I like this fix – what flaws there are are incredibly small, and with the limit of one million sides per dice, aren’t practically going to be detectable.

I’d prefer a more direct wording - something like “the dice roller hosted at https://blognomic.com/dice/roll.php is considered to produce sufficiently random results for any random generation required by the ruleset”.

I’m assuming that people with server access deliberately manipulating the dice roller would be covered by Fair Play?

I’d think so. “should not take any action using such heightened access for the purpose of causing any Agent or Agents to gain, receive, maintain, or preserve gameplay advantage” seems pretty straightforward in that regard.

Somebody noticing and exploiting some other quirk (“the first roll on a Monday is always a 3, but the rules say that such statistical flaws are to be treated as non-existent!”) would also be against Fair Play (“unexpected behaviours in the software”).

I think I prefer my wording over yours Josh, sorry. Your wording could be seen as more a suggestion to prefer the dice roller over other randomisers, while mine is explicitly a command to ignore the faults.

@kevan I worry that adding this might actually make that problem worse?

By mentioning the statistical flaws, someone could argue they are no longer unexpected behaviors in the software but instead non-existent and thus okay for exploitation.

I think this doesn’t solve the problem. The issue is whether the dice roller is sufficient when the rules call for a random selection; this doesn’t address that question, as it specifies that errors don’t exist “for the purposes of generating random results”, which isn’t quite the same as “for the purposes of meeting the requirement for randomness”.

I can see both sides of this argument, so I’m leaving this to the Concierge. 

Sigh.

(requesting it be a REVISE rather than AGAINST if not a FOR, because we do need to get this sorted out one way or another)

I prefer Josh’s general version (maybe no need for the “for any random generation required by the ruleset”) as being less cryptic to someone reading this rule a year from now.

Okay, I’m willing to think on the wording and submit a revised version. 

Ah, actually now that I’m looking at the revise rules, it looks like it only works if I revise it at a moment where there’s enough revise votes. I can’t revise then have others come in later to push it up to the right number.

Also, I think DEF with an emperor’s Revise technically turns the DEF into an against vote?

I’d say so, on the DEFs; my REVISE vote is treated as an AGAINST “with respect to all other rules”, so when another rule tries to set someone’s vote to “the same as the Concierge’s Vote”, that rule just sees an AGAINST.