==============================  CFJ 2086  ==============================

    I CFJ on the previous statement.

========================================================================

Caller:                                 ehird

Judge:                                  avpx
Judgement:                              


Judge:                                  G.
Judgement:                              UNDECIDABLE

Appeal:                                 2086a
Decision:                               REASSIGN


Judge:                                  ais523
Judgement:                              FALSE

Appeal:                                 2086b
Decision:                               REMAND


Judge:                                  ais523
Judgement:                              FALSE

========================================================================

History:

Called by ehird:                        15 Jul 2008 22:27:55 GMT
Assigned to avpx:                       17 Jul 2008 06:15:06 GMT
avpx recused:                           22 Jul 2008 03:09:17 GMT
Assigned to G.:                         22 Jul 2008 04:36:49 GMT
Judged UNDECIDABLE by G.:               22 Jul 2008 16:35:01 GMT
Appealed by root:                       22 Jul 2008 19:46:50 GMT
Appealed by G.:                         22 Jul 2008 20:02:09 GMT
Appealed by Zefram:                     22 Jul 2008 20:54:47 GMT
Appeal 2086a:                           22 Jul 2008 21:06:30 GMT
REASSIGNED on Appeal:                   08 Sep 2008 22:48:05 GMT
Assigned to ais523:                     14 Sep 2008 01:29:15 GMT
Judged FALSE by ais523:                 16 Sep 2008 18:31:33 GMT
Appealed by ehird:                      16 Sep 2008 18:47:38 GMT
Appealed by root:                       16 Sep 2008 18:54:44 GMT
Appealed by woggle:                     16 Sep 2008 19:01:49 GMT
Appealed by G.:                         16 Sep 2008 19:10:36 GMT
Appeal 2086b:                           16 Sep 2008 19:10:36 GMT
REMANDED on Appeal:                     02 Oct 2008 15:45:33 GMT
Assigned to ais523:                     02 Oct 2008 15:45:33 GMT
Judged FALSE by ais523:                 03 Oct 2008 17:54:04 GMT

========================================================================

Judge G.'s Arguments:

Past precedents have held that these statements should be trivially
true (the cross-reference is a trivial perturbation of previous cases).
However, despite much discussion, past precedents have not considered
the timing of speech acts.  So that must be considered, and there is
no precedent on that issue whatsoever.

The following things are correct:
  1.  Immediately prior to a "speech act", a speech act is false
  (it hasn't happened yet).
  2.  Immediately after a successful "speech act", the speech act
  is true (it has happened).

But the rules are silent about the "instant" of the speech act.  That
instant of time is the one in which the CFJ must be evaluated.  Approaching
the jump from true to false as a limit case from either side shows us that
there is a discontinuity at the "instant of action".  This discontinuity
is not defined as either true or false in any literature, common defintion,
or Agoran ideas that I can find.  It is, however, very in keeping with fine
Agoran traditions to treat linguistic discontinuities by applying loose
mathematical principles.  While the past precedents give the implication
of that the limit case is approached from the "true" side, this is not
actually supported, and in any case has not been carefully considered in
those precedents (why not the false side?  No precedent has said).
It would be possible to define the discontinuity as true or false in the
rules, but we have not done so. So, quite literally, the truth value of
the CFJ is that it is
                                    "not capable of being accurately
        described as either false or true, at the time the inquiry
        case was initiated"

and so I find UNDECIDABLE.

Note that this judgement does not address the legality of an action, and
it also does not make it undecidable whether an invalid action is a lie
(that would be approaching the limit with a FALSE from both sides, so no
discontinuity, though whether this leads to false at the instant of action
isn't convered here), so I don't think this "type" of undecidable would
lead to wins by paradox in the current system.

========================================================================

Appellant root's Arguments:

The judge argues for a judgement of UNDECIDABLE on the basis that no
prior precedent has examined the merits of either TRUE or FALSE in the
light of eir mathematical analysis.  However, e also makes no attempt
to examine the merits of either TRUE or FALSE and makes no effort to
positively show that the statement is incapable of being accurately
described as either true or false.

Additionally, the judge's leading statement ("Ha, ha... My
judgements...") leads me to believe that the judge either has lost eir
marbles or intended this whole judgement as a joke.

========================================================================

Appellant G.'s Arguments:

While I have expressed previous opinions on the truth value of speech
acts, it was the looped timing of the cross-referenced CFJs here (within the
instantaneous message) which led me to think hard about this timing of
transition vs CFJ, and kept me from following the trivially assumed precedent
of "true".  So to me, at least, it was a "fresh look".  Appeals court or new
judge, please take a good hard look at the issue of the discontinuity and
help me understand why it should go on one side or the other (besides the
fact that "we've done it that way so far"...so far we haven't considered
the issue while glossing over this timing issue, so to the extent that its
part of custom it might be due to collective ignorance and bad assumptions
as much as anything).

