Why are there negs in QB?

[cwasims]

After some occasional and maybe not especially rigorous thought about the gameplay purpose of negs in QB, I will confess that I’m not really convinced of the overall benefit of having them in the game. I think the gameplay benefit is very situational and that they usually end up penalizing actions that are pretty unobjectionable from a gameplay perspective.

I think the only principled reason to have negs is to prevent the following situation: team A knows that team B is much better than they are at some category X, and so when team A realizes a tossup is in category X, they buzz in early with a random guess. Since this random guess is very likely to be wrong, team B will then be able to wait until the end of the question to answer, and (in a format without powers) almost certainly maximizes their expected points by doing so. However, the players on team B who are good at category X enjoy getting good buzzes on questions in that category, and it is frustrating for them that they cannot do so because their opponents will just randomly guess before they have the chance to hear a clue that they know.

Instead, with negs, the expected value of randomly guessing is (most likely) somewhat negative instead of slightly positive, so people will refrain from randomly guessing in most cases. As a result, team B will get the opportunity to get more good buzzes in category X, and team A will not really be worse-off since their chances of randomly guessing the correct answer were extremely low anyways.

Although this sort of argument is certainly not a slam dunk (I could easily see someone not seeing any reason to discourage random guessing specifically), I think most QB players would say it is probably a good thing from the perspective of keeping players interested and excited about playing the game that those players can in fact get good buzzes in categories they know. The problem, though, is that I think very few negs fall into this category of randomly guessing because of your knowledge of your opponents’ strengths. Certainly before the advent of detailed stats there were few ways to gain much knowledge about your opponents’ strengths (especially at the Nationals level) and yet negs were alive and well.

Instead, I would conjecture that the vast majority of negs are educated guesses of some variety, and it seems strange that we should want to penalize people for doing this above and beyond the (much larger) opportunity cost of incorrectly interrupting. It also seems strange that incorrect educated guesses should end up being penalized in the middle of a question but not at the end. As a tool for disincentivizing random guessing, negs seem like a pretty poor solution from my point of view: they serve their purpose, but at the cost of penalizing a much more common type of (in my mind) unobjectionable action.

I can see a few other minor benefits of negs – it is somewhat useful to track which players incorrectly interrupt more than others and negs can make the cost of incorrectly interrupting more salient to new players. It does not really make much sense to modify gameplay to serve either of these ends though, since incorrect interruptions could still be recorded somehow if there was an interest in doing so and it hopefully isn’t too problematic to explain the opportunity cost of incorrectly interrupting to a newcomer once they’re more serious about QB.

TL;DR (and apologies for the probably excessive length) – although negs penalize random guessing, they also penalize educated guessing, which seems bad to me, and I can’t think of any other gameplay reason why you would want to have negs.

[ninjernacci]

I've always felt that the purpose of negs was, like you said, to punish random guessing with more than just the opportunity of cost of giving the other team a clean chance to hear the entire question. Although, I will disagree that punishing educated guessing is bad. While we don't necessarily want to punish educated guessing, we do want to disincentivize educated guessing over actually knowing the clues being asked about. For instance, we want someone to be rewarded for knowing a lot about, say Salvador Allende, than we want someone to be rewarded for hearing stuff that sounds like a South American socialist and lateral thinking their way to the right answer using their knowledge about the difficulty level of the set and what may or may not come up in the packet being played.

[Skepticism and Animal Feed]

I played without negs in high school and with negs in college. I can't say that the presence or lack of negs really changed my thinking or approach to the game. The other team getting to hear the entire tossup and buzzing in at the end was such a bad outcome that I was sufficiently disincentivized from guessing randomly even without negs. The 5 points is paltry compared to this.

I suspect the only meaningful impact of a negs is that they allow more comebacks to happen by making it possible for a team to actually lose points. Having once been eliminated from a tournament where my team was in a "if we don't neg we make it to the finals" situation after my teammate took an exceptionally stupid neg, it was a very unpleasant experience for me, but very exciting for the half dozen or so observers. Of course making things interesting for observers is pointless in modern quizbowl, where at best somebody's B team is watching if they happen to be on a bye that round, but it makes sense if you recall that this all started as a TV show.

