Prompt and Anti-Prompt: A Tale of Woe

[DumbJaques]

It's been a while since we've had a good rules change (and it seems we're still struggling to gain full acceptance of the "be lenient on equivalent myth figures" thing), but it's time the quizbowl community came up with some semi-codified guidelines for our ever-evolving organic law. By far the most important one:

The Anti-Prompt. Look guys, its time has come. We need to all hop on board this train, quickly and openly, so we stop boning people out of good, deep knowledge. Especially considering we're all expecting a (much-needed) reeling in of overall difficulty this year, which means that the frequency of perhaps atypical common link questions is probably going to increase. I think the best way to see this point is in case studies, so here you go:

Ex. (transcribed from my unreliable memory of the 2008 NSC finals):

One of these, known as Heaven’s Will, was retrieved by a golden-shelled turtle from the Vietnamese king Le Loi for the dragon king Long Vuong. The souls of twelve berserkers were contained in Skofnung, the one owned by Hrolf Kraki. Amir Arsalan exploited the weakness of Fulad-zereh to one called Shamshir, while Susanowo took one from the body of Orochi called Kusanagi. For 10 points, Sigurd’s Gram, Roland’s Durandal, and Beowulf’s Hrunting are all what type of object, as is King Arthur’s Excalibur?
ANSWER: swords

The illustrious Shantanu Jha, if memory serves, buzzed in on the Shamshir clue and answered "scimitar." Now, Heaven's Well and that weird Norse thing aren't scimitars (though really I'm not sure how you'd prove Heaven's Sword couldn't be characterized that way, it's not like you can get on the phone and call Le Loi's consulate in DC). But this seems to me such an overwhelmingly clear display of precise knowledge (indeed, even more precise than the question strictly requires) that negging it, which is just inconsistent with the spirit of good quizbowl. The point of a prompt is to allow someone who demonstrated clear but incomplete knowledge to expand their answer to include the required information. That shouldn't at all exclude generalizing your statement, you're still adding information to make the answer "more complete."

I'd also like address briefly the argument that answers like "scimitar" have to be wrong, since they don't work for the earlier clues. This just doesn't make logical sense. The whole idea of a common link question (or, indeed, any tossup at all) is that people will buzz on X clue only as a result of not knowing (or sufficiently knowing) the clues preceding X. When you write a 10-line common link on "cattle" in myth, I sincerely doubt you expect more than like one person to get the first clue. How is it fair to punish someone who buzzes in on, say, an Apis Bull clue and says "Bull?" It makes no sense, you're requiring the player to do unrealistic mind-reading and, much worse, you're enticing the person who knows more to buzz in, but making it likely they'll get it wrong. This is in my opinion the single worst outcome of a tossup - it's the very antithesis of what pyramidal quizbowl seeks to do.

Now, in the cattle example, the writer should just be diligent enough to note in the answerline that one should accept either "cow" or "bull." But there are situations where the possible specific answers are a bit more numerous/impractical to list out. And let's face it, writers are not always going to be on the ball with this, even good ones - sometimes you just overlook something that only comes out when people actually play a tossup. If the answer was "mare" and somebody said "horse," but there was no prompt, even the least cavalier of competent readers wouldn't hesitate to prompt. I've seen a few situations this year already where a reader encountered an antiprompt situation, but just didn't know how to antiprompt/didn't think it was established convention. And, well, it isn't. So let's fix that.

Proposal: "Anti-prompt" shall enter the quizbowl vernacular as merely another form of prompting, to be used as freely and naturally. I'd encourage NAQT and ACF to make official provisions for this, but in the meantime I'd like to get just a community consensus here in this thread, to help get the word out. Specific terminology that seems appropriate would be "anti-prompt" or, to quote Matt Weiner, "can you be less specific?" Other suggestions are welcome.

I intend to include anti-prompts in all my questions from this point foward, and encourage all writers to do the same; I also encourage readers to consider using the antiprompt in appropriate situations, just as they would prompt if an equally obvious classical prompt situation arose but the answerline was lacking.

