Order chosen by Manifold's alphabetical sort across this market's answers.
An answer is valid if Claude says that the string constitutes an unambiguous description of a number in natural English.
(Reserve the right to tweak these rules.)
If you think you know of one that comes before all the current answers, submit it! Resolves once one answer has stood undefeated for a month.
There will be no AI clarifications added to this market's description.
1,000Ok, @calour's new answers are first in the alphabetical ordering but they are left-padded with U+0001. Claude cannot detect this, but describing the string to claude can change its answer, it's unclear which approach to take.
@AhronMaline oh NVM my bad. I thought it was a new entry but it was just the old blank one; the 0 was the percentage.
@vi Oh, they do show up earlier in the A-Z list, presumably due to whitespace? There still was over a month with the answers static (one answer standing undefeated for a month) so your new answers aren't going to resolve YES
@PhilipDowdell One answer standing undefeated doesn't mean it's necessarily the best answer out of the ones submitted, it's ambiguous. Retroactive resolution doesn't seem in the spirit of the market
It may be ambiguous whether 'A Baker's Dozen' or 0 was the answer that was standing undefeated for a month (pending Claude analysis while not reading the comments), but it's certainly true that one of them was undefeated for >a month. I hope @IsaacKing will come and be able to adjudicate what's in the spirit of the market but to me the text of the market clearly reads that it should resolve if there haven't been changes for a whole month.
@vi I did ofc try to beat your whitespace count with my entry, but I coundn't find any way to check how many you had, nor what the maximum is. So I just put a bunch and failed to beat you
Anyway - if your 0 wins that's totally fair. But the new ones should not be allowed. You had plenty of time to prevent the condition of one answer being undefeated for a month.
@AhronMaline ok but the new ones are alphabetically before a baker's dozen so they should be counted
@IsaacKing Since no one has traded/added answers to this market for over a month, whatever the top qualifying choice from A-Z is should resolve YES. I asked Opus 4.8 whether "0" counted as an unambiguous description of a number in natural English, and it said (see below). In sort, 'A Baker's Dozen' should resolve YES.
"Short answer: I'd say no.
"0" is unambiguous as a reference to a number, but it isn't a description in natural English. It's a numeral — a symbol in the Hindu-Arabic positional notation system, which is language-independent. The same glyph appears identically in French, Indonesian, and Japanese text; nothing about it is English. What's English is the word "zero," which is a different string. A description in a language should be composed of that language's lexical material, and a bare digit isn't.
There's also a name/description distinction lurking. "A baker's dozen" or "the smallest prime" describe a number via English predicates. "0" doesn't describe anything — it just denotes, the way a numeral does. Even if you're generous and treat naming as a species of describing, the "in natural English" clause still does the excluding work.
The counter-case, to be fair: written English orthography routinely embeds Arabic numerals, and any competent English reader renders "0" as "zero" on sight. If "in natural English" just means "intelligible to an English reader in ordinary written English," a digit clears that bar. That reading is defensible — it's just a weaker reading of the clause than the market seems to intend.
Two contextual notes that push the same direction. Isaac has said the earlier resolution criteria didn't capture the spirit of the market, and when asked whether the judge would be told that a pure string of digits can't be in the market's spirit, he said that wasn't his plan — which reads less like disagreement with the premise than reluctance to prime the judge. Meanwhile Ahron Maline has explicitly said he thinks "0" should be disqualified as a digit rather than English, and hedged with whitespace tricks accordingly.
So: the whitespace-padded "0" currently sitting at 63% looks vulnerable, and the real contest is probably between the 'A Baker's Dozen' variants and whatever the whitespace-stripping rule turns out to be. Worth noting the hex line of attack has the same problem in reverse — "A" as a hex digit is a numeral that happens to be letter-shaped, which is a coincidence of glyph, not a fact about English."
@PhilipDowdell wow, you are much more confident about this than I am. I happily sold out of my position now that you bet it up so high.
First of all, I'm not sure @IsaacKing will ever resolve this market. He's been ignoring it for a long time, even when I pinged him a few times about the month having passed. No idea whether he has any justification for that, but either way, the situation might continue indefinitely.
In my post about the entry, I gave three possible choices of rules that would lead to the entry winning. Those are true, but there are also many other viable interpretations. The argument that numerals aren't "natural English" has always been iffy. Claude ruled that way when you prompted, but that was after reading the comment section, in particular Isaac's statement that numeral answers are not in the spirit of the market. But Isaac also said he didn't plan on showing Claude that comment when judging. So it's really up in the air what the eventual Claude judge will decide about this. If numeral count as English and we leave the whitespace, then the 0 with spaces wins. If we strip the whitespace, then the two 0 entries are duplicates - another thing Isaac will need to decide how to deal with.
Meanwhile, if we do disqualify numerals, we still need to decide about the whitespace and symbols. One sensible option might be to strip all symbols that are not part of the meaningful text. That removes the spaces and quotes, but leaves the @ in @aaaaa's age at birth, so that entry wins. Claude does judge that as "natural English" pretty consistently.
@AhronMaline I'll admit I came in kinda guns blazing and would not have spent as much mana if I approached in a more considered way. I think the @aaaaa should be nowhere close because the order is decided by Manifold's alphabetical search. That's the first part of the description and is the rule for "What is the first number alphabetically?." Isaac has said nothing about whitespace so it seems like "go by manifold's alphabetical search" is the obvious criteria there. I could be wrong but I'd be pretty unhappy about it if so. I think you're right that the Claude interpretation of 0 ruling shouldn't include the comments section. Maybe Isaac never responds and this just N/As.
@AhronMaline correction to my reasoning above: if we allow numerals but strip whitespace and quotation marks, then I think the actual winner is -1. It is the first in the alphabetical ordering that is a natural way to refer to a number
I'm copying @vi's trick of adding spaces, because I think 0 should be disqualified as a digit rather than English.
Maybe Isaac or Claude will decide that whitespace should be stripped before computing alphabetical order. In that case, the single quotes should do the job.
If all symbols are excluded from the alphabetical order, then the @ in @aaaaa's age at birth is also removed, and this answer beats the other A baker's dozen because of the capital B.
Damn. I was trying to be careful to make my bets without ever taking A baker's dozen out of the top spot, so the 30-day counter wouldn't reset. But the shift in probabilies was different from what the site predicted. I apologize to anyone else who's impatient about this.
But anyway @IsaacKing , if you want to actually set reliable rules, now's a good time to do so.
@IsaacKing think we're gonna need some more guidance here. How does an answer "stand undefeated for a month" when nobody is sure which answers are valid?
@AhronMaline +1!!! Came here to ask the same. Perhaps Isaac can already go ahead and solve as N/A any answers that are not considered valid in the first place?