On the Continuity of the Agoran Judicial System

An Agoran Thesis by G.
6 June 2020

[
The main body of this thesis (minus the introduction) was first published
in March 2020, as part of Gratuitous Arguments for CFJ 7:
https://faculty.washington.edu/kerim/nomic/cases/?7
and Caller's Arguments for CFJ 3822:
https://faculty.washington.edu/kerim/nomic/cases/?3822
]

INTRODUCTION

The resolution of disputes, particularly interpretive disputes about the
rules, is a key aspect of nomic.  If nomic is considered solely as a board
game, this aspect represents "house rules" that are developed to allow the
game to be played with a degree of consensus between players when faced
with textual uncertainty.  Grand Hero of Agora Nomic Peter Suber, in eir
original version of nomic
(https://legacy.earlham.edu/~peters/writing/nomic.htm), relied heavily on
the metaphor of the judiciary as it related to legislative systems:

> Nomic includes provision for Judgment (Rule 212), not merely to
> imitate government on another front, but also for the reasons that
> government must make provision for judgment: rules will inevitably be
> adopted that are ambiguous, inconsistent, or incomplete, or that
> require application to individual circumstances not specified in the
> rule or not anticipated by the framers. "Play" must not be
> interrupted; some agency must be empowered to make an authoritative
> determination so that "play" may continue.

E also stressed that such a judicial system might come to rely on
continuity and consistency:

> Judgments in Nomic are not bound by rules of precedent, for that would
> require a daunting amount of record-keeping for each game. But the
> doctrine of stare decisis may be imposed at the players' option, or
> may arise without explicit amendment as successive judges feel
> impelled to treat like cases alike. Without stare decisis players are
> put upon to draft their rules carefully, make thoughtful
> adjudications, overrule poor judgments, and amend defective rules.
> This is one way in which Nomic teaches basic principles and exigencies
> of law, even while it vastly simplifies.

Clearly, GHAN Suber had not anticipated the tenacity of Agorans when it
comes to a "daunting amoung of recordkeeping"!  It is not wholly uncommon
for Agorans to go back through years of Agoran judicial archives when
writing a new judgement.  But how consistent has that record been?  In
Agora, if a rule is changed so that it no longer supports the "existence"
of a game-created object - are we relying, in quoting ancient precedents,
on judges and judgements that no longer "exist"?

To answer that question, where better to turn than the courts themselves?
On March 19, 2020, acting as Arbitor (officer of the courts) I attempted
to assign CFJ 7 to a new judge.  CFJ 7 had been called on August 5, 1993,
just over a month before Agora began, with the notes on the case
indicating it had never been judge.  Did it still exist as an open
judgement, and could it be judged, bringing closure to a case that was
over 26 years old?  To approach this, there are two questions to consider
- the ability to factually resolve the case, and whether the ancient
dispute continues to "exist".


QUESTION 1
Do the facts of the case exist to make a determination, and are there
appropriate judgements for today that 1993 Agorans would recognize?

Again, the tenacity of Agoran recordkeepors comes to our aid - the
information required to judge this case is available.  We don’t know the
exact ruleset on August 5, 1993.  However, Zefram’s archives have two
ruleset versions that bracket this time reasonably closely:

Initial ruleset, dated June 30, 1993
Original: http://www.fysh.org/~zefram/agora/chuck0_nr_19930630.txt
In evidence as:
https://faculty.washington.edu/kerim/nomic/cases/7_ExB.txt

Ruleset dated August 28, 1993
Original: http://www.fysh.org/~zefram/agora/usenet0_nr_19930828.txt
In evidence as:
https://faculty.washington.edu/kerim/nomic/cases/7_ExC.txt

An examination of these two rulesets shows that the rules in question in
the statement (111, 112, and 219) did not change during this period;
further an examination of two rulesets suggests that other supporting
rules on precedence, etc. were relatively unchanged.  Therefore, the
ruleset to use for judging is relatively clear.