I made it clear, and analyzed in the judgement, that (a) before the speech
act is made, it is FALSE, and (b) after it is made, it is TRUE.  I tried
(and failed) to find linguistic/logic evidence of what would be considered
at the "moment of transition."  You can get as close as you like on either
side with a true/false but not to the actual "time of action"!  I am not an
expert in these fields so I relied on general mathematical consequences of
undefined continuities, but literally, that in-between-moment *is* neither
true nor false as far as I can tell, and so literally and precisely fits the
definition of the given judgement.  So I look forward to becoming better-
educated  (and not just opinied to "cause it's obvious because it's, um
obvious") on why I am incorrect!

========================================================================

Appellant Zefram's Arguments:

I support, and point to the analysis that I just gave in a message to a-d.

[Said message is included below.]

Kerim Aydin wrote:
>The following things are correct:
>  1.  Immediately prior to a "speech act", a speech act is false
>  (it hasn't happened yet).
>  2.  Immediately after a successful "speech act", the speech act
>  is true (it has happened).

No.  The statement of action is in the present tense (otherwise it
wouldn't take effect as an action).  Prior to the time the statement
is made, it is false, because the action is not happening (it has yet
to occur).  After it has been made, it is *again* false, because the
action is again not happening (it has already occured).  The statement
is true at, and only at, the instant it is made.  (Well, it can also
be true at times that similar actions are taken.  But in general, it is
only true at this instant.)

For the record, I am dubious about this interpretation of a statement
being made, and action being taken, at a particular instant.  Making the
statement is a process which takes non-zero time, and the statement's
truth is evaluated (and any actions take effect) in the context of that
process.  This context is not pinned to any particular instant in time,
although we have historically assigned a particular instant for many
legal purposes.  Some day this distinction will matter, but I think in
this case the "instant" model is workable.

========================================================================

Judge ais523's Arguments:

According to rule 478,
{{{
Any action performed by sending a message is performed at the time
date-stamped on that message.
}}}

Therefore, it is incontrovertible that, according to the rules, CFJs
2086 and 2087 were filed simultaneously. (It's also worth pointing out
that the Holiday rules interact with various timing rules, but I won't
consider the interaction here because no Holidays are involved.)

Rule 591 says, in part:
{{{
      * FALSE, appropriate if the statement was factually and
        logically false at the time the inquiry case was initiated

      * TRUE, appropriate if the statement was factually and logically
        true at the time the inquiry case was initiated

      * UNDECIDABLE, appropriate if the statement was logically
        undecidable or otherwise not capable of being accurately
        described as either false or true, at the time the inquiry
        case was initiated
}}}

This makes it clear that the statement's truth must be evaluated at the
time that the CFJ is initiated. So therefore, CFJs 2086 and 2087 must
have the same truth-value, which is the truth value of the statement "At
the time CFJs 2086 and 2087 were initiated, they existed.". This seems
to me to depend mostly on the definition of the word "initiate" that is
used here; the word is not defined by the rules, so the standard English
definition is meant here. My understanding of the meaning of that word
in general (when used as a verb) is that either you initiate something
that already exists to cause it to start in some way, or that you
initiate something that doesn't exist to cause it to exist; the first
meaning is more common, but does not seem to apply here (the CFJs
clearly didn't exist /before/ they were initiated), so I take the
second. My reading is that CFJs are initiated /at/ a moment in time when
they do not exist, and come to exist as a result; therefore, the CFJs
came into being because they were initiated, and did not exist at the
moment at which they were initiated. (Goethe's argument that it is
undecidable whether something exists at the moment it is initiated is
interesting, but ultimately I think this is a matter of semantics, and
the semantics of 'initiate' being unclear is not a reason to claim that
they are undecidable. Maybe a rule that defined it explicitly would help
to clear things up, here.)

Arguably, IRRELEVANT is also appropriate, but I may as well clear this
point up now to save the same arguments again once the CFJ does become
relevant.

Therefore: I judge CFJ 2086 FALSE, CFJ 2087 FALSE.

========================================================================

Appellant ehird's Arguments:

See, "initiation" is the process by which something not existing is
made to exist.

{moment before initiation | doesn't exist}
{moment of initiation | transitioning from not existing and existing}
{moment after initiation | exists}

If something is transitioning from A to B it is neither A or B.