[btressler]

I believe the historical context of the neg comes from College Bowl. They wanted to discourage early buzzing so the audience heard almost all of the question. And those tossups were at most 3 sentences, and often less.

Perhaps someone else with better knowledge can support or refute this.

As a coach, I liked that PACE didn't have them.

[AKKOLADE]

One argument off the top of my head is that at an event like PACE, where the field is so deep, the penalty for missing a question early is that your opponent will almost assuredly pick it up by the end and get the advantage on the bonus.

I suppose another argument coming off of that is that at a tournament with a field with bigger disparities in talent, missing a question isn't guaranteed to lead to your opponent getting the tossup, and a five point penalty would act as a further penalty for over-aggressive play.

This is all before my caffeine kicks in, so I'm not sure how good of an argument it is.

[Cheynem]

I don't really particularly care if there are negs or no negs in a bonus format--the penalty of missed bonus points, as some people have alluded to, looms larger anyway.

I do think in an all tossup format (or at the very least in a shootout format), negs are obviously more significant and I think are utilized to prevent wild mass guessing in the shootout format or keep the aspect of risk-reward that you get in the bonus format.

One idea that might be interesting is--in tournaments with powers, what if during the in-power portion you could get a neg for an incorrect buzz, but once you moved out of power, there would be no neg for an incorrect buzz? This would still try to discourage risky guesses, but would perhaps allow for more later, educated guesses, as well as avoid unfortunate scenarios in which someone buzzed with literally only a minor word left in the question and got negged.

[Gene Harrogate]

I agree with your post Chris. Derek So recently brought up another reason negs might be a bad gameplay feature: gender bias in punishments for guessing. Research has shown that the SAT's wrong answer penalty disproportionately affects women, as they are on average less willing to put down a guess than men even though test-takers can usually eliminate enough answers to make guessing worth it. It could be that negs in quizbowl have a similar differential effect, while adding very little to gameplay. Anecdotally it seems to me female players tend to play more conservatively on tossups than men, or focus their contributions more on bonuses. I think it's pretty well established that the quizbowl environment can make women feel unwelcome, and punishing someone who already feels disproportionately judged or out of place for trying to play the game seems like an undesirable mechanism.

[ThisIsMyUsername]

I think Chris's original post (directly and indirectly) raises multiple questions about negs. Included among these are: (1) As a matter of abstract principle, is guessing undesirable enough to merit punishment? (2) How do negs factor into strategically correct gameplay? (3) What is the psychological effect of negs upon people who are not thinking strategically? My thoughts on this will be a little scattered, I'm afraid. I may have more to contribute later.

(1) The first of these questions is unanswerable without making broader assumptions about the purpose of this game, about what it exists to reward. There are those on one extreme, who think the game should be about as close to a pure test of knowledge as is possible given some kind of buzzer format. These are the sorts of people who are liable to call most forms of lateral thinking "fraud," even when that lateral thinking depends on quite a bit of background knowledge. In this school of thought, lateral thinking is not something that quizbowl should actively reward, and is maybe something it should actively discourage, as it might disadvantage the person who has more direct knowledge of a tossup's answerline. Then there are those on the other side who view quizbowl as first and foremost a game, which should be as much about strategy as it is about unadulterated knowledge. These are the people who are liable to write questions that give "context" clues, and who complain about questions in which the clues seem like detached pieces of information, each of which is intended to elicit a buzz in isolation. Most people, of course, exist somewhere on the spectrum between these two extremes. Thus, there is a wide range of views as to whether guesswork is positive, and to what degree. My own view is that the five-point penalty is by no means excessive.

