Negs

[David Riley]

This is a philosophical iniquiry more than anything--I don't really expect to see change--but:

Why do we have negs?

For a long time, many in the quizbowl community have reailed against Illinois' silly behavioral rules (e.g. matching tops, pencils down, etc.)and rightfully so. How are negs any different? It's almost like a player is being told (naah, naah, naah, you buzzed in too early-I can picture the finger-wagging schoolmarm now).

I suppose I can see it as a check against buzzer-happy players who don't really listen to the questions, but is this problem really widespread, and are there other reasons? Converesely, I can understasnd powers as a reward for deeper knowledge of a question and getting early clues, but negs? Just curious.

[Captain Sinico]

Negs change the competitive equilibrium, having the effect of discouraging wild guessing somewhat. A proper analysis indicates that the points cost by negs matter only in a very small fraction of games anyway and the same is true of points gained by power. They're different from Illinois' stupid behavioral rules because they implement a point change for an in-game action; if negs are of that ilk, then so are all scoring rules.

MaS

[Nuclear Densometer Test]

Okay, so this is probably wrong in every aspect, but it's what came to mind.

Negs are like the -1/4 point deduction on the SAT. They are there to encourage "guesses" based on Real knowledge only and to deter players from answering via illegitimate means (frauding, etc.). When paired with powers, it creates a risk/reward system. Since, ideally, a player ought to know the answer at least after the power mark (where most players' knowledge will lie). So the player has a +5 possibility to answering it correctly early as well as a -5 possibility for incorrectly answering it. Now from here it is clear that the powers only extend to say, half the tossup, while the negs are throughout. I believe that the reasoning behind this is also via the risk/reward system. Players don't know where the power mark is so if they are only 50% sure of the answer, it gives them a choice to either a) buzz, hope for a 15 and risk a -5, b) Know that the possibility of power is over, assume that the other team does not know it, and wait for a clue that narrows your choices down or c) Buzz anyway, risk a -5 on a 50% shot for only 10 points (assuming they know the power mark is over).
Negs add another form of competition within the activity. It makes players weary of the possible consequences and opens up the door for players with Real knowledge to answer, with less of a risk.

Now, most of this will only apply when there are both powers and negs. Although this does encompass many high school tournaments such as NAQT and HSAPQ, college tournaments operate under a different methodology of negs, no powers which has been mimicked in many "advanced" high school tournaments such as Prison Bowl and Harvard Fall. I would say that this adds yet another layer of competition to deter bad guesses throughout the entire tossup without the reward of power.

[quizbowllee]

After deep consideration and meditation on this issue I have come to the following conclusion:

Powers and negs make the game more fun.

[Stained Diviner]

The definitive post on this topic is here.

Negs are useful even without powers.

[David Riley]

Thanks everyone, especially to Lee for his succinct answer

[the return of AHAN]

At the St. Viator Septemberist, Barrington A had the most negs of any team, yet finished 11-0. Is there a lesson to be taken from that? I think so.

[Cheynem]

A good question would be how to help novices overcome the stifling fear not to neg. I would argue that for a lot of new people who get involved in quizbowl, assuming a basically strong level of knowledge, their early stats tend to look like: 20/0 or whatever, indicating that they knew things, buzzed when they knew them, but were a little tentative sometimes.

Here's some things I have thought about in helping people overcome neg fear (or buzzer fear in general, I suppose):

1. Point out that the best players do neg, sometimes at a pretty reasonable clip. To use terrible sports analogies, good baseball players still strike out a lot and good quarterbacks throw interceptions.

2. Stress the difference between a BAD neg and a "GOOD" neg (no neg is good, but you know what I mean). A bad neg is a neg made by not paying attention, like just buzzing on a reflex word or just taking wild, frivolous guesses. "Good" negs might be good-faith attempts to match information with clues, a good buzz somewhat foiled by other circumstances (wrong pronoun, wrong proper name), or other reasons. Discourage the bad negs and stress why "good" negs were tolerable.

3. Make a neg into a learning experience (I mean, not in a didactic, pompous way). Like I screwed up Zulu and Ashanti at IO last year. A quick recap of what the Ashanti thing is about got me on the right track.

[New York Undercover]

Hmm, why is buzzing in after the tossup ends not worthy of a neg? What's the difference in guessing incorrectly after and during the tossup?

[Mechanical Beasts]

Hmm, why is buzzing in after the tossup ends not worthy of a neg? What's the difference in guessing incorrectly after and during the tossup?

I assume it's because when you buzz in in the middle, you're inherently overestimating your own knowledge and letting your opponent hear a bunch of easier clues for free. When you buzz in at the end, it's your last chance to get your team the tossup--the only reason to penalize you at all is if I'm buzzing on history and Bruce Arthur is on my team. (And then, not converting the tossup and Bruce's icy gaze is enough punishment.)

[btressler]

I could be wrong on this, but I think that negs were to discourage interrupt buzzing on the early College Bowl shows so that the audience would hear the entire question.

I personally wouldn't mind if negs went away. You're already punished in that the other team can hear the balance of the tossup.

[silverscreentest]

I personally wouldn't mind if negs went away. You're already punished in that the other team can hear the balance of the tossup.

I agree. It's Academic used to have toss-up round with no penalty and no bounce back to the other teams. I was in the Washington championship where the teams figured this out. So Mac McGarry would get about four words out and a team would buzz in with something far away from the answer, so nobody scored anything. The rules were immediately changed for the first game of the following season.

[DumbJaques]

Negs and powers offer a much more informative slate of statistical results, while being minimally difficult to implement. Independent of any other arguments, I've always found this reasoning to be justification enough to have both tossups and negs (though I think inherently, such reasoning does imply that you should have both, rather than just one).

[Mechanical Beasts]

Well, you could always have neg-zeros and power-zeros for statistical purposes, then.

[cvdwightw]

I still stand by my post in that other thread (as well, I suspect, I should). I will now present a quick addendum explaining why there should not be negs at the end of a tossup.

In that