Therefore both neither exist or don't exist.

Therefore they cannot be judged TRUE or FALSE.

Therefore they should both be judged UNDECIDABLE.

========================================================================

Appellant root's Arguments:

On Tue, Sep 16, 2008 at 12:50 PM, Ian Kelly <ian.g.kelly@gmail.com> wrote:
> On Tue, Sep 16, 2008 at 12:47 PM, Elliott Hird
> <penguinofthegods@googlemail.com> wrote:
>> 2008/9/16 ais523 <ais523@bham.ac.uk>:
>>> Therefore: I judge CFJ 2086 FALSE, CFJ 2087 FALSE.
>>> --
>>> ais523
>>>
>>
>> With two support I intend to appeal this judgement.
>>
>> Arguments:
>>
>> See, "initiation" is the process by which something not existing is
>> made to exist.
>>
>> {moment before initiation | doesn't exist}
>> {moment of initiation | transitioning from not existing and existing}
>> {moment after initiation | exists}
>>
>> If something is transitioning from A to B it is neither A or B.
>>
>> Therefore both neither exist or don't exist.
>>
>> Therefore they cannot be judged TRUE or FALSE.
>>
>> Therefore they should both be judged UNDECIDABLE.
>
> Why is your interpretation of the moment of initiation better than ais523's?

Never mind, I thought of a good reason.  Often we assume that we can
initiate a CFJ / submit a proposal / agree to a pledge / etc. and then
further manipulate that newly created instance within the same
message.  ais523's judgements imply that we can't, which is contrary
to game custom.  On those grounds, I support tusho's appeal of CFJs
2086-7.

========================================================================

Appellant woggle's Arguments:

I also support the appeal of CFJ 2086-7. As root has stated,
long-standing game custom allows the actions of a message to have some
ordering in legal time even though they take effect at the same
instant of physical time. (I do not believe that tusho's arguments
have merit: although a light that is turning on may be neither off or
on, it is one of on or not on.)

========================================================================

Appellant G.'s Arguments:

I support this appeal and thus appeal it.  This should be considered
as a fencepost problem; things that are considered "at the time of"
are being considered in terms of continuous time (be it legal or real
time).  Instants in time that border two periods of continuous time
are in neither, except by arbitrary definition.  The arbitrary
definition chosen should reflect game custom if it is not explicitly
defined.

========================================================================

Judge ais523's Arguments:

Some light can be thrown on this case by considering the history of the
relevant part of rule 478.

It first shows up in rule 478/11:
{{{
      If the Rules require a Player to publish certain information,
      then e is considered to have satisfied that requirement at the
      time e publicly sends a message containing the information e is
      required to publish.
}}}
Rule 478/13 removed this paragraph again.
The modern phrase was added in rule 478/15:
{{{
      Any action performed by sending a message is
      performed at the time date-stamped on that message.
}}}
So before 6 August 2005, the rules were sufficiently different that the
current situation wouldn't apply.

There also used to be a rule 1527 which is very relevant to this
discussion, so I provide its history here, even though the rule in
question no longer exists.

The earliest recorded version is either 1527/0 or 1527/1, and looks like
this:
{{{
      Whenever a message contains more than one notification, report,
      or other communication in which the Rules place some legal
      significance, the communications in that message shall be taken
      to have been sent sequentially, separated by infinitesimal
      increments of time, in the order which they appear in the
      message.
}}}

The last version, before it was repealed, was 1527/4:
{{{
      Whenever a message contains more than one note on which the
      rules place some legal significance, the notes shall be taken to
      have been sent sequentially in the order they appear in the
      message.

      A tangle is any set of two or more notes in a message, none of
      which can be determined to appear in the message before the
      others.  If a message containing a tangle does not provide an
      order for the notes in that tangle, then the notes in that
      tangle shall be considered ambiguous and without effect if the
      effects of those notes on the gamestate would depend on the
      order in which they take effect, discounting the mere fact of
      their taking effect in a different order.
}}}
"Notes" seems not to have the present meaning; a comparison to 1527/3
seems to imply that the meaning then was something like "action,
notification, report, or other communication", although I can't be sure
without investigating more of the ruleset. The rule was repealed on 27
August 2006, relatively recently, by proposal 4866; until then, the CFJ
would have been ambiguous.

A copy of proposal 4866 is archived at
<http://www.agoranomic.org/cgi-bin/mailman/private/agora-official/2006-August/
002683.html>. The proposal made radical changes to the ruleset, along with two
others submitted at a similar time, and seemed to have been part of a large
ruleset simplification. (Goethe was responsible; I invite em to shed more
light on the circumstances.) Rule 478 was amended at the same time, but mostly
to replace "Registrar" with "Herald", and the relevant parts of the rule did
not substantively change.

