Index ← 3282a CFJ 3282 3283 → text
==============================  CFJ 3282  ==============================

    Proposal 7174 was enacted.

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

Caller:                                 omd

Judge:                                  G.
Judgement:                              FALSE

Appeal:                                 3282a
Decision:                               AFFIRM

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

History:

Called by omd:                          14 Oct 2012 03:33:07 GMT
Assigned to G.:                         14 Oct 2012 04:13:27 GMT
Judged FALSE by G.:                     01 Nov 2012 22:47:44 GMT
Appealed by omd:                        01 Nov 2012 23:31:07 GMT
Appealed by Kolja:                      05 Nov 2012 09:47:24 GMT
Appealed by G.:                         05 Nov 2012 14:57:55 GMT
Appeal 3282a:                           05 Nov 2012 14:57:55 GMT
AFFIRMED on Appeal:                     15 Nov 2012 01:30:13 GMT

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

Caller's Arguments:

On Sat, Sep 29, 2012 at 10:37 PM, omd  wrote:
> This removes the two instances of "take actions that are not otherwise
> IMPOSSIBLE by announcement".  My attempted scam is to interpret that
> as allowing taking actions normally permitted without objection by
> announcement, which happens to be the exact scam that I tried 7 months
> ago, and was ruled to fail in CFJ 3171.  (That was the one where G.
> called my arguments "either purposefully or idiotically obtuse".)
> However, some weeks later I noticed an argument that at least calls
> the judgement into question.  To quote myself:
>
> On Fri, Mar 30, 2012 at 11:30 PM, omd  wrote:
>> [The current mechanism has weird interactions with higher-powered
>> Rules and actions that are performed by sending a message rather than
>> by announcement, and I am skeptical that my recent scam failed in a
>> way that is not patchable (unless Golems are generally unable to act).
>>  In particular, even if the mechanism is part of the specification of
>> an action, since Golems are *not* sending messages and no rule is
>> pretending they are, it's equally valid to say "I cause my Golem to do
>> X without objection" as "I cause my Golem to do X by announcement"
>> (since "by announcement" is defined with reference to sending
>> messages) - in both cases the purported mechanism is a lie.  I would
>> have brought this up in CFJ 3171, but I wasn't thinking straight.
>> Durr.]
>
> You could argue that "without objection" is a more complex mechanism
> than "by announcement", but I think there's a decent chance that
> either Golems and the President have never been able to take (most?)
> actions, or the permission to "take actions that are not otherwise
> IMPOSSIBLE by announcement" extends to the mechanism of
> without-objection.

On Tue, Oct 2, 2012 at 5:04 PM, omd  wrote:
> [Proposal 7174]
> was the proposal I submitted back in February and attempted to
> distribute after causing a Golem to make everyone inactive.  I listed
> myself as the only eligible voter in the distribution and this was
> contested within a week, so the distribution was only effective if the
> inactive-making was; since I then attempted to retract the proposal
> (for simplicity), there was no purported resolution to self-ratify.
> If it happens that the scam was effective after all (I'm not claiming
> it necessarily was, but I may as well go all the way):

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

Caller's Evidence:

"The owner of a Slave Golem CAN cause it to take actions that are not
otherwise IMPOSSIBLE by announcement."

http://zenith.homelinux.net/cotc/viewcase.php?cfj=3171

Relevant messages from when 7174 was proposed:
http://www.mail-archive.com/agora-business@agoranomic.org/msg23089.html
http://www.mail-archive.com/agora-business@agoranomic.org/msg23112.html
http://www.mail-archive.com/agora-business@agoranomic.org/msg23156.html
The message in which I purported to resolve it (and shamefully lied
about my non-laziness):
http://www.mail-archive.com/agora-business@agoranomic.org/msg23943.html

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

Judge G.'s Arguments:

First, as I was an earlier judge on the same subject, if this is to be re-
examined, I request an Appeal (rather than a Motion to Reconsider) and I
request that the Appeals Court come up with a joint uphold or overturn
with some consensus, rather than a simple reassign.