(2) I don't agree with most of Chris's descriptions of or assumptions about quizbowl strategy. I don't think that one can draw a dichotomy between "random" and "educated" guesses, or that those terms capture the most relevant aspects of guesswork. If you're playing well, your thought process should look something like QANTA's display: multiple possible guesses, and some sort of confidence rating. What confidence threshhold you decide to meet before you pull the trigger should be situational. People who tell you things like "always/never buzz on a 50/50" are insane. If you are playing correctly, your calculation should be primarily about whether the next clue is more likely to help your team or the opposing team. (For this reason, stats are not always a good indication of whether someone "negs too much," since they don't tell you anything about whether a neg was well-calculated.)

The potential downside of the five-point penalty increases the lower the expected bonus conversion is (which is one of many reasons why your team's aggression should vary with tournament difficulty). Assuming you're playing a regular-difficulty tournament and your team has decent bonus conversion, the number of situations in which the five-point penalty should change whether or not you make a guess are actually pretty small. And no matter what the scoring system was, you would have to perform similar calculations whenever you decide whether to buzz.

I don't know where this conjecture comes from:

Instead, I would conjecture that the vast majority of negs are educated guesses of some variety

Why assume this? Surely a lot of them are just due to misremembering, misparsing, forgetting a name, thinking the tossup is describing something it's not, etc.

I also don't agree with this:

The problem, though, is that I think very few negs fall into this category of randomly guessing because of your knowledge of your opponents’ strengths. Certainly before the advent of detailed stats there were few ways to gain much knowledge about your opponents’ strengths (especially at the Nationals level) and yet negs were alive and well.

I came to quizbowl without any high school experience and until very recently, I did not spend any time in any quizbowl internet setting outside of the forums. For this reason, the forums, the QB wiki, recordings of past matches, and word of mouth were my main sources of knowledge about other players when I started out. Nonetheless, by my second or third year of quizbowl, I was able to learn the relative strength of my opponents, including their category specialization. This did not require advanced stats. I took correct play seriously, and so I made an effort to do this.

(3) Most inexperienced players--and even many more experienced players--are more worried about the five-point penalty than they should be. But I would say that the net effect is positive! They may be too worried about those five points, but they are probably still not grasping just how large an effect negging has. While I have known some newer players who were frightened to buzz, even when they knew things (which might be exacerbated by the penalty, in some cases), I've encountered a greater quantity of newer players who neg too much. Were the disincentive removed, I think their play would be even less ideal. They may have the wrong reasons for refraining from negging, but they are still adopting the right behavior.

[cwasims]

Thanks for the discussion so far! I would be interested in hearing more about whether people have observed that negs might have a greater impact on female player's playing style - if that's indeed the case, then I think that alone would provide a good reason for experimenting with no-neg formats.

I would agree with Mike that negs have a massively disproportionate impact in tossup-only formats and some more thought should be given to why they're present. Especially in shoot-outs, it's very easy for a tossup to result in a net decrease in aggregate scores, which seems pretty unideal. At the Canadian mirror of IL5, we decided not to play with negs and I enjoyed that quite a bit (I also think we agreed not to engage in the kind of "random guessing" I've not mentioned at length).

I will respond to a few of John's points, though, since they've made me think a bit more about what I said in my initial post. I do think that it is possible to draw a distinction between "random" guesses and "educated" guesses even while acknowledging that in this case (and any other) there is inevitably going to be a continuum. I would think of random buzzes, at an extreme, as what a player would say if they accidentally interrupted a question and are now forced to give an answer. They will presumably say something at least vaguely related to the category if possible and, if they were paying enough attention, one that matches the question's pronoun. Of course, these types of "ideal" random buzzes are very rare in practice, but I think are illustrative of a potentially-optimal play where you give what is essentially a random buzz-type answer to a question because you are certain that the other team will convert if you don't at least guess right now.

