On 4/6/26 10:00 AM, Katherina Walshe-Grey via agora-official wrote:
I assign CFJ 4125 to Trigon.
CFJ 4125 was called by snail, and reads: "snail tabled an intent in the
above quoted message."
Judging any CFJ relating to the level of clarity required for performing
actions is one of the greatest lose-lose situations in Agora. The ways
in which this specific clause can be stress-tested are innumerable. No
question can be asked without summoning five more counterexamples.
In this situation, there are interesting arguments for all sides. This
judgement seeks to answer every question put forth in discussions
directly following the calling of this CFJ as well as addressing
relevant passages from prior CFJs with related precedent.
First, though, I want to talk a little bit about the history of
leetspeak, leet, or 1337. In its original implementation among the
hacker groups that originated the term, it was used both as a
shibboleth, as a means to circumvent text filters in text chats, and as
a way to exclude newbies or "n00bs" from conversations that they are not
meant to understand. Over time, these substitutions grew in popularity
among general internet users and were no longer just used by hackers in
limited circles, but leetspeak certainly has a legacy of exclusion.
From EarlyRetirement's original judgement:
Here,
the required elements for both rules are written with digit-for-letter
substitutions, which requires decoding, (decoding only requires mental
effort of some sort beyond what is required to read normal text) before a
reasonable player can understand the text presented.
This is true. The text in question is written in an atypical form of
English orthography. As mentioned above, it is accurately described as a
substitution cipher, albeit not a difficult one.
That some readers
decoded the meaning or even objected is irrelevant because the test is
objective and attaches to the text itself, not anyone's comprehension of
it.
This is partially true. The test is not predicated on whether Janet and
ais523 (two players who sent objections to snail's message) are able to
decode the message. But I disagree that the test is objective; it is
subjective on the part of the paradigm of "a reasonable player," as it
has been called by various contributors to this thread. What this
entails is itself a subjective element, and one that changes over time
as conceptions of Agora, Agoran culture, and even English speaking
culture change.
But this paradigm is not "the average Agoran" as some other contributors
allege. From Janet's arguments on 26 October:
R478 specifically does not say "reasonably" in the definition of "by
announcement". This is why R869's registration mechanism, which *does*
use "reasonably", has a lower standard.
I would like to push back on the "reasonable player" wording as well in
favor of what I will call the "least privileged player." The least
privileged player is the theoretical person who is capable of playing
Agora functionally but who has access to the fewest resources enabling
em to do so. In this case, let the resource in question be English
language comprehension. Then the least privileged player is the person
whose English language comprehension is high enough that e is capable of
understanding and following the Agoran rules as they are written (but
may not be high enough to serve in a real-life public office in an
English speaking country). The standard of "without obfuscation" must be
strict enough to allow for this person to be able to decode any tabled
actions that e comes across.
From snail's arguments on 26 October:
The first requirement is met in that just squinting at the encoded intent
is enough to "get the gist" of which dependent action is being done. If
some letters were replaced with random characters, for instance, it would
be harder to understand, but this case is more similar to a typo, or the
Transposed Letter Effect, which would likely be equally readable barring
additional confounding factors.
The second requirement is also met in that the "decoding" process is very
simple (3 and E look very similar, as do I and 1, A and 4, O and 0) and can
be deduced with the context clues as the replacement is consistent across
each word.
The question is, can we assume that any player who can understand the
rules is also familiar with a small standard set of digit-to-letter
substitutions? The proposition is not absurd; in order to play Agora it
is likely that players must be familiar enough with English-speaking
culture, in particular how it is used on the internet (where leetspeak
and related letter-to-digit ciphers rose to the height of their adoption).
From Automaticat's arguments on 26 October:
I agree with these arguments entirely. I believe that a reasonable agoran coule
read the text as it was intended to be read without even needing to decode - it
just looks like the text.
