Primordia

Primordia

View Stats:
tsevech May 7, 2014 @ 1:33pm
The puzzle with Oswald and Cornelius is ambiguous
Spoiler alert: I'm referring to the puzzle where Horatius must help Oswald and Cornelius solve their legal issue. Question 1 and 3 are extremely easy, but I fear that Question 2 is broken (or ambiguous at the very least) and should be clarified. Since I know that some of the developers are reading this forum, I hope that this feedback is appreciated.

Basics:
seven robots
each is factor or sturn
each has Q, L, M
# factor > # sturn
all L are factor
no sturn have Q

Assume that "exactly two Factorbuilt robots have the same processor type", we would like to prove that we have exactly four Factorbuilt robots (hence the correct answer from the game: there are three Sturn robots).

However, the following distribution violates the thesis.
Factor: LLLLQQ
Sturn: M

Indeed, notice that the sequence LLLLQQ features "exactly two Factorbuilt robots of the same type", i.e., it contains the sub-sequence QQ. Based on my understanding, the requirement by Oswald is just a constraint on having the sub-sequences LL, QQ or MM in the Factor's distribution, but it seems that developers were meaning something different.

I think that a non-ambiguous formulation in natural English would be: "exactly ONLY two Factorbuilt robots have the same processor type" or something along these lines, which implies that two processor types must agree, while all the others must be different. This allows to conclude that Factor can build at most four robots by a cardinality argument (but Factor must build at least four robots to satisfy the initial constraint #factor > #sturn, hence exactly four robots by Factor is the only possibility).

What do you think? Anyway, great game so far! ;)
Last edited by tsevech; May 7, 2014 @ 1:45pm
< >
Showing 1-15 of 54 comments
Mark Y.  [developer] May 7, 2014 @ 9:27pm 
I believe I have come across this complaint from one other person. It seems fair to assume that two commenters represent some dozens or hundreds of non-commenters. Still, I'm not sure that it's really ambiguous. If you ask a person, using ordinary English, "In the set LLLLQQ, how many items are the same?" I think the answer would be, initially, some confusion as to what "same" meant, but I can't imagine anyone would say that the answer is "two." I think the answer would be either, "The same as what?" or "Six."

I don't think "exactly only two" is "natural English." I've never heard "exactly only" in my life, of which all but about six months and maybe another couple hundred hours have been spent speaking English. (The rest would be speaking Spanish or sundry "guidebook phrase" in foreign languages.) (Googling the phrase points to this meme -- http://www.quickmeme.com/meme/35s4ha -- but using cat memes use deliberately flawed English.)

I guess I can also say that the question is based on a real LSAT question, and I would be surprised if a test for aspiring lawyers wasn't carefully scrutinized by logicians and linguists to make sure that the answer could not be litigated as ambiguous. Since no one likes to bring grievances over ambiguous language like budding lawyers!

Still, given that the phrasing confused a couple players, it probably could've been worded better. Glad you're enjoying the game, and I'm tickled -- and more than a little outclassed -- by the way you've dissected the puzzle. :)
bliss May 7, 2014 @ 11:14pm 
I'd be surprised if "exactly only" doesn't appear in any legal speak such as T&C or EULA :D:

Most, if not all, spoken languages would be ambiguous if you were to interpret them literally without the context of common use. Thank goodness these robots have preserved English to idiom proficiency.
tsevech May 8, 2014 @ 12:05am 
Hi, Mark, thanks for your answer. I'm not a native English speaker, so my proposal "exactly only two items" may be actually bad. Retrospectively, I think that a better sentence than "exactly two Factorbuilt robots have the same type" would simply be "only two Factorbuilt robots have the same type".

This would rule out the guilty sequence LLLLQQ, since it contains four Factorbuilt robots of the same type. It would also rule out LLQQM, which is similarly problematic to solve the puzzle, but admitted by the ambiguous interpretation. What do you think?

That said, I still think that the puzzle you designed is very entertaining! I'm always impressed by the ability of game designers in creating puzzles which are funny and not frustrating.

If I got confused, that may be a side-effect of completing a PhD on formal methods :) I guess that only mathematicians are worse than lawyers when coming to definitions...
wcc May 15, 2014 @ 10:31am 
I can affirm this semantic problem for me and for a friend - but we're both non-native speakers.
Mark Y.  [developer] May 16, 2014 @ 9:09pm 
Huh. I wonder if it's an idiomatic English thing, or if I just picked a poor wording.
noname Jul 7, 2015 @ 12:20pm 
I'm also non native speaker. For me main ambiguity in this test was why is first answer can't be true? One of the conditions is that each of the robots have either of presented processor architectures. So it must be true that one of them has quad-core processor. Am I understanding word "either" wrong? I understood this condition as "each of mentioned processor architecture must be in this group of robots".
Last edited by noname; Jul 7, 2015 @ 12:24pm
Mark Y.  [developer] Jul 8, 2015 @ 9:03pm 
I think you're understanding "either" wrong -- or, if not "wrong," differently from how I understand it. At least to me, "all are either X or Y" means "each member of the set can be X or it can be Y" not "some members of the set are X and some are Y." (In fact, the question posed is "each robot is X, Y, or Z.")