I would lump under "educated guess" all the other types of reasons for negging that you would describe, which maybe is not really accurate but I think is helpful in comparing with random guesses. I don't think any of the kinds of educated guessing are particularly worth penalizing - in all of those cases, the player thinks they have some idea what is going on, and are confident enough in their thought process that they think the expected value of buzzing is positive. Of course, how confident they need to be to buzz is going to be endogenous to their opponents' strengths, but I think it is fairly clear how this differs from the sort of random guessing I described above and is much less objectionable from a gameplay perspective than random guessing. In addition, the 5 point penalty alone should change very little about assessing when to buzz if a player is playing optimally (unlike in the case of random guesses, where I contend it does meaningfully change the calculation), so it fails to really achieve any gameplay purpose.

To respond to your point (3), there is almost certainly a "utilitarian" sort of argument to made in favour of having negs, since they tend to make the cost of incorrectly interrupting more salient and thereby encourage people to play in more optimal ways. I think there are other reasons, including those that Henry outlined and perhaps an Ockham's razor-esque approach to game design, for not having as part of a game aspects which primarily exist because of this sort of consideration. Maybe I'm getting a bit too philosophical here, though.

[Cheynem]

I'd be interested to hear more about the impact of negs on female players (maybe through survey?). I do agree that stereotypically and traditionally, female players are less aggressive on the buzzer. In my own experience (I am not female, though) and through talking to women I have been teammates with (as well as men), I have found that the specter of losing 5 points on a neg is usually not the main reason for buzzer fear--rather, it is locking your team out of points ("I don't want to buzz because ___ knows it") or anxiety over embarrassment ("I don't want to make a dumb guess"). In those cases, those fears would remain even if the neg five aspect was eliminated. However, I also don't know how much more the "neg" aspect contributes to those fears.

In regards to shootouts or tossup only formats, I've played a few such tournaments that didn't use negs. They seemed fine--nobody turned into a buzzer maniac or anything. I've also played shootouts that DID use negs and featured people being the most insufferable random guessers ever (i.e., wracking up horrendous neg totals but then finally getting a question right). In that case, the negs didn't deter them anyway. I do think that negs do act as a deterrent to particularly reckless behavior though.

[naan/steak-holding toll]

I would agree with Mike that negs have a massively disproportionate impact in tossup-only formats and some more thought should be given to why they're present. Especially in shoot-outs, it's very easy for a tossup to result in a net decrease in aggregate scores, which seems pretty unideal. At the Canadian mirror of IL5, we decided not to play with negs and I enjoyed that quite a bit (I also think we agreed not to engage in the kind of "random guessing" I've not mentioned at length).

Why is it not ideal for a tossup to reduce aggregate scores? It seems like that's a fair outcome if a bunch of people are coming in with wrong answers. In particular, it seems like the neg increases the risk-reward tradeoff in this and other tossup-only formats, which strikes me as a good thing.

In general, I find negs useful to help players understand their own behavior through statistics. At least from a personal perspective, controlling my number of negs has been made much more concrete by the presence of negs in statistics, and provided a useful benchmark for improvement as a player.

[BenWeiner27]

In general, I find negs useful to help players understand their own behavior through statistics. At least from a personal perspective, controlling my number of negs has been made much more concrete by the presence of negs in statistics, and provided a useful benchmark for improvement as a player.

In addition to helping a player understand their own statistics, neg stats are also super useful for coaches to help their players improve their statlines by either being more aggressive or restrained. At large programs (such as Wayzata) we only had two coaches to manage 6-10 teams depending on the tournament. At most one of them might be able to see a game or two if you are on one of the lower teams, so having the neg stats are extremely useful in understanding how players play the game and helping them make adjustments accordingly.

[Cheynem]

Theoretically you could still track this statistic by marking an "interrupt" instead of a neg and not awarding neg five. I do think John is correct though, that a neg does teach especially newer players of the importance of restraint in buzzing.

[Carlos Be]

I don't think that losing five points on an interrupt matters very much one way or the other. Rather, I think negs cause anxiety because they're recorded. People often obsess over stats, so it can be demoralizing when the stats say "LOOK HOW MANY TIMES YOU FAILED." Similarly, if you neg a bunch of times in an early round, those negs will never go away, and you know that you'll be stuck with a "bad" statline no matter how you play in later rounds. If interrupts weren't recorded, then it would be easier to move on from a bad round and enjoy the rest of the tournament.