Let's consider the extent to which the message "just looks like the
text." To many of us, reading the coded message was trivial. The
characters *do* bear a certain resemblance to their unencoded
counterparts, made more obvious by familiarity. However, we have to
consider that commonly substituted letterforms that are familiar to
those Agorans who were able to decode the message may not be as obvious
to players whose first language is not English. I think a critical part
of deciding whether the message is too obfuscated to work lies in
investigating the extent to which these are intelligible to the least
capable English speaker that can play Agora.
3 can be either be reflected or rotated 180° to visually resemble
uppercase E, but also looks like the Cyrillic letter З or з. 0 can
appear as an unfilled oval like O or an oval with a cross within it like
Ø, depending on the viewer's font. This has the potential to confuse
speakers of certain Scandinavian languages. 1 replaces both i and l in
different dialects of leetspeak because it looks similar to both.
snail's message is consistent in only using 1 for i; if it was used for
both we'd have an entirely different problem on our hands. 4 only barely
resembles uppercase A. The horizontal/diagonal angles are different
between the glyphs, and 4's left line does not continue all the way
down; this can also be obscured by fonts that render 4 without the top
lines meeting, further distancing the glyphs from each other.
But you can nitpick any kind of substitution in this way. I'm not clever
for coming up with these attacks on the system. And the fact that the
substitutions *are* being used in an English-speaking environment means
that certain assumptions about the set of available letters to decode to
are more valid than others, as snail pointed out. But I do not think
that we can claim that it "just looks like the text" or that its meaning
is obvious "at a glance" or that we can read them "intuitively."
From juan's arguments on 27 October:
Saying the thing
is in code and so is “obviously” obfuscated is kinda worring. Is
a typo a form of obfuscation? Even so, banning that, it's not far from
saying particular grammatical constructions are “obfuscated”.
Anecdotally, I, as a native English speaker, remember coming across a
sample of leetspeak as a younger teenager and being utterly baffled. I
had to have it explained to me before I could fluently read the text.
And obviously this can be explained to any player unfamiliar with the
concept in less than a minute using a table.
But, then again, the ROT-13 cipher can be explained using a table. If we
set our standard low enough that anything that can be explained like
this is not obfuscated, then we allow ROT-13, a full-on cipher, so we
can't do that. I think the main difference here is that leetspeak
*becomes* intuitive after enough time, just as any jargon will, in a way
that ROT-13 will not.
In fact, Agorans are *prone* to this behavior. It is not unusual for us
to collect common cultural norms (such as Spivak pronouns and terms like
TTttPF) into a glossary that we hand to players upon entry, and this
knowledge will become intuitive over time.
However, this line of reasoning reminded me of another landmark case,
CFJ 3663, where standards for "jargon" are set out. In CFJ 3663, a
player attempted to define in one message the term "reiterate" for
general usage. It was deemed not to have worked because the jargon was
not defined by a legal document, already had a natural language
definition, and was given more or less without context. As Judge G.
pointed out,
In general, Agorans use Jargon a fair amount, and there's nothing wrong
with that. However, each player is not free to adopt Humpty Dumpty's
maxim of choosing words to mean just what *they* want them to mean, that's
chaos rather than communication.
Leetspeak is not strongly associated with Agora or Agoran culture, has a
dubious level of understanding as a form of English, and was not
introduced with very much context to help others understand it.
I've gone back and forth on this one. At the start, I was more inclined
to support the original judgement, then I flipped back and forth
multiple times while writing this document. But I think I've come to a
conclusion.
The original message, written in a regular form of leetspeak, is
obfuscated. Regardless of the fact that the cipher is regular and can be
learned over time by those who don't know it, this specific usage of
leetspeak is obfuscation. The code is presented without context and is
not otherwise given special semantic meaning by other parts of Agoran
culture. Finally, the code is potentially out of reach of certain people
who could otherwise play the game: even though it is teachable, it is
not well tutorialized and it cannot be expected that players know it.
This judgement explicitly DOES NOT ban codes in general, but I opine the
Agoran public's best interests are served by limiting usage of codes in
the same way that we limit usage of jargon.
I judge CFJ 4125 FALSE.
--
Trigon
Popular Polygon and Treasuror Extraordinaire
