[Dixonary] Dealing out of turn

[Daniel B. Widdis]

Paul Wrote in the "real rules":
> Communication delays may result in two players each considering that
> the obligation to deal the next round has fallen to them. According
> to a broadly accepted proposal, the first player to post The Word is
> the dealer of the new round, irrespective of the exact operation of
> the rules.

Dodi Replied:
> The principle seems very murky to me. I don't understand
> "irrespective of the exact operation of the rules."

My recollection of the event which drove the particular consensus decision
was the occurrence of properly cast (within the deadline) votes which did
not arrive in the dealer's mailbox in time to be scored. The consensus was
that the scores would be corrected, but if a new word had already been
dealt, the original choice of dealer would stand.

It was a "communication delay" but only one player at any given time thought
they were the new dealer. The "broadly accepted proposal" was that if the
player who initially thought they were dealing had dealt, the round would
continue even if the rules/scores said otherwise after correction of the
scores.

I'm not sure if it was discussed, but it stands to reason that any other
situation (e.g., dealer scoring error, or incorrect announcement of the next
dealer) which might cause someone to start the round thinking they were the
dealer, would be handled similarly: if a deal has begun, let it go.

If, on the other hand, late-arriving (but valid) votes, or a scoring error,
or a dealer's incorrect announcement of the next dealer is detected and
corrected prior to the next deal beginning, the "correct" dealer should deal
the woid.

I do believe that this is a consensus opinion, and is in the spirit of the
game (let's play and not obsess over scores). I'm sure I'll be corrected if
it's not.

--
Dan

[Paul Keating]

> Dodi Replied:
> > The principle seems very murky to me. I don't understand
> > "irrespective of the exact operation of the rules."

The situation was as follows:

The winner of a round (A) had apparently not posted a new word within 24
hours of the announcement.

In default of the winner posting a new word, following rule 1(c), the next
player in line (B) posted a word.

But A either had posted a word, and his message got delayed; or had received
the announcement late, and so had posted late. I forget which. In any case,
A posted a word after B did.

What if A did in fact announce a word within 24 hours, but the announcement
only appeared in other mailboxes after 36 hours? And in the meantime B had
already posted? Then according to the rules, A's word would take precedence
over B's and we would have to roll back 12 hours of play. (The rule talks
about the dealer failing to announce, which A did not, so rule 1(c) does not
apply here.)

On the other hand if A received the results late, and so posted a new word
late, then he did fail to announce (though through no fault of his own) and
so rule 1(c) does apply, and B's word should stand.

And how could players tell which situation had arisen? The only way would be
to halt play and nominate a committee to examine timestamps in mail headers
and give an opinion on whether A posted in time and before B, and so whether
rule 1(c) should apply or not.

The consensus was that the first word to appear, in this case B's, was to be
regarded as the new word. This pragmatic approach relieves us of the
obligation of determining the precedence of the two announcements, which we
would have to do if we were to apply the rules exactly.

That is what I meant by "irrespective of the exact operation of the rules".

The root of the problem is that the 1990 rules assume nearly instantaneous
delivery. That was true of CompuServe, but it's never been true of SMTP, and
alas, with 91% of all email being spam, it's becoming less true every day.

--
Paul Keating
The Hague