Second, the context for the case is clean.  The case involves the
resolution of a direct and straightforward conflict between initial rules.
This was very early in the life of Agora, and none of the prior CFJs (1-6)
touch on the rules involved here.  Therefore, this case provides a unique
opportunity for a “pure rules” adjudication, made free of the weight of
judicial precedent, game custom, or similar historical clouding of the
issue.

Third, I believe that TRUE and FALSE are the only 100% consistent
judgement options for the duration of the game.  In terms of alternatives
to TRUE/FALSE, the initial rules included UNDECIDED; in 1995 this was
changed to UNDECIDABLE and UNKNOWN, then later DISMISS was the only
choice.  Choices proliferated in 2006/2007 (including IRRELEVANT), then in
2014 we cut back again to DISMISS only, before moving to what we have
today.  In 2007, Proposal 5371 (https://mailman.agoranomic.org/cgi-
bin/mailman/private/agora-official/2007-August/003161.html) actively
replaced some judgements with others with this clause:
  * For each pre-reform CFJ with an unappealed judgement, the
    post-reform inquiry case has a judgement assigned to its question on
    veracity, according to this translation:
    + Pre-reform FALSE judgement corresponds to post-reform FALSE.
    + Pre-reform TRUE judgement corresponds to post-reform TRUE.
    + Any other pre-reform judgement corresponds to post-reform
      UNDETERMINED.

But this was after several of the changes in valid judgement types
mentioned above (for which full proposal texts are unavailable), so it is
not clear whether these past alternatives to TRUE/FALSE hold up today or
transitioned through different rule versions.   Throughout it all, TRUE
and FALSE have been consistent.  Therefore, a finding of either TRUE or
FALSE can be made, with a reasonably degree of factual accuracy, such that
either the 1993 Agorans or present Agorans would recognize it as a "valid"
consensus judgement on the matter.


QUESTION 2
Does the original case exist?  Can it be assigned to a judge?

For these arguments, I will use the term "likely" as shorthand for facts
that the documentary evidence supports and for which I found no
contradictory documents.

Around March 1 2020, the full text case log for CFJ 7 from the CotC
archives appeared as follows:

> Call for Judgement from Wes (Thu Aug  5 22:26:27 GMT 1993)
>
>   "Because of Rule 111, Rule 112 takes precedence over Rule
> 219. Therefore, Rule 219 has no legal force."
>
> { This CFJ was never judged.}
https://faculty.washington.edu/kerim/nomic/cases/7_ExA.txt

The parenthetical {} comment is interesting.  Similarly-delimited {}
comments appear in several cases in the early CFJ history, and all such
comments supply historical context that would be known only after the case
was called (e.g. CFJ 28 {} comments document the full history of that
case).  The current CotC (G.) inherited these case logs from omd, who
inherited them from Murphy.  Murphy (pers. comm.) recalled that e
inherited these cases from Taral.  Murphy also noted that some of these
represented after-the-events attempts (possibly long after), by an unknown
editor, to reconstruct original history.  I suggest that the editor had
access to archives of the original mailing list (i.e. to reconstruct CFJ
28 history) and performed reasonable due diligence in research before
making these annotations.  Therefore, it is likely that the indicated
statement was called for judgement by Wes on August 5 1993, and that it
was never judged.

We don’t know the exact ruleset on August 5, 1993.  However, Zefram’s
archives have two ruleset versions that bracket this time reasonably
closely:

Initial ruleset, dated June 30, 1993
Original: http://www.fysh.org/~zefram/agora/chuck0_nr_19930630.txt
In evidence as:
https://faculty.washington.edu/kerim/nomic/cases/7_ExB.txt

Ruleset dated August 28, 1993
Original: http://www.fysh.org/~zefram/agora/usenet0_nr_19930828.txt
In evidence as:
https://faculty.washington.edu/kerim/nomic/cases/7_ExC.txt

I assume these rulesets are likely accurate for the indicated dates, and
for rules quoted in the below arguments I cross-checked to see if there
were changes between the two dates.  The initial and 28-Aug-1993 rulesets
differed in capitalization of key terms, with the later version
capitalizing many terms such as Players and Statement throughout.  There
were no other changes in the rules quoted below - I quote the initial
(uncapitalized) versions.