Of course, I guess people could disagree, so this thread seems the natural platform to do that as well.

[Cheynem]

I'd agree with Chris. You have to be careful so that the anti-prompt doesn't just reward people saying things based on incomplete knowledge, but in cases where people directly say what a question refers to, an anti prompt is fine in my opinion.

[Down and out in Quintana Roo]

I feel like i like this idea but i'm confused as to when it would be implemented.

For example, this was a question in practice today, from HSAPQ set #16:

One of these can be issued by the Supreme Court to obtain immediate review of exceptional cases
and requires agreement from four of the justices. Another of these orders commands an official to
perform a specific act required of his office. In addition to (*) certiorari and mandamus, another of these
can be issued to direct one who holds another in custody to produce that person before the court. For 10
points, name these common law orders requiring the performance of a specific act, an example of which is
habeas corpus.
ANSWER: writs

A player buzzed in right at "required" and said "writ of mandamus." Should she have been de-prompted or asked to give a less specific answer?

[Mike Bentley]

It's been a while since we've had a good rules change (and it seems we're still struggling to gain full acceptance of the "be lenient on equivalent myth figures" thing), but it's time the quizbowl community came up with some semi-codified guidelines for our ever-evolving organic law. By far the most important one:

The Anti-Prompt. Look guys, its time has come. We need to all hop on board this train, quickly and openly, so we stop boning people out of good, deep knowledge. Especially considering we're all expecting a (much-needed) reeling in of overall difficulty this year, which means that the frequency of perhaps atypical common link questions is probably going to increase. I think the best way to see this point is in case studies, so here you go:

Ex. (transcribed from my unreliable memory of the 2008 NSC finals):

One of these, known as Heaven’s Will, was retrieved by a golden-shelled turtle from the Vietnamese king Le Loi for the dragon king Long Vuong. The souls of twelve berserkers were contained in Skofnung, the one owned by Hrolf Kraki. Amir Arsalan exploited the weakness of Fulad-zereh to one called Shamshir, while Susanowo took one from the body of Orochi called Kusanagi. For 10 points, Sigurd’s Gram, Roland’s Durandal, and Beowulf’s Hrunting are all what type of object, as is King Arthur’s Excalibur?
ANSWER: swords

The illustrious Shantanu Jha, if memory serves, buzzed in on the Shamshir clue and answered "scimitar." Now, Heaven's Well and that weird Norse thing aren't scimitars (though really I'm not sure how you'd prove Heaven's Sword couldn't be characterized that way, it's not like you can get on the phone and call Le Loi's consulate in DC). But this seems to me such an overwhelmingly clear display of precise knowledge (indeed, even more precise than the question strictly requires) that negging it, which is just inconsistent with the spirit of good quizbowl. The point of a prompt is to allow someone who demonstrated clear but incomplete knowledge to expand their answer to include the required information. That shouldn't at all exclude generalizing your statement, you're still adding information to make the answer "more complete."

I'd also like address briefly the argument that answers like "scimitar" have to be wrong, since they don't work for the earlier clues. This just doesn't make logical sense. The whole idea of a common link question (or, indeed, any tossup at all) is that people will buzz on X clue only as a result of not knowing (or sufficiently knowing) the clues preceding X. When you write a 10-line common link on "cattle" in myth, I sincerely doubt you expect more than like one person to get the first clue. How is it fair to punish someone who buzzes in on, say, an Apis Bull clue and says "Bull?" It makes no sense, you're requiring the player to do unrealistic mind-reading and, much worse, you're enticing the person who knows more to buzz in, but making it likely they'll get it wrong. This is in my opinion the single worst outcome of a tossup - it's the very antithesis of what pyramidal quizbowl seeks to do.

Now, in the cattle example, the writer should just be diligent enough to note in the answerline that one should accept either "cow" or "bull." But there are situations where the possible specific answers are a bit more numerous/impractical to list out. And let's face it, writers are not always going to be on the ball with this, even good ones - sometimes you just overlook something that only comes out when people actually play a tossup. If the answer was "mare" and somebody said "horse," but there was no prompt, even the least cavalier of competent readers wouldn't hesitate to prompt. I've seen a few situations this year already where a reader encountered an antiprompt situation, but just didn't know how to antiprompt/didn't think it was established convention. And, well, it isn't. So let's fix that.