There are two arguments for what happened at this point. One is that the
repeal of the in-order rule meant that from then on multiple actions in
the same message happened simultaneously rather than in order. The other
is that the rules before the change would imply actions in order, and
even after the repeals this was still game custom. I have been trying to
establish evidence as to which, but Agora went into a lull after the
Repeals in question. I've looked through the a-b archives from then, and
surprisingly the next message for which it matters was sent on 18
December 2006, several months later (and is visible at
<http://www.agoranomic.org/cgi-bin/mailman/private/agora-business/2006-Decembe
r/005977.html>. The situation was that Agora was going very slowly, and had
slowed down even further due to players deregistering for the holidays (as
going On Hold had been repealed back then). Goethe sparked off a flurry of
activity by publishing a Writ of FAGE, and Murphy (then CotC) took multiple
actions in a response a bit later, creating a CFJ, assigning it, rotating, and
assigning CFJ 1594 all in the same message (and incidentally winning by
paradox, due to a split gamestate in which each of the judges of CFJ 1594
ruled that the other's gamestate was correct). It now seems apparent that
Murphy's assignment was part of a scam, to get the CFJs assigned in just this
fashion. There was sufficient controversy around other things (mostly around
whether Writs of FAGE worked at all) that nobody questioned the use of
multiple actions in a message at the time; most of the players active had been
around since before the Repeals, which changed a lot, and it's likely that
they merely assumed that the multiple-actions thing would work.

There were many other breakages caused by such a large change to the
ruleset, some of which were fixed by proposal; there is also evidence
that many players around that time weren't really paying attention to
the rules and assumed that they continued working the same way. So I
would argue that, in late 2006 and early 2007, game custom was as if it
would have been if rule 1527 has still existed.

Since then, the game custom in question seems to have perpetuated via
new players picking it up from old players; the custom of performing
multiple actions one after another is an obvious habit for people to
fall into. It's not clear if any of this was legal, but in passing I
note that rule 1551 stood relatively intact through all this, so most of
the things that might have been affected by this has since been
ratified, and it seems unlikely that we need a massive gamestate
calculation.

There is already support in the ruleset for things happening
simultaneously (as in 'at the same time') but ordered, in the resolution
of proposals that cause multiple rules changes. Rule 105 says:
{{{
Rule changes always occur sequentially, never simultaneously.
}}}

Before proposal 4866, it was possible for actions to happen "at the same
time" but in a sequence (rules 478 and 1527 co-existed happily for
ages), and it seems that game custom that this is possible is relatively
strong. (This is similar to root's and woggle's arguments on the case.)
Things could have gone quite differently if back in 2006 someone had
tried to scam the absence of rule 1527; but by now, it's reasonably
established what game custom says the new ruleset means. I think that
under the current ruleset, when multiple actions are given in a message,
it is up to the author to specify which order they happen in to avoid
ambiguity; but that there is a very strong presumption that they are
intended to happen in the order given in the message, in the absence of
evidence to the contrary. (This is the best interpretation I can make of
a missing rule 1527; in the absence of a rule specifying ordering,
people sending the message can choose.) When ehird filed the CFJs in
question, there was no indication that any intention was meant other
than the CFJs being filed one after the other.

With that established, all that is needed is to look at the exact
meaning of the statements ehird CFJd on, to determine the correct
FALSE/TRUE/UNDECIDABLE status of the statements.

The statement of CFJ 2086 is "I CFJ on the preceding statement". The
preceding statement was "I CFJ on the following statement"; at the time
CFJ 2086 was filed, ehird had not up to that point generally CFJ'd on
that statement (and was in the process of CFJing on the statement "I CFJ
on the preceding statement".) Therefore, CFJ 2086 is FALSE.

The statement of CFJ 2087 is "I CFJ on the following statement". The
following statement was "I CFJ on the preceding statement"; ehird had
just CFJd on that statement, so at the time that CFJ was called it was
true that ehird CFJd on that statement. (Note that the most natural
interpretation of "I CFJ on the following statement" is as a speech
action for the filing of the CFJ itself, but as a statement for the
statement of the CFJ.) Therefore, CFJ 2087 is TRUE.

I judge CFJ 2086 FALSE and CFJ 2087 TRUE.

TLDR: Under the current ruleset, actions in one message do not happen
simultaneously unless explicitly specified (and sometimes this is
prevented by the rules anyway), and by default happen in the order given
in the ruleset.

========================================================================