Mutable rule 213 read:
      If players disagree about the legality of a move or the
      interpretation or application of a rule, then a player may invoke
      judgement by submitting a statement for judgement to the Speaker.
      Disagreement, for the purposes of this rule, may be created by the
      insistence of any player.  When judgement is invoked, the Speaker
      must, as soon as possible, select a Judge as described in the Rules.
      The Speaker must then distribute the statement to be judged, along
      with the identity of the Judge, to all players.

In the original R213, “Invoking judgement” is done by “submitting a
statement for judgement”.  Requests (i.e. calls) for dispute resolution
are tied to the submission of a single statement, with the intent that the
statement is “for judgement”.  Further, it ties the selection of a Judge
to a particular statement.

If we ignore the all intervening history for the moment, it can be seen
that the initial rules for judgement were/are essentially synonymous with
the current R991/31 (In evidence as:
https://faculty.washington.edu/kerim/nomic/cases/7_ExD.txt).
In R991/31, judgement is initiated (synonymous with the old “invoked”)
with the announcement of a single statement, and that statement is tied to
the selection of a judge.  I posit that these endpoints (again, setting
aside intervening history) are sufficient such that the original
definition matches the current one within the bounds of the current
R1586/9 (Definition and Continuity of Entities), to the extent that these
concepts are “the same” in the initial and current rulesets (including
concepts like statement, judge, caller, etc.).  Further, current
definitions such as “open” would continue to apply to old cases.

In the current ruleset, we know from CFJ 3488
(https://faculty.washington.edu/kerim/nomic/cases/?3488)
that judgeship is sticky - you need to be an eligible player to be
assigned, but once assigned, registration status doesn’t matter and you
are the judge whether or not you are a player.  An examination of the
original judgement rules (213-217) shows the same situation - judges only
change when the officer in charge changes them.

Additionally, if CFJ 7 had a judge assigned at some point in history, and
was never assigned a judgement, and that person was never explicitly
removed as a judge, it should be POSSIBLE for the current Arbitor to
recuse em.  The current R2492/4 reads in part:

>     The Arbitor CAN recuse a judge from a case by announcement, if that
>     judge has violated a time limit for judging the case and has not
>     judged it in the mean time; the Arbitor SHALL do so in a timely
>     fashion after the time limit expires, if able.

At the time CFJ 7 was called, R215 read:
>     After the Speaker has distributed the statement to be judged and the
>     identity of the Judge, the Judge has one week in which to deliver a
>     legal judgement.

Therefore, there was a time limit back then, and if a judge was assigned,
it is likely that the judge violated the original time limit, which would
allow the current recusal rule to function such that the Arbitor CAN
currently recuse em.  (A second possibility is that the original judge is
indeterminate, which under the current switch rules would result in the
judge being “unassigned” - so ability to recuse would be irrelevant).

It’s also important to note that the “Arbitor SHALL [recuse]” in R2492 is
a “SHALL in a timely fashion after the time limit expires”. This SHALL is
tied to an Office and a SHALL that didn’t exist when the original time
limit expired, so there is no REQUIREMENT for the Arbitor to act to
perform recusals on ancient cases.  So e CAN recuse an ancient judge who
has not judged, but NEED NOT do so (if e *does* recuse someone to make the
case clearly unassigned, R991 does require that e assign a new judge).

So the start and end points are consistent with CFJ 7 being a currently-
open CFJ, that (one way or the other) was unassigned after the Arbitor
recused the previous judge, and thus a new judge could be assigned to it.
The remaining question is: did anything in the intervening 26+ year
history of Agora inject a discontinuity that “destroyed” CFJ 7?