I'd also like to push back on the idea that recorded negs help players learn good buzzer strategy. Presumably, negs are helpful to some players. But I know that there are other players for whom negs interfere with good strategy. If you are afraid of a recorded neg, then that fear can creep into your mind and convince you not to buzz on clues that you definitely know, or even multiple consecutive clues that you know. Personally, I am far less likely to buzz on a clue I know if I already have negged twice in a round, even if I've buzzed on the clue before.

[whatamidoinghere]

Based on my sample size of one when I asked my newer teammates, assessing a penalty for negs did help them reduce their neg rate, but I feel like once a newer player learns about the dynamics of the game, it's less necessary to use negs. The reasoning I have is that if you interrupt a question incorrectly, even if you don't get assessed the penalty, you still have to deal with the opponent getting to hear the whole question, convert it, and hear the bonus, possibly either closing what lead your team may have had or maybe them gaining a lead. Plus, it feels bad to know that your team may have lost a really close game at a tournament (say a 5 point game where the neg made all the difference) based on a single neg that may have been wholly reasonable (say in an aggressive move to try to beat your opponent to a topic you like but you had a brain fart). PACE NSC runs without negs (and with bouncebacks too to possibly make up for lost points due to the interrupt) and seems to run fine for teams.

[everdiso]

I agree with a lot of posters that it's important to explain to new players the substantial opportunity cost of a wrong buzz during a question when your teammates might still have been able to get it. However, I think negs might not actually do a good job of this. They probably cause a lot of newer players to think that the cost of a wrong buzz is... well, 5 points. Those players might look at two games in which they went 4/2 and 3/0, getting 30 points each time, as equally good performances. As John said, there are "smart" negs, but in general, 2 negs tend to be more costly than 1 tossup gotten is beneficial (especially in high school, where more players are generalists and your teammates would usually have a decent chance of picking up tossups that you neg later on). If wrong interrupts didn't cost 5 points, I think it might be clearer that the real cost of them is losing the opportunity to get the tossup and hear 30 points' worth of bonus parts. In fact, if wrong interrupts were still tracked as a stat (which I think they should be, since that would require 0 additional work over tracking negs), they might lead newer players to look at those two columns - gets and negs - next to each other as equal, and therefore instinctively think that a 4/2 line is only as good as a 2/0 line, which is probably a lot more accurate, broadly speaking.

Additionally, I have a more philosophical objection to negs. In the pyramidal tossup format, I think the penalty for buzzing in and being wrong is that your team can't get the question anymore and the other team gets to buzz at the end, just like the reward for knowing a question on an earlier clue is that you're less likely to be beaten to it. The addition of a bonus -5 points for being wrong seems artificial - why is it worth half as much as getting the tossup? Why does it apply uniformly throughout the tossup, unlike a power, before disappearing when it ends (leading to weird buzzer races on the last syllable)? I think there should have to be very good reasons to introduce something like this, and as Chris has argued, I think the majority of negs are not things that we need to mess around with the structure of the game with to discourage. (Or even things that can be discouraged - a lot of negs, like John said, come from forgetting the names of things you know, and that can't really be disincentivised.)

[Mike Bentley]

A couple of thoughts:

As someone who negs a lot, I'd personally benefit from this policy. I'm not really convinced it's a huge problem that needs to be solved but as I've argued elsewhere I would like to see more experimentation in quizbowl formats. Particularly to make tossups more interesting as a team activity. While we're throwing wild ideas out there, why not consider scenarios where someone's teammates can still buzz after a wrong answer with some penalty attached to it (fewer points, something like a -20 on a second neg)?