Proposal: "Anti-prompt" shall enter the quizbowl vernacular as merely another form of prompting, to be used as freely and naturally. I'd encourage NAQT and ACF to make official provisions for this, but in the meantime I'd like to get just a community consensus here in this thread, to help get the word out. Specific terminology that seems appropriate would be "anti-prompt" or, to quote Matt Weiner, "can you be less specific?" Other suggestions are welcome.

I intend to include anti-prompts in all my questions from this point foward, and encourage all writers to do the same; I also encourage readers to consider using the antiprompt in appropriate situations, just as they would prompt if an equally obvious classical prompt situation arose but the answerline was lacking.

Of course, I guess people could disagree, so this thread seems the natural platform to do that as well.

I'm pretty much 100% in favor of this. Without getting into specifics, getting negged on a too specific answer happened multiple times at our MOO mirror last weekend and was very frustrating. I hate when this happens to me and I hate as a moderator having to give teams -5 for this.

[Windows ME]

What if Shantanu buzzed in on a different clue from the one he buzzed in on and said "scimitar"?

...and it wasn't a scimitar? How can a moderator hope to tell this information?

EDIT (Don't want to double post): And what if he needed to think o the scimitar clue and it took him a good three quarters of the way into the next clue before he could buzz in?

To use a very simple example, if the common link is in "wars England participated in" and you buzz in with "Thirty Years War" because it took you 2 lines to recognize a clue in the lead-in, but the question is no longer talking about the Thirty Years War, should you be reverse-prompted?

[dtaylor4]

What if Shantanu buzzed in on a different clue from the one he buzzed in on and said "scimitar"?

...and it wasn't a scimitar? How can a moderator hope to tell this information?

EDIT (Don't want to double post): And what if he needed to think o the scimitar clue and it took him a good three quarters of the way into the next clue before he could buzz in?

To use a very simple example, if the common link is in "wars England participated in" and you buzz in with "Thirty Years War" because it took you 2 lines to recognize a clue in the lead-in, but the question is no longer talking about the Thirty Years War, should you be reverse-prompted?

In general, I think any specific answer that could fit one of the clues should be reverse-prompted. I think the player should get benefit of the doubt as to the particular clue being buzzed on.

[Kyle]

It seems to me that NAQT's policy of obsessively listing acceptable alternate answers on common-link tossups is already a step in the direction toward what Chris is suggesting. There is one tossup in particular that I have in mind as an example, but it's from a current set and I can't mention it here. I poked around to try to find an older tossup to illustrate my point and I came across a "short" question (#271005) from 2010-2011 Quizbusters E, which I hope I am allowed to post here?

Which of the {classical elements} names Long Island's longest {barrier island}, an object built in a Jack London title, and consists of rapid {combustion}?

answer: _fire_ (accept _Fire Island_ or _To Build a Fire_)

The idea, basically, is that the specific answer corresponding to any single clue in a common-link tossup is a correct answer to the tossup as a whole. So it seems like NAQT's policy -– at least in theory -- already allows for what Chris is suggesting. Or am I misinterpreting something?

[theMoMA]

It seems to me that this thread might go by the better name of "The Modern Quizbowl Answer Line: An Exercise in Insufficiency."

[Important Bird Area]

question (#271005) from 2010-2011 Quizbusters E, which I hope I am allowed to post here?

That's fine.

In general, NAQT's recent policy on common link tossups has been to list all of the components as separately acceptable. (There was some controversy about inconsistent application of this in the HSNCT staff meeting a few years ago, so we've really tried to make it better.)

The fire tossup, though, is a bit different because all plausible answers already contain the word "fire." The kind of anti-prompt Chris is talking about is considerably rarer, and I don't know what NAQT's policy is. (It may very well depend on the phrasing of individual questions.)