First off: I admit this could never be 100% answered.  There is always the
possibility that a statement like {“There are no Open CFJs”} was ratified
at some point to fix a game bug or something.  However, given no explicit
memory of such an event, I’ve reviewed the history of the judgement rules
to determine if continuity is reasonably likely.  Starting with the
historical annotations of R991/31 as a reference, I used Zefram’s ruleset
history:
Original: http://www.fysh.org/~zefram/agora/rules_text.txt
In evidence as:
https://faculty.washington.edu/kerim/nomic/cases/7_ExE.txt

for amendments 1-7 (through 2002) and for later amendments, found each
proposal resolution to include proposal actions as well as rule texts in
the analysis.

Starting with the original R213 text above, the changes from the
historical annotations on R991 were:

Amended(1) by P407 (Alexx), 03 Sep 1993
  Changed the judge selection job from Speaker to Clerk of the Courts,
  no impact on the existence of CFJs or judges.

Amended by P991 'Ultimate Reward for Achievement of Goal' (Fnord.), ~12
Aug 1994
  Adds the following text:  “No Player shall submit more than five
  CFJ's per week.”  This is an important piece of evidence.  It used the
  term “CFJ” without further explanation, while the main body of the rule
  still talked about “submitting a statement for judgement.”  This
  indicates that the term CFJ was synonymous with “submitting a statement
  for judgement” rather than being a separate or new concept.

Amended(2) by R750, around 12 Aug 1994
Infected and amended(3) by R1454, 23 Oct 1995
Amended(4) by P2042, 11 Dec 1995
  Amendments 2-4 were minor/not relevant.

Amended(5) by P2457, 16 Feb 1996
  Makes “Call for Judgement” the leading term instead of “invoking
  judgement”.  Since the two terms were used simultaneously following the
  P991 amendment above, this is not a discontinuity.  For reference, the
  full text of 991/5 is:
      Any Player who seeks formal resolution of any dispute pertaining
      to this Nomic shall be permitted to request such by submitting a
      Call for Judgement to the Clerk of the Courts.  For the purpose
      of this and other Rules, the submission of a Call for Judgement
      shall constitute proof of the existence of a dispute.

      Any document submitted to the Clerk of the Courts and which is
      clearly marked as a Call for Judgement is a Call for Judgement.

      The Clerk shall distribute the text of a Call for Judgement,
      along with any additional material submitted by the Caller
      (including, but not limited to, Arguments and Evidence) not
      later than the time e announces the identity of the first Judge
      assigned to Judge it.

Mutated from MI=1 to MI=2 by P2669, 19 Sep 1996
    Power change only.

At this point in history (1998) another rule is created that is relevant -
Rule 1868.  It reads in part:
     Once assigned as the Judge of a CFJ, that Player remains the
     Judge of that CFJ until e is recused from that CFJ, or e ceases
     to be a Player.

Here, explicitly, ceasing to be a player removes a judge automatically
(this wouldn’t affect the existence of the CFJ, just the judge).  However,
it’s not clear what this would have done if the judge of CFJ 7 had ceased
to be a player *before* the rule took effect.

In 2001, this text became:
     Once selected as the Judge of a CFJ, that Player remains the
     Judge of that CFJ until e is Recused from that CFJ or becomes
     ineligible to Judge that particular CFJ.
(which included methods other than deregistration, such as zombification,
that could make a person ineligible).

In 2007, recusal stopped being automatic:
     Whenever a CFJ that has not been judged has no trial judge
     assigned, the Clerk of the Courts shall as soon as possible
     assign an eligible judge to it by announcement.  The assigned
     judge remains its trial judge until recused.

But the point remains that the judge of CFJ 7 may have been “auto-removed”
between 1998 and 2007.

Back to Rule 991 amendments:

Amended(6) by P4170 'Justiciar Opinon' (Elysion), 26 Jun 2001
    Added the justiciar option, no effect on continuity.

Amended(7) by P4298 'Massive Judicial Reform' (Murphy), 17 May 2002
  Major rewording/restructuring, but fundamental definitions remained
  intact:
      Any person may request formal resolution of a dispute pertaining
      to this Nomic by submitting a Call for Judgement (CFJ) to the
      Clerk of the Courts.  The submission of a CFJ constitutes proof
      of the existence of such a dispute.

