Index ← 3795 CFJ 3796 3797 → text
===============================  CFJ 3796  ===============================

      Gaelan has the patent title "The Powerless."


Caller:                        Gaelan

Judge:                         omd
Judgement:                     FALSE



Called by Gaelan:                                 20 Jan 2020 04:42:11
Assigned to omd:                                  26 Jan 2020 03:02:25
Judged FALSE by omd:                              31 Jan 2020 10:08:44


Caller's Arguments:

Here’s a summary of my view on the issue. This question has been debated 
at length in the thread from my original proposal, which the H. Judge may 
want to take a look at.

The central issue here is described in a 1996 thesis by Andre, reproduced 
in full in Appendix A. The jist is this: Generally, precedence between 
Agoran rules is determined by Rule 1030, which sets out the power system 
we know and tolerate. In particular, it purports to makes any attempt by a 
lower-powered rule to claim precedence over a higher-powered rule 
INEFFECTIVE. However, what happens if a lower-powered rule claims 
precedence over rule 1030? Both rules claim precedence over the other, and 
there’s no inherent reason to prefer one, other than Rule 1030 itself.

Andre proposed a solution to the issue, which remains, sans some changes 
in wording, as the last paragraph of rule 1030/13: {
      No change to the ruleset can occur that would cause a Rule to
      directly claim precedence over this Rule as a means of determining
      precedence. This applies to changes by the enactment or amendment
      of a Rule, or of any other form. This Rule takes precedence over
      any Rule that would permit such a change to the ruleset.

I proposed what is now Rule 2604/0, which appears to circumvent that 
clause of rule 1030: {
      This rule takes precedence over That One Rule, provisions of That 
      One Rule notwithstanding. That One Rule is defined as the rule that 
      Gaelan has most recently declared, by announcement, to be That One 
      Additionally, this rule takes precedence over all rules other than 
      Rule 1030, provisions of That One Rule notwithstanding.
      Gaelan CAN, by announcement, award emself the patent title of “The

At the time of the rule’s passage, That One Rule was undefined, and 
therefore Rule 1030 would have no reason to prevent 2604’s passage. Soon 
afterward, however, came the big reveal: That One Rule is 1030! It’s been 
tricked! Now that rule 2604 is in the ruleset and claiming precedence over 
1030, we’ve reached Andre’s paradox: two rules claim precedence over each 
other, and there’s no neutral third-party telling us which to prefer.

It’s worth considering whether or not my declaration of That One Rule 
constitutes a “change to the ruleset… by the enactment or amendment of a 
Rule, or of any other form” (1030/13). I find this interpretation 
unlikely: Agora has plenty of examples of gamestate that effect the 
functioning of the rules, but aren’t part of the ruleset. For instance, 
nobody would argue that a person registering is a change to the ruleset, 
even though it grants the person new rights and obligations under the 

So, do I have the patent title? Rule 1030 would claim that I don’t: Rule 
649/41 claims that "Awarding or revoking a Patent Title is secured at 
power 1.” Of course, Rule 2604 claims otherwise. With no mechanism to 
decide between the two, I believe this is “logically undecidable as a 
result of a paradox or or other irresovable logical situation” (591/46), 
and therefore PARADOXICAL.

It’s worth noting that in the latest edition of the Ruleset, the H. 
Rulekeepor has placed rule 2604 before rule 1030. One might argue that in 
a natural reading of the rules, an earlier rule would take precedence over 
a later one. This isn’t a particularly strong argument, because Agoran 
don’t typically view the ruleset as an ordered document, but this isn’t a 
typical sitauation.