>From the Caller's arguments:
>>  since Golems are *not* sending messages and no rule is
>> pretending they are, it's equally valid to say "I cause my Golem to do
>> X without objection" as "I cause my Golem to do X by announcement"
>> (since "by announcement" is defined with reference to sending
>> messages) - in both cases the purported mechanism is a lie.

As the judge, I disagree with this critical core of the argument.
The Fora rule allows any player to make an announcement by sending a message
via a public forum.  It is a simple inference from slave golem mechanisms to
say that the slave's Master is part of the sending mechanism, i.e., a slave
sends a message "via" the mechanism of eir master, which is no more strange
that saying it happens "via" wires or "via" an email client.

This is similar in precedent to saying that a golem's R101 rights were not
violated as e could deregister via eir master.  (before R101 was amended
to exclude golems, of course, but the principle is part of precedent).

This is not true of without objection, which require specific announcements
by first-class players to work.  That is, there is no way that this
mechanism can work "via" a master's single announcement alone.

The Caller later added:
> I don't think your [referring to the judge's arguments above] leap from a
> rule with a condition (actions that are not otherwise IMPOSSIBLE) and a
> mechanism (directly causing the Golem to take the action through a
> rule effect) to a different mechanism (causing the Golem to send a
> message) makes sense.  A rule saying you can cause the Golem to send a
> message (like the Promises rule) would make (more) sense, but that's
> not what this rule says.  Unless you're arguing that the "action" in
> question is sending a message.

to which I reply:  we have to make some such inferences
that this is possible; for example, the only mechanism for deregistering
is making announcements, and the previous precedent on R101 said that
this had to be possible for a golem to avoid R101 being violated, so the
strong R101 "no interpretation of Agoran law" kicks in there to strongly
we should interpret thusly.  Not to mention the right to participation
in the fora!!

Thus, overall, I hold that including mechanism as a part of "where the rules
permit" (i.e. requiring that the mechanism be followed, in accordance with
earlier judgement) does not paralyze golems, as inferring their "announcement"
ability is in keeping with the Rules on Fora, and also was required by R101.

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

Appellant omd's Arguments:

> From the Caller's arguments:
>>>  since Golems are *not* sending messages and no rule is
>>> pretending they are, it's equally valid to say "I cause my Golem to do
>>> X without objection" as "I cause my Golem to do X by announcement"
>>> (since "by announcement" is defined with reference to sending
>>> messages) - in both cases the purported mechanism is a lie.
>
> As the judge, I disagree with this critical core of the argument.
> The Fora rule allows any player to make an announcement by sending a message
> [...]
(This restates what I addressed in my later arguments.)
> The Caller later added:
>> I don't think your [referring to the judge's arguments above] leap from a
>> rule with a condition (actions that are not otherwise IMPOSSIBLE) and a
>> mechanism (directly causing the Golem to take the action through a
>> rule effect) to a different mechanism (causing the Golem to send a
>> message) makes sense.  A rule saying you can cause the Golem to send a
>> message (like the Promises rule) would make (more) sense, but that's
>> not what this rule says.  Unless you're arguing that the "action" in
>> question is sending a message.
>
> to which I reply:  we have to make some such inferences
> that this is possible; for example, the only mechanism for deregistering
> is making announcements,

Well, more precisely, "A player CAN deregister by announcement", which
is the same form as any other announcement action.  My argument is
that "CAN take actions that are not otherwise IMPOSSIBLE by
announcement", for it to have any effect at all, must be interpreted
as "if it would be possible for the Golem to take an action by
activating some mechanism, this rule provides a separate announcement
mechanism to take that action", so there is nothing preventing Golems
from having their owner deregister them.

[Historical note: Golems existed for only ten days before R101 was
amended to remove Golems' rights, but multiple cases (3169, 3173)
about those rights were called in the meantime.  3169 concluded, among
other things, that giving the owner a mechanism to deregister the
Golem was sufficient to satisfy its right to deregister, and 3173 that
this was true even if the Golem publicly protested that e wished to
deregister.]