      A CFJ should be a single clearly-labeled Statement whose truth
      or falsity can be determined using logical reasoning, assuming
      perfect knowledge.  A CFJ may be accompanied by Arguments,
      Evidence, or other related material; the Judge is encouraged,
      but not required, to take notice of these things.

      The Clerk of the Courts shall publish the text of a CFJ, along
      with any additional material submitted by the Caller (including
      but not limited to Arguments and Evidence), no later than the
      time e announces the identity of the first Judge assigned to
      that CFJ.

Amended(8) by P4867 'Judicial Reform 2.0' (G.), 27 Aug 2006
  This created a subclass of CFJ called “civil” CFJs that were designed to
  adjudicate contracts.  However, these were a subclass of general CFJs
  which kept a consistent definition with the past, the resulting R991
  was:
       Any person may request formal resolution of a dispute pertaining
       to this Nomic by submitting a Call for Judgement (CFJ) to the
       Clerk of the Courts.  The submission of a CFJ constitutes proof
       of the existence of such a dispute.

       A CFJ should be a single clearly-labeled Statement whose truth
       or falsity can be determined using logical reasoning, assuming
       perfect knowledge.  A CFJ may be accompanied by Arguments,
       Evidence, or other related material; the Judge is encouraged,
       but not required, to take notice of these things.

Amended(9) by P5015 'CFJ by announcement' (Zefram), 24 Jun 2007
  Changed the longstanding “submit to the CotC” method to by
  announcement.

Retitled by P5086 'judicial reform' (Zefram), 01 Aug 2007
Amended(10) by P5086 'judicial reform' (Zefram), 01 Aug 2007
  Proposal 5086 was a major change in definitions that created “inquiry
  cases” (the traditional kind) and “equity cases” and removed the general
  CFJ definition.  The proposal (resolved here:
  https://mailman.agoranomic.org/cgi-bin/mailman/private/agora-
official/2007-August/003161.html)
  had a specific and detailed continuity method, including the following
  text:
   * Each pre-reform CFJ continues to exist as a post-reform inquiry
       case.
   * Each pre-reform CFJ with a judge assigned has the same judge
       assigned in its post-reform form.
   * Each pre-reform CFJ with no judge assigned has no judge assigned in
       its post-reform form.

  My reading of these clauses is that they successfully transitioned pre-
  reform “CFJs” to post-reform “inquiry cases” without loss of continuity.

Amended(11) by P5110 'Regulate ID numbers' (Murphy), 02 Aug 2007
  Made ID numbers more official - no effect on continuity.

Amended(12) by P5317 'Judicial fixes' (Murphy), 28 Nov 2007
  Clarified the hierarchy of types of judgements by making this the first
  paragraph of R991:
      A judicial case, also known as a call for judgement (CFJ), is a
       procedure to settle a matter of controversy.  There are
       subclasses of judicial case with particular features defined by
       other rules.  Subclasses of judicial case exist only as defined
       by the rules.
  The “inquiry” subclass remained continuous with the past.

Amended(13) by P5464 'Secure more judicial aspects' (Murphy), 13 Mar 2008
Assigned to the Committee on the Judiciary by P6053 'Committees'
   (Murphy, woggle, ais523), 23 Jan 2009
Amended(14) by P7050 'No more warp drive' (omd), 16 May 2011

  None of these affected basic definitions or continuity.

Retitled by P7616 'Mostly Simple Judging v2.3' (G.), 16 Jan 2014
Amended(15) by P7616 'Mostly Simple Judging v2.3' (G.), 16 Jan 2014
  Vastly simplified the Judicial system.  Converged the different types of
  cases (while being explicit about synonyms for continuity).  Resulting
  Rule:
      Any person (the initiator) can initiate a Call for Judgement
      (CFJ, syn. Judicial Case) by announcement, specifying a statement
      to be inquired into.  E may optionally bar one person from the
      case.

      At any time, each CFJ is either open (default), or has exactly one
      judgement.