> A case of problematic precedence
> In this thesis I will present a paradox in the precedence rules for a 
Nomic game. One example of these is of course known to us all: Suppose the 
following rules would exist:
> 4000 The Virus is green. This rule takes precedence over rule 4010.
> 4010 The Virus is blue. This rule takes precedence over rule 4011.
> 4011 The Virus is yellow. This rule takes precedence over rule 4000.
> What colour would the Virus have? Rule 4000 takes precedence over 4010, 
4010 over 4011 and 4011 over 4000. So all have another rule of higher 
precedence conflicting with it. It's a problem that is known in real-life 
situations as well: If three vehicles come to a crossing, situations can 
occur in which each has to let one of the other two go first. This, 
however, is only an omission in the precedence rules. It could even be 
argued that our Rule 1030 has already overcome it, although in my opinion 
it does not (although it does alleviate some other precedence problems).
> This is not the point I want to discuss here. My example is in a way 
more disturbing, because no change in the precedence rules will 
effectively correct it.
> Suppose a rule would be enacted, with MI=1, and the following text:
> Rule 9999/0
> Andre may not deregister.
> Will it have effect? Of course not. It is in conflict with Rule 113, 
which has higher precedence. New try:
> Rule 9999/1
> Andre may not deregister. This Rule takes precedence over Rule 113.
> Again Rule 9999 and Rule 113 are in conflict. Let's look at the 
precedence Rules. Rule 1482 says: 
>       In a conflict between Rules with different Mutability Indices, 
>       the Rule with the higher Mutability Index takes precedence over
>       the Rule with the lower Mutability Index. [1 
> So, at first sight this would be no problem: Rule 1482 specifies that 
Rule 113 still takes precedence. There is more here than meets the eye, 
though. Rule 9999 says it takes precedence over Rule 113. Rule 1482 says 
Rule 113 takes precedence over Rule 9999. So, in fact there is a conflict 
between Rule 1482 and Rule 113. This Rule is of course solved by looking 
which of the two Rules takes precedence. This is Rule 1482, so in fact 
Rule 113 takes precedence over Rule 9999. However, this leads to the 
following, paradox causing Rule:
> Rule 9999/2
> Andre may not deregister. This Rule takes precedence over Rule 113 and 
> 1482.
> Again Rule 9999 and Rule 113 are in conflict. Rule 9999 and 1482 are in 
conflict, but this time - Who wins this conflict? Both claim to be the 
winner, and who is to arbitrate? The Rules can't tell you.
> As with so many paradoxes this paradox revolts around self-reference. In 
Nomic we already have the Paradox of Self-amendment, which in fact kind of 
started the whole game, which appears when the rule changing rules are 
changed [2 ], and the pardoxes that occur when a CFJ has to regulate its 
own legality or application. Here it is the precedence between precedence 
rules that causes the problem.
> What can we do about this? As will be clear from the preceding 
discussion, adding or changing precedence Rules will not solve the 
problem, and can even deteriorate it. Two ways are still open:
> Add a meta-rule (does anyone have a quasi-official set of them?) to this 
effect, for example: If two Rules regulating precedence conflict on the 
subject which of them takes precedence, then the oldest one does.
> Disallow the creation of this kind of disturbing Rules, for example by 
adding a high-MI (3 seems most logical) Rule with a text like:
>     Any Rule Change which would cause a Rule with an MI lower than three 
>     claims precedence over Rule 1482 is not allowed to take place, any 
>     to the contrary notwithstanding.
> References:
> [1 <>] Agora ruleset.
> [2 <>] Peter Suber, The Paradox of Self-Amendment: A Study of Logic, 
Law, Omnipotence, and Change. Peter Lang Publishing, 1990.


Judge omd's Arguments:

I really, really want to judge this PARADOXICAL.  It's a great scam,
firing a Chekhov's gun that has been hanging on the wall for _24
years_.  I agree with the caller that declaring a rule to be That One
Rule is not a "change to the ruleset", not even one "of any other
form", and hence bypasses the protection in Rule 1030.

However, e made a mistake.  There is one other clause in the ruleset
that can be used to resolve precedence between rules, and e neglected
to have eir rule mention it.  It's this clause (emphasis added):

Rule 217/12 (Power=3)
Interpreting the Rules

      When interpreting and applying the rules, the text of the rules
      takes precedence. Where the text is silent, **inconsistent**, or
      unclear, it is to be augmented by game custom, common sense, past
      judgements, and consideration of the best interests of the game.

Normally, when two rules conflict, "the text of the rules"
(specifically, the text of Rule 1030) states precisely how to resolve
the conflict, so we need not and cannot proceed to the second
sentence.  However, Gaelan has engineered a situation where "the text
of the rules" contains two conflicting conflict resolution mechanisms.
(E then creates a conflict between eir rule, which states that e CAN
award Patent Titles, and Rule 649, which secures Patent Titles at
Power 1.  That arguably creates an additional conflict, between eir
rule and Rule 1030.  In both cases, eir rule and Rule 1030 disagree on
which of the two takes precedence.)

This leaves the text as "inconsistent", and Rule 217 informs us to
augment it by the usual factors.  In this case, "game custom" clearly
supports higher-power rules taking precedence over lower-power ones.
So does "common sense": the ruleset as a whole is clearly designed
with that expectation, it's also consistent with the ordinary-language
meaning of 'power', and Gaelan's rule is clearly contorted to exploit
a loophole rather than being designed as a rational game mechanic.  On
the other hand, the "past judgements" factor arguably supports the
scam: Andre's thesis itself is tantamount to a "judgement" that this
case would in fact be PARADOXICAL.  But "the best interests of the
game" oppose the scam, for reasons I will explain later.  As a whole,
then, the Rule 217 factors counsel that Rule 1030 should take

Now, you might be thinking, if Rule 217 wants Rule 1030 to take
precedence over Gaelan's rule, but Gaelan's rule wants itself to take
precedence, doesn't that just create a conflict between Gaelan's rule
and Rule 217?  Rule 217 has higher Power than Gaelan's rule, so just
like Rule 649 and Rule 1030, it wins according to normal procedure,
but loses according to Gaelan's rule itself.  Are we again left
without a resolution?