[magin]

I think that a lot of this could be avoided by making sure to list alternate answers; if the clue refers to a scimitar and the answer is "sword," making sure to say "accept scimitar" on that clue is pretty important. I also agree with Chris that if a player gives an answer which is correct for a clue, it's not reasonable to require him or her to know the earlier clues; it's the responsibility of the player to know clues and buzz on them, not to read the question's mind.

After all, if no one buzzes on the leadins (which happens all the time), not accepting a correct answer like "scimitar" because those earlier clues referred to swords in effect requires the player to know those earlier clues as well as the clue he or she buzzed off, which seems silly to me.

[Kyle]

The fire tossup, though, is a bit different because all plausible answers already contain the word "fire." The kind of anti-prompt Chris is talking about is considerably rarer, and I don't know what NAQT's policy is. (It may very well depend on the phrasing of individual questions.)

Yeah, but isn't it the same principle? The idea behind NAQT's answer line is that, if "To Build a Fire" is a correct answer for the third clue only, then it's a correct answer to the tossup as a whole. Chris is saying that, if "scimitar" is a correct answer for the third clue only, then it's a correct answer to the tossup as a whole.

[vcuEvan]

I think that a lot of this could be avoided by making sure to list alternate answers; if the clue refers to a scimitar and the answer is "sword," making sure to say "accept scimitar" on that clue is pretty important. I also agree with Chris that if a player gives an answer which is correct for a clue, it's not reasonable to require him or her to know the earlier clues; it's the responsibility of the player to know clues and buzz on them, not to read the question's mind.

After all, if no one buzzes on the leadins (which happens all the time), not accepting a correct answer like "scimitar" because those earlier clues referred to swords in effect requires the player to know those earlier clues as well as the clue he or she buzzed off, which seems silly to me.

I heartily agree with the sentiment of Chris' post, but I think the solution proposed here is better.

[Mike Bentley]

I think that a lot of this could be avoided by making sure to list alternate answers; if the clue refers to a scimitar and the answer is "sword," making sure to say "accept scimitar" on that clue is pretty important. I also agree with Chris that if a player gives an answer which is correct for a clue, it's not reasonable to require him or her to know the earlier clues; it's the responsibility of the player to know clues and buzz on them, not to read the question's mind.

After all, if no one buzzes on the leadins (which happens all the time), not accepting a correct answer like "scimitar" because those earlier clues referred to swords in effect requires the player to know those earlier clues as well as the clue he or she buzzed off, which seems silly to me.

I heartily agree with the sentiment of Chris' post, but I think the solution proposed here is better.

By "accept scimitar on that clue", do you mean that "scimitar" should only be acceptable while that clue is being read and at no other point in the question? There are times where I'll delay a buzz on an earlier clue because I can't remember it, for instance.

[vcuEvan]

I think that a lot of this could be avoided by making sure to list alternate answers; if the clue refers to a scimitar and the answer is "sword," making sure to say "accept scimitar" on that clue is pretty important. I also agree with Chris that if a player gives an answer which is correct for a clue, it's not reasonable to require him or her to know the earlier clues; it's the responsibility of the player to know clues and buzz on them, not to read the question's mind.

After all, if no one buzzes on the leadins (which happens all the time), not accepting a correct answer like "scimitar" because those earlier clues referred to swords in effect requires the player to know those earlier clues as well as the clue he or she buzzed off, which seems silly to me.

I heartily agree with the sentiment of Chris' post, but I think the solution proposed here is better.

By "accept scimitar on that clue", do you mean that "scimitar" should only be acceptable while that clue is being read and at no other point in the question? There are times where I'll delay a buzz on an earlier clue because I can't remember it, for instance.

Scimitar should just be accepted. I'd rather be overinclusive and give someone points for thinking excalibur is a scimitar than be underinclusive and punish someone for thinking the scimitar is a scimitar.

