 2021 01:46:27
Judged FALSE by ais523:                           01 Sep 2021 03:58:51


Caller's Arguments:

For the "can" part of the cfj, Rule 2655 reads in part:

       The Mad Engineer's weekly duties include the performance of the
       following tasks, in order:

       a) Randomly select exactly one rule.

Now, we've generally accepted that random picks can be separate from
action emails (e.g. use a dice server to get a confirmably random number,
in a follow-up email perform the action that uses that number).  But what
if there's weeks of ahead-warning for what the pick will be?  Are random
picks well in advance on the dependent action still random, once published?

The counterargument is that "weekly duties...randomly select" means that
the random result can't be known to anyone (must still be random) until
you get to the appropriate week.

For the "needs to" part:

If a draw is published, and later taken back (assuming the ruleset hasn't
changed etc.), then the full selection process was biased in terms of
discarding the previous pick, and wasn't determined uniformly.

Judge ais523's Evidence:

Rule 2505/1 (Power=1)
Random Choices

      When a Rule specifies that a random choice be made, then the
      choice shall be made using whatever probability distribution among
      the possible outcomes the Rule specifies, defaulting to a uniform
      probability distribution.
      The choice can be made using any physical or computational process
      whose probability distribution among the possible outcomes is
      reasonably close to that required by the Rules, and for which the
      final choice is not trivially predictable by the selecting person
      in advance. The selecting person SHOULD make the selection method
      public, and SHOULD use a method for which the final probability
      distribution can be readily confirmed.

Judge ais523's Arguments:

The first issue is to establish the time at which the random selection
would/could be performed, under these assumptions. There are two strong
points of evidence that imply that in order for this sort of random
selection to work at all, it would need to be considered to have been
performed at the time of the weekly duty (as opposed to the time of the
dice roll or equivalent): a) its validity cannot be determined too far
in advance, due to the possibility that the ruleset changes; and b)
rule 2143 requires weekly duties to be performed "each week", implying
that performing them in advance doesn't fulfil the requirement to
perform them.

However, that would imply that the Mad Scientist would have information
about what the result will be, in advance of the time at which the
action was considered to be performed. This would violate the
requirement in rule 2505 (which is a hard requirement, not a SHOULD)
that the final choice is not trivially predictable by the selecting
person in advance.

In other words, there's no timing with which the rules would be
satisfied that this sort of randomization process is actually random;
they require the randomization to not "officially" occur until the week
of the duties, but they also require the Mad Scientist to not know what
the result will be in advance. So rolling the dice prior to the week in
question is not, by the rules, treated as a valid method of