Perhaps.  However, two factors suggest interpreting the ruleset to
prioritize Rule 217, one unimportant and one important..

The unimportant factor: Rule 217 and Rule 1030 versus Gaelan's rule is
a case of two against one.  Granted, there is no rule that
contradictions should be resolved by majority vote; nor is such
principle found in mathematical logic, which indeed would be
scandalized by the concept.   Still, from an ordinary language
perspective, it does strengthen the case somewhat.

The important factor is structural, having to do with how the rules
fit together.  As I mentioned, the normal precedence procedure fits
comfortably into Rule 217's "text of the rules" step.  But Gaelan's
rule is clearly worded to compete with Rule 1030 as an alternate
version of the precedence procedure, so it must also fit into that
step.  Since the four factors are a second step, one to apply after
the first step fails, they win out without needing Rule 217 to
literally 'take precedence' in the normal sense.  Treating precedence
as part of the first step is the most harmonious interpretation of the
ruleset: any other interpretation might imply that Rule 217 and Rule
1030 provided competing precedence mechanisms even before the scam,
and thus either conflicted or were redundant.

Imagine the rules involved as a stack, with each level explaining and
interpreting the levels below it:

│         Rule 217          │
│  Rule 1030  │Gaelan's rule│
│        other rules        │

It feels very natural to me for the rules to slot in this way.  The
other possible orders, with Rule 217 side-by-side with or under the
other rules, feel considerably unnatural in comparison.  For example,
if Rule 1030 were "unclear" on some point, it would be strange to say
Rule 217 didn't apply.  But it would also be strange if Rule 217
governed interpretation in some cases but not others.

This doesn't mean that Rule 217 is some kind of impregnable defense
against precedence scams.  The only reason why it matters what's
'natural' or 'harmonious' is the text of the rules is unclear about
how Rule 217 interacts with Gaelan's rule.  But Gaelan's rule could
easily have addressed the matter explicitly.  For maximum clarity, for
example, it could have stated that any inconsistencies between it and
Rules 1030 or 217 are not to be augmented by Rule 217 factors.  Even
if it had simply claimed precedence over Rule 217 specifically, as
opposed to all rules, that would at least strongly suggest that it
wanted to interact directly with Rule 217's ruleset interpretation
clauses, as opposed to letting those clauses be used to interpret it.
(Though Rule 217 also has a clause that isn't a ruleset interpretation
clause, in the last paragraph.)

And there's the rub: the reason the best interests of the game favor
defeating the scam.  The fact that the rule could have mentioned Rule
217, but did not, means that Gaelan made a mistake.  Even if e didn't
know how much importance a judge would give Rule 217, or if e thought
there was no need to declare precedence over Rule 217 if e was
declaring precedence over all rules, the mere possibility that Rule
217's mention of "inconsistent" could affect precedence would warrant
addressing it anyway.  The fact that e didn't explicitly address it
means that e probably didn't think of the possibility.  (My apologies
if I'm wrong.)

If not for that, just like in CFJ 3783, I'd be tempted to diverge from
the arguably-traditional assumption that it's always in the best
interests of the game to defeat scams.  Instead, I'd say that if a
scam is clever enough, if it identifies a real flaw in the ruleset,
it's in the best interests of the game to reward the scamster with
success.  Besides, this particular "scam" was actually passed by
legitimate popular vote, 15 to 6.  Not enough of a majority, as it
happens, for an AI-3 proposal that could have legitimately altered the
behavior of Rule 1030, but still a considerable majority.  Arguably,
then, the scam reflects the will of the playerbase, which is always a
critical factor in the interests of the game.  Now, it seems that
players mostly voted for it to test the outcome, not because they
wanted it to succeed *per se* – but that still suggests they were at
least okay with it succeeding.

However, I do think it's in the best interests of the game to punish
scammers for mistakes: it makes the scamming process more exciting if
you know that one mistake can spell your doom.  And this *is* enough
of a scam for that to apply here, popular vote notwithstanding.

Better luck next time!  Rule 649 takes precedence over Gaelan's rule,
so Gaelan could not award emself a patent title by announcement.