[cvdwightw]

I think that potentially the best way out of this is to use the rarely-seen "accept [x] after [y]" construction. For instance, this example would read:

ANSWER: swords [accept scimitars after "Arsalan"]

This gets around every problem that has cropped up in this thread, and around a couple of other ones - e.g., what if a player buzzes in with "scimitar" off the lead-in, and what if a player buzzes in with something like "katana" that is a type of sword but doesn't match any of the clues?

[theMoMA]

Yeah, I think people have pinpointed the more important issue in the later posts of this thread. Quizbowl's answer lines need to do a much better job anticipating possible answers and telling moderators what to do when those answers are given.

[DumbJaques]

Well, sure, the best solution would be for people to write ideal answerlines (and if you're writing a common link tossup, you should be extra diligent about this). But the reality is that people won't always do that, and even good writers can overlook some possible buzzes. I would feel terrible as a writer if I'd neglected such a potential answer, and someone was negged by a well-meaning moderator for something that everyone knew was basically correct.

We don't hesitate to prompt in obvious cases even if the answer line doesn't explicitly say to do so, so I don't think there should be any difference in how we treat antiprompts. The only difference is the "it doesn't specifically apply to the earlier clues" thing, which most people seem to agree runs counter to how pyramidal quizbowl is designed to work.

I'd love it if we could trust every writer to be all over this stuff, but come on. . . everyone here plays quizbowl. We know that's not how it's always gonna play out. So why not extend the general discretion of the prompt prerogative to encompass the antiprompt?

[grapesmoker]

Honestly, in that situation, I would have just accepted "scimitar" outright. But I think Chris' idea is a worthwhile one.

[Mike Bentley]

I would be in favor of a rules change rather than just placing the onus on question writers.

[theMoMA]

I would be in favor of a rules change rather than just placing the onus on question writers.

I disagree. I think that questions should specify where people can prompt (or anti-prompt, as it were). If there's a question on swords and one of the clues is about a scimitar, I'm fine with the question saying "prompt on scimitar [on that particular clue]." I'm not fine with saying that all more general answers should be anti-promptable always; this shades way too overinclusive for me. Perhaps there could be an anti-prompt protest rule (i.e. the clue I buzzed on applied to buckyballs but the answer was fullerenes, I should have been anti-prompted), but I don't think that the general rule should be "anti-prompt at will," because I think that the general presumption should be that more specific answers are wrong unless they're verifiably right.

The onus continues to be on question writers to write answer lines that eliminate these problems, not moderators to know the panoply of correct anti-promptable answers in existence (obviously, a moderator anti-prompting based on specific knowledge and discretion is a different story).

[Skepticism and Animal Feed]

I challenge proponents of the anti-prompt rule to draft a rigorous statement of the rule, as they would like it to appear in the ACF rules or wherever else. My suspicion is that it would be difficult to draft such a rule that isn't overly complicated and prone to literally allowing people to earn points in questionable ways.

I'd rather settle the issue on a case-by-case basis; if people truly demonstrated real knowledge and suffered an injustice by not being reverse-prompted, they can protest and it can be resolved that way. Yes, this means that if it ends up not mattering, sometimes your PPG might be 1.5 points lower than it otherwise ought to be. The world isn't perfect.

[Sima Guang Hater]

Honestly, in that situation, I would have just accepted "scimitar" outright.

This. Can't we just make a general statement to moderators and question writers along the lines of "be generous" instead of trying to codify this down to the absolute last letter?

[DumbJaques]

Yeah, what I'm advocating for is sort of this sequence of reforms, in order of preference:

1) People write tossups with the principles of the Anti-prompt in mind (and note in answerlines to accept or at least anti-prompt on various specific answers, "be generous" with specific answers, etc.)

2) We establish a sort of general consensus that moderators should use the same common sense in anti-prompting or accepting that they would use in prompt situations, where there was similarly no explicit answer line notation provided but nobody in the world would question the prompt.
Basically, the idea here is to make "anti-prompt" as mainstream as "prompt," so if someone buzzes in on a clue about the Siege of Syracuse in a tossup on "this effort" but the answer is "Sicilian Expedition," a mod who can see the evident relationship from the info in the tossup can be free to dish out an anti-prompt without feeling like they're being cavalier about the rules or anything.