I'm a little skeptical of the argument that new players can't be expected to understand that negging a tossup is worth more of a penalty than 5 points. Every single time this happens, the player will have the experience of the bonus going to another team. Quizbowl already has a high barrier to entry--you need to know at least some facts about some things to play. Are there really people out there that can meet that criteria but can't grasp that their team is penalized by wild guessing? In my experience, the most egregious over-buzzers at the first few college practices of a year come from people who have experience in other formats that more reward wild guessing on the first clue.

Removing negs from shootouts seems like a bad policy to me. It effectively removes any penalty for guessing on a category you're not good at. If I know that I'm almost never going to get a legitimate buzz on, say, a classical music question on a country against John Lawrence, it's heavily in my interest to buzz early with a plausible country even if I only have a 5% chance of getting it right.

[aurochs-and-angels]

PACE NSC runs without negs (and with bouncebacks too to possibly make up for lost points due to the interrupt) and seems to run fine for teams.

Not only does PACE not have negs, but on top of this powers are worth 20 points. Furthermore, given that it is a high school tournament, a sizeable number of the answer lines are things which could be tossed up at any high school tournament.

Though I know some things from AP classes and accidental packet osmosis, I’m not good at science questions. I have a very solid grasp of the sort of things which get tossed up, but never read about them in depth or learn the clues because I don’t care about science: against a top team at PACE with multiple science players, I will get maybe 3-7 % of the questions.

As a solo team, there is no reason that I should not guess in power on every single science tossup (given an elite opponent) at PACE, and indeed would have done so this year if not robbed by coronavirus of my experiment. Though many questions are unguessable, many can be frauded due to the smaller high school canon. If I hear “this kind of radiation” then I should immediately buzz and randomly guess. Ditto with “this phylum,” “these numbers,” “this geological period,” “this distribution,” “these particles,” “this planet,” and dozens more that I won’t list (sometimes editors are more vague until dropping these descriptors in the second/third line). Already my chances of getting these are much higher than 5%, plus I get 20 points for a power AND if I’m wrong, the other team now will likely not go for a power. Nor does it have to be a super transparent three word buzz. If I hear “this equation” I can usually deduce it is chemistry (or whatever) by the second/third line without real knowledge, and should then immediately guess some chemistry equation. “This quantity” can begin to sound like some circuit thing very early: hey, let’s just buzz and guess resistance! I am of course oversimplifying this tremendously too: the more intuitive sense* you have, the better you can take something like "this molecule" and narrow it down quickly beyond merely "biology." Some may object that this itself takes knowledge, which I suppose is true, but not necessarily more knowledge than gets me the "genuine" 3-7%.

Again, getting all 4 of these wrong is no big deal: I would have got destroyed anyway, and perhaps the opponent will score even less total points (!!!) by not going after 20s. But getting even one right, let alone two, could enormously swing a game, especially considering that all my correct buzzes will be worth 20 points. If nothing else, it would be exceedingly amusing: presumably one would have to justify every round why they inexplicably first line negged every science question, and will have to defend themselves of cheating accusations on any of the correct five word reflex guesses**.

*It goes without saying, I hope, that the calculus in this post is not limited to "guessing randomly on science questions." Teams have all sorts of peculiar lacunas in their knowledge bases, and players frequently buzz by "guts" due to a number of factors like specific opponent knowledge and their current lead/deficit. The continuum between "educated" and "random" has rightly been noted: if nothing else, my post should demonstrate that having no negs, in addition to powers worth 20 and a high school canon, drastically would alter the calculus involved in general risk taking.
**I'm really getting ahead of myself, but someone could also make a list the weekend before PACE of the various answerlines which could accompany "this molecule" or "this scientist" and so on. Then, after each round, they could cross these out, given that repeats are unlikely. They could also attend every game in their bye, and perhaps by playoff time their guess probability will be nontrivially higher! I relegate this observation to a measly footnote because doubtless someone will question the quizbowl ethics of someone with the effrontery to try such a thing instead of partaking in the sublime joy of learning science. (I am unsure if I would have tried this strategy this year or not)