I vote conditionally as follows: AFFIRM if the judgement is correct,
otherwise REMAND.
More seriously - I support REMAND but disagree substantially with
ais523’s line of reasoning. Reasoning on appeal seems somewhat
old-hat since we switched to voting on them, but here's mine.
First of all, I'd like to note that Gmail displays the message
differently from ais523's images. I see
{
[arabic text] : I call for judgement on the following statement
}
except right-justified. This leads to the somewhat amusing result
that the Arabic text wins under English ordering rules, and the
English wins under Arabic formatting rules. If anyone is wondering,
this is caused by Gmail adding an explicit <div dir="rtl">; I don't
know why. The algorithm specified by Unicode TR9 makes the paragraph
left-justified, because it looks for the first character in the string
for which directionality can be determined.
Anyway, I'd argue that this doesn't matter, nor is it necessary to
inspect the Unicode to determine the 'correct' ordering. The effect
of the message can be decided on much simpler grounds:
1. The text "I call for judgement on the following statement" should
be interpreted as an action, not implicitly quoted by the Arabic text.
This does not depend on the ordering; it would be true even if the
Arabic version unambiguously came first, as in
{
:أدعو إلى إصدار حكم بشأن البيان التالي
I call for judgement on the following statement
}
[note: I don't believe that this would succeed in calling for
judgement, as there would be no "following" statement, see below.]
Why? Because we’d accept this as performing an action:
{
nkeplwgplxgioyzjvtxjnncsqscvntlbdqromyeyvlhkjgteaqnneqgujjpwcbyfrpueoydjjk
I do an action.
}
And this is so even if the author later explains that "nkep[..]" is a
word in eir secret language that means, say, "I do not do the
following". The same applies to Arabic, for while it is understood by
more people than that one, CFJ 1460 clearly established that messages
must be intelligible without knowing any languages other than English.
Okay, that was a bit unfair, since I dropped the colon. A closer
analogy would be:
{{
nkeplwgplxgioyzjvtxjnncsqscvntlbdqromyeyvlhkjgteaqnneqgujjpwcbyfrpueoydjjk:
I perform an action.
}}
But even though the colon vaguely hints at quotation here, I wouldn't
say it establishes one clearly enough to prevent "I perform an action"
from being effective. This would:
{{
nkeplwgplxgioyzjvtxjnncsqscvntlbdqromyeyvlhkjgteaqnneqgujjpwcbyfrpueoydjjk:
"I perform an action."
}}
So would this:
{{
nkeplwgplxgioyzjvtxjnncsqscvntlbdqromyeyvlhkjgteaqnneqgujjpwcbyfrpueoydjjk:
{{{
I perform an action.
}}}
}}
But not a bare colon.
2. In the original message, "the following statement" refers to the Arabic text.
To conclude this we don’t have to decide that the Arabic text actually
*is* "following", only that it it’s reasonably unambiguous what the
author *means* by "the following statement". Given that there's only
one other piece of text in the message which could be a "statement",
i.e. the Arabic text (G.'s message has no other text), it’s enough if
that text has *some* broadly plausible claim to being "following", so
that the message isn’t completely nonsensical. Does it? It depends
on how email clients display the text. In the image ais523 posted,
where the Arabic text is on the right, it is "following" under English
rules, so in a primarily English context, it clearly has a claim to
being "following". (It's a primarily English context both because
Agora is primarily English and because the action, the 'main' part of
the sentence, is in English.) But what about my version, where it's
on the left? Gmail is a common enough email client that if it renders
the message nonsensical, I'd say it should be treated as ambiguous,
with the historical exception for fixed-width formatting.
Well, in that version the text is right-justified, which suggests RTL,
so even if the reader has never heard of Arabic, anyone with an
understanding of the abstract concept of right-to-left languages could
perceive the Arabic as potentially "following". This demands a bit
more thinking of the reader, but not too much, I think. Again, it
only has to be potentially "following", not certainly.
One might wonder what would happen if the message said "preceding".
In that case, in my version with the Arabic on the left, the Arabic is
"preceding" in an English context, which is good enough. In ais523's
version, though, the reader would have to recognize that the paragraph
could be seen as right-to-left based only on the characters used. By
contrast, I'd say this is just nonsensical:
{{
I call a CFJ on the preceding statement : nkep
}}
even if the author later explains that eir secret language is right-to-left.
But I'd say it is reasonable to expect readers to recognize Arabic as
right-to-left. For one thing, it's arguably reasonable to expect
Agorans to know what Arabic looks like and that it's right-to-left,
just as common knowledge, but that's not necessary; anyone who is
unsure of the script used can just look it up. CFJ 1460 noted that
machine translation is far from perfect, but we can certainly use
online tools/sources to reliably determine that the text is Arabic,
and that Arabic is right-to-left.
What if the message said "following" and there was Arabic on both
sides? Well... again, because of the English context, including the
fact that the 'main' part of the sentence is in English, I'd call it
too ambiguous to be effective.
Note that this all relies on the assumption that CFJ statements don't
have to be grammatical "statements", as opposed to, say, expressions
or questions or nonsense; there's precedent to this effect. If there
were such a requirement, it would be impossible to CFJ on Arabic text
directly (as opposed to inside a quotation within the statement)
because readers could not be expected to verify it as grammatical.
Oh, and I'd say Unicode basically doesn't matter, only visual
appearance in common email clients. The following text:
{
I do X. I do Y.
}
uses the Unicode control characters LRO, RLO, and PDF; a pseudo-HTML
equivalent would be:
{
<rtl><ltr>I do X.</ltr> <ltr>I do Y.</ltr></rtl>
}
In other words, two LTR blocks within a RTL block. Semantically, it's
quite clear that "I do X." comes first, but visually the text should
look identical to
{
I do Y. I do X.
}
It is not reasonable to expect Agorans to inspect the Unicode of every
message they read, so I'd say the fancy text would have the same
effect as its plain lookalike: the message is effective and action Y
is performed first.