Would you see the same ambiguity in this question?

Five people are in a room. They are all either English or American. There are more Americans than English. Which of the following cannot be true?

(1) There are two English.
(2) There are four Americans.
(3) There are five Americans.
(4) There are two Americans.

(The answer it seems to me must be 4.)

If the question were posed as "Five people, some English and some American, are in a room . . ." then it could be 3 or 4.

It's interesting to me, though, that these questions run cause so much trouble for non-native English speakers. Or, let me rephrase that: I can imagine non-native English speakers getting hopelessly lost in the confusing terminology ("binominal biennial" and so forth) and throwing up their hands in disgust. But I never thought that "either" or "exactly" would be ambiguous!

Notwithstanding the suffering it has caused my dear players, I do appreciate the window it has offered into linguistic challenges.

I'm curious: are any of you guys French speakers? If so, I'd be interested whether you think Flavien's French translation (soon to be released) contains the same ambiguity or resolves it!

De plus, chaque robot est architecturé autour d'un processeur, soit à quatre coeurs, linéaire, ou multiplexé.
Les Factorfab sont plus nombreux que les Sturnweilerfab.
Ceux dotés d'un processeur linéaire ont tous été fabriqués par Factor.
Enfin, Sturnweiler n'a jamais fabriqué de robots dotés de quatre coeurs.

Si exactement deux robots Factorfab sont dotés d'un même type de processeur, laquelle des propositions suivantes est vraie ?
tsevech Jul 9, 2015 @ 11:33am 
Hi, Mark.

I'm not a French speaker, but I'm Italian and my language is not so far from French. Looking at this thread again with fresh eyes, I still find "exactly two" ambiguous for the very same reasons :-) Since "exactement" is just "exactly" in French, I guess the ambiguity is still there. Switching back to English, I still think "only two" is a better wording for this puzzle.

To make the ambiguity I see more concrete, take this sequence of letters: AABBB. If I ask you: "are there exactly two coinciding letters", you may answer me either yes or no. On the one hand, you may notice the sub-sequence "AA", which consists of exactly two coinciding letters; on the other hand, you may observe that the sub-sequence "BBB" features more than two coinciding letters.

But, if I ask you "are there only two coinciding letters", you can only answer no, since the sequence contains three B's which violate the constraint. What do you think?
Mark Y.  [developer] Jul 9, 2015 @ 8:33pm 
Huh, I would think the answer to that quest is definitively "no." It's "no" in a tricky way -- like a question that asks about a "rectangle" but accepts a square as an answer because a square is a subset of rectangles.

One of my favorite things about this is that (1) the question is taken directly from the law school admissions exam and (2) lawyers (like me!) pride themselves on precise language, and yet even still it contains ambiguity. There is ambiguity in all living things, and no matter how much we may try to constrain them, languages -- like Jurassic Park's dinosaurs -- are alive, "and life finds a way."
Last edited by Mark Y.; Jul 12, 2015 @ 9:32am
bliss Jul 12, 2015 @ 4:17am 
After revisiting this, I agree that 'only' would technically be much better than 'exactly' because it would exclude subsets.

I accept that it's literally ambiguous whether 'exactly' is inclusive or exclusive, but I stand by it being unambiguously exclusive in the context of native contemporary English usage. I suggest prideful lawyers avoid depending on that constraint as it's not future proof, and evidently, not culturally robust.

However, one could argued that the current wording is perfect as a question because it tests students contextual judgement of language at the same time as their logic.