      When a CFJ has no judge assigned, the Speaker CAN assign any
      player to be its judge by announcement, and SHALL do so within a
      week.  The players eligible to be assigned as judge are all
      players except the initiator and the person barred (if any).  The
      Speaker SHALL assign judges over time such that all interested
      players have reasonably equal opportunities to judge.


Amended(16) by P7629 'Do Things' (Alexis), 07 Apr 2014
  Created the Arbitor office.

Amended(17) by P7647 'formatting/misc fixes' (omd), 03 Jun 2014
Amended(18) by P7777 'Simplified moot fix' (omd), 14 Aug 2015
Amended(19) by P7803 'Self-service judiciary' (G.), 31 Jul 2016
Amended(20) by P7867 'Economics Overhaul v2' (nch, o, grok, Aris), 30
   Jul 2017
Amended(21) by P7876 'Float On' (o), 26 Sep 2017
Amended(22) by P7888 'BILLY MAYS HERE' (o, R. Lee), 26 Sep 2017
Amended(23) by P7899 'Arbitor's Union' (G.), 03 Oct 2017
Amended(24) by P8004 'Rusty' (G.), 30 Jan 2018
Amended(25) by P8005 'Hopefully routine fixes' (Murphy), 06 Feb 2018
Amended(26) by P8014 'Putting Agora on a Map v7' (Trigon, Aris, ATMunn,
   G., o, R. Lee), 24 Feb 2018
Amended(27) by P8017 'RTRW Cleanups' (Alexis), 06 Mar 2018
Amended(28) by P8065 'No undead courts' (twg), 18 Jul 2018
Amended(29) by cleaning (Trigon), 31 Oct 2018
  All of Amendments 17-29 were minor (or side-things like adding and
  removing economics) that didn’t affect continuity and without any side-
  effects from Proposals.

Amended(30) by P8134 'The judge switch' (G.), 02 Dec 2018
  Turned Judge into a CFJ switch.  Importantly, the proposal included a
  continuity clause:
      For every CFJ that was assigned to a judge immediately before this
      proposal took effect, that CFJ's judge switch is flipped to that
      judge.

Amended(31) by P8221 'Usage de-capitalization' (Jason Cobb, Aris), 05
   Sep 2019
  Decapitalized all the stuff.

I hope this thorough review is sufficient evidence to prove the truth of
the statement of CFJ 3822  - that CFJ 7 has existed continuously as a CFJ
(on occasion called an inquiry case), is very likely (to the preponderance
of evidence) existing, open, that it either had no judge or the judge was
successfully recused by Arbitor G., and as such, CFJ 7 has recently been
assigned to Jason.


CONCLUSIONS

The above arguments (to question 2) were found TRUE by Judge Aris in CFJ
3822, and CFJ 7 was resolved by Judge Jason (using the TRUE/FALSE options
that were consistent under Question 1).  Judge Aris pointed out that,
technically, DISMISS might be appropriate given that we could never be
sure that ancient judgements weren't explicitly "ratified away" or
otherwise broken by a hidden proposal or other game actiion.  However,
Judge Aris sides with the historical record (thus extending a bit of
precedent on standard of proof for historical documents):

> Given all of the emphasis we've placed on preserving our history,
> judging this case DISMISS merely because of the slight chance that
> there was some change that was missed would be an admission of defeat.
> It would be an admission that despite the efforts we've taken to
> preserve our records of the past, we could nevertheless not trust
> them. This is not how we play Agora. We play on the assumption that
> the records of the past are true, unless there is some special reason
> to doubt their validity. If we cannot find any intermediate evidence
> invalidating past evidence about the state of the game, and there is
> no particular reason to expect that changes have happened in the
> meantime, we will continue on the assumption that the state of affairs
> remains the same until it is proven wrong, thus preserving the chain
> of time linking us to our past.

Thus, as per the findings of the current legal system, the Agoran system
for resolving disputes (the Call for Judgement on the truth value of a
particular statement) has a legal consistency, and thus represents a
formal stare decisis, that stretches (at the time of this writing) for
nearly 27 years.