3) We accept legitimate anti-prompt protest grounds, so if it's at the point where this kind of stuff is deciding a game the player has some recourse to take the matter to the TD and everything. I assume if it gets to this point there was a serious flaw with the construction of the question, but as is I don't know that there'd be an established standard for accepting anti-prompt protests. I'd surely accept reasonable ones, but getting something of a consensus about adding this to our collective body of "shit you can protest" seems like it would give people a last resort against any particularly dire situations.

Again, making sure we're all conscientious about answerlines is the most basic (and ideal) step to fix this problem, but I see nothing but positive results from establishing a precedent of leeway on this front. I mean, consider the only real drawback, that a moderator is over-generous with anti-prompting (as can easily happen in the natural course of the game with regular prompting, see 2010 Nats Finals). This is EASY to fix on protest, because the protest resolution is just that the player should have been negged, and a replacement tossup can be read to the other team only if needed.

Conversely, consider the situation in which a player wasn't anti-prompted, protests, and everyone readily agrees they should have been. So now, even if the anti-prompt would have almost assuredly led to a correct answer, the whole thing has to be yanked and a live replacement tossup read, which blows as the person basically had the original answer right. I think this general culture change (and that's really what I'm advocating, as opposed to a hard and fast rule, which I admit would be hard to explicitly formalize given how various common link questions can be) is a clear moderate benefit-zero cost situation, which is why I'd like to see it become a part of mainstream quizbowl.

[jonpin]

Honestly, in that situation, I would have just accepted "scimitar" outright.

This. Can't we just make a general statement to moderators and question writers along the lines of "be generous" instead of trying to codify this down to the absolute last letter?

As someone who currently attends quiz bowl matches solely as a moderator, I don't think this is a good idea. Saying "Be generous" and leaving it at that would go poorly for me and absolutely awfully for a lot of the people who tend to moderate, especially at the high school level, simply because of the "katana" example. I don't know mythology, and in the heat of the moment, I wouldn't want to have to make a split-second decision "Is anything in this question a katana?" keeping in mind that a moderator should ideally have a response within a second and not spend 5 seconds looking at the packet with a quizzical look on his face. The current approach I frequently see in NAQT (which would have specifically listed "accept 'writ of marandus' " on that example listed above) works perfectly fine, which I show here with an example I recently saw in last year's BHSAT:

4. This adjective partly names Mirabeau Lamar’s worthless currency in the Republic of Texas and a group which massacred Fort Mims, but lost at Horseshoe Bend to Andrew Jackson; those Creek Indians had this type of “sticks.” Métis leader Louis Riel rebelled near Canada’s river of this name. One period named for this color included the Palmer raids; in another, Joseph Welch asked “Have you no decency?” to a Wisconsin senator, a member of the House Un-American Activities Committe. For 10 points, give this color against which Joe McCarthy led an anti-Communist “scare”.
ANSWER: red [accept redbacks, Red sticks, etc. while the clue referring to that specific thing is being read]

The one thing, though, I'd specify, rather than leave it to "etc."

I am completely against the idea of a moderator ever saying "anti-prompt", especially when cardinal rules currently say that when a moderator "prompt"s, he is not to give any more idea as to the clarification being sought. With less experienced players, saying "anti-prompt" is just going to confuse the hell out of them.

[No Rules Westbrook]

I vote yes on Chris's series of principles above. The time for instant replay has come...er, or something like that.

I regret giving a kneejerk reaction of "no" to Magin's Carl Rogers protest at Chicago Open, because I was too schooled in the traditional bias against anti-prompt.

Obviously, writers should be cognizant and try to help as best they can. But people should be aware that these rules place a lot of responsibility on moderators - many of whom have no idea about the subject matter of the question. I think a "be generous" injunction to most moderators is the best we can do - something akin to how the fumble is handled in the NFL - play it conservatively and prompt a lot, then let people object later, rather than just saying no right off the bat. Create grounds for protest. Because most moderators just don't have the qb background to be making logical decisions about anti-prompting.