It's true "life finds a way", and that's why scientists and engineers use mathematical and programming languages for expressing logic. It's not only that technical languages are easier but that they're actually unavoidable because natural languages lack the capability. While lawyers do a great job of shoehorning logic into natural language, they'll never make themselves redundant because they're also bridging the languages deficiencies by providing common context and reasonable judgement when reading them back.
Mark Y.  [developer] Jul 12, 2015 @ 9:36am 
For what it's worth, "only" has a faint connotation of "no more than" to me, whereas "exactly" more squarely means "no more, no less." (I assume that's why the LSAT uses "exactly" for this purpose.) Though I agree that "only" would also work. :) In short, I surrender!
Last edited by Mark Y.; Jul 12, 2015 @ 9:36am
contrivable Jan 7, 2017 @ 6:37pm 
For me the ambiguity, if any, lies wholly in "have the same type". This has been touched on before. We all seem to agree on what "exactly two" means. Now, "have the same type" in my mind becomes basically:
architecture( robot_1 ) = architecture( robot_2 )
A hidden assumption here is that we only consider cases of
robot_1 != robot_2
for sameness of architecture. Otherwise each robot is the same architecture as itself and the statement "Exactly two [..]" cannot hold true.
Due to "solved it" blindness I find it hard to see any more possible interpretations of how a robot or robots can have the same type of something. Maybe one of you will comment on that despite the time gone by. If so, I'd prefer you stated your interpretation of the relevant set by way of a formula rather than easily arguable natural language. Here's how I understood (and can now not un-understand) the set:
{r1 in factorbuilt_robots | exists r2 in factorbuilt_robots such that ( r1 != r2 and arch(r1) = arch(r2) )}
That covers cases like Factorbuilt: LLLQQ. The set, then, is required to have cardinality 2. As stated by OP that leads to 4 Factorbuilt and 3 Sturnweilerbuilt robots and their architecture distribution.

Now, my problem actually lies with the answers in round 2. When I read them I saw that three of them were correct and immediately attempted the two that I suspected wouldn't be accepted.
Note, the 3 Sturnweiler 'bots together with at least one Factor 'bot are necessarily designed with multiplex processors. Also, at least one Factor 'bot needs to have a quad-core processor. The former verifies "Four of the seven robots have multiplex processors," and the latter "One of the seven robots has a quad-core processor."
I hope LSAT would allow for multiple answers or prepend both "offending" answers with "exactly."

Anyway, nice discussion you got going here.
Mark Y.  [developer] Jan 7, 2017 @ 9:48pm 
You're right. And I'm fairly certain I took the 'structure" of the text unchanged, so I assume "exactly" was missing there. (The LSAT does not accept multiple answers.) Unfortunately, it's impossible for me to remember what the original question was, or to find it, at this point.
Murray Jan 8, 2017 @ 4:19am 
I noticed it being open to interpretation.

Had to work backwards from the answers and assume there even was a right option to understand how the question was supposed to be interpreted.

Saying "exactly two factorbuilt robots have the same processor architecture" didn't mean there couldn't be multiple factorbuilt robots with one of the other architectures. Not to me. Not to anyone who isn't following some lawyer syntax.
Last edited by Murray; Jan 8, 2017 @ 5:11am
contrivable Jan 8, 2017 @ 4:50am 
Originally posted by Murray:
Had to work backwards from the answers and assume there even was a right option to understand how the question was supposed to be interpreted.
For reference, facts, questions and anwers/options:
F1) At the Binomial Biennial, seven robots in the Probabilitists' Circle are gathered for a party.
F2) Each robot is either a Factorbuilt or a Sturnweilerbuilt.
F3) And, moreover, each robot is designed with either a quad-core, a linear-type, or a multiplex processor architecture.
F4) First, there are more Factorbuilt robots than Sturnweilerbuilts.
F5) Second, every linear-type robot was build by Factor.
F6) Finally, Sturnweiler nerver built a robot with a quad-core.
Q1) Which of the following statements about the Probabilitists' Circle cannot possibly be true?

A1-1. Five robots have linear processors.
A1-2. Five robots have quad-core processors.
A1-3. Four robots have multiplex processors.
A1-4. Four robots were built by Sturnweiler.
A1-5. Five robots were built by Factor.

F7) The preceding facts are unchanged.
Q2) If exactly two Factorbuilt robots have the same type of processor architecture, then which of the following must be true?

A2-1. One of the seven robots has a quad-core processor.
A2-2. Two of the seven robots have linear-type processors.
A2-3. Three of the seven robots were built by Sturnweiler.
A2-4. Four of the seven robots have multiplex processors.
A2-5. Five of the seven robots were built by Factor.

F8) The facts are still unchanged.
Q3) Which of the following types of processor architecture might you find in a Probabilitist built by Sturnweiler?

A3-1. Only multiplex.
A3-2. Any of the three types.
A3-3. Only multiplex and linear-type.
A3-4. Only quad-core and multiplex.
A3-5. Only linear-type.

Originally posted by Murray:
Saying there are "exactly two factorbuilt robots with the same architecture type" didn't mean there couldn't be multiple factorbuilt robots with another architecture type.
Agreed, otherwise Factorbuilt Probabilitists couldn't be in the majority.
< >
Showing 1-15 of 54 comments
Per page: 1530 50