You said that you interpret the clause as "requiring that the
mechanism be followed", but that can only mean that the clause is
interpreted as "CAN take actions that are not otherwise IMPOSSIBLE,
using a mechanism that is a combination of 'by announcement' and the
mechanism that makes them POSSIBLE" (you said: "requiring that the
mechanism be followed"), but aside from making the clause pretty
useless, this poses a chicken and egg problem: the original mechanism
is for the Golem to make an announcement, but although a player might
generally be able to cause em to do so through an informal
act-on-behalf agreement, the Golem would have to express its assent
somehow; that requires publishing a message.  (And even if you take
"sending a message" as the "action", which is clearly not the intent
of the clause, doing it by announcement *still* requires faking the
method, because the method for sending a message is (the Golem itself
possibly or someone previously authorized to act on behalf of em)
sending an email, turtles all the way down.)  Or else you're just
turning "can act on behalf of X to perform actions" into "can send
messages on behalf of X to attempt to perform actions", which I think
is not precedented and too vaguely defined.

Someone might argue that without any mechanism for causing a person to
act, everything is "otherwise IMPOSSIBLE" for em to do, which would
cause Golems to be paralyzed, but the interests of the game in
avoiding paralyzation, never mind the R101 rights cited by G. suggest
that we should consider this "POSSIBLE" as "something that's legally
possible, regardless of the situation of the guy on the other end of
the line" rather than "something that's practically possible for the
person to do", just as it's not rules-IMPOSSIBLE for me to take some
morally heinous action just because I could not bring myself to do it,
and wasn't rules-IMPOSSIBLE for Bayes to submit a CFJ even though he
had no code to do that.

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

Appellant Kolja's Arguments:

I support the appeal, I think it's important to get maximum clarity and
consensus on this topic.

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

Appellant G.'s Arguments:

Agreed, it deserves more than myself and omd; I support and do so.

Appeals argument, just to see if it puts the issue as clearly and neutrally
as possible [omd please let me know if this captures it neutrally]:

1.  The rule in question [Rule 2361/1] says "The owner of a Slave Golem
     CAN cause it to take actions that are not otherwise IMPOSSIBLE by
     announcement."

2.  Let's say another rule says "Players can do X without objection".

3.  The caller claims that it should be parsed "[Players CAN do X]
     without objection", meaning that, since "Players CAN do X", the
     owner can cause the golem to do X by the owner announcing that the
     slave does X, ignoring the 'without objection' requirement.

4.  The Judge's arguments state that it should be parsed "[Players
     can do X without objection]", meaning that the owner can set up
     for the golem to do X, but only via the golem using the without
     objection mechanism, because it is "otherwise IMPOSSIBLE" without
     using the without objection method.

5.  The caller counters that, if the judge's arguments were correct,
     that, since many actions are written "Players can do Y by
     announcement", reading it "[Players can do Y by announcement]"
     instead of "[Players can do Y] by announcement" makes it impossible
     for the golem to do anything at all, because golems literally cannot
     announce anything.

6.  The judge counters that R478(Fora) sufficiently strongly implies
     that golems CAN announce things 'via' their owner (i.e. the owner
     is part of the mechanism, an extra step in the email client, in the
     'via a public forum' clause).  The judge states that this is
     (a) generally in keeping in past precedents on what 'acting on
         behalf of' means; and
     (b) at the time this case arose, R101 was protective of the R101
         rights of golems;
     so R101(v) (Participation in the fora) required us to infer the
     ability of golems to send messages to fora and therefore announce
     stuff.

7.  Caller's appeal states that 6(a) and (b) are not enough at all
     to infer that golems can announce anything, so we either have to
     accept the caller's reading of "[Players can do X]" (the caller's
     preferred argument) or accept that golems never could do anything
     at all (rewinding months of gamestate, counter to just about
     anyone's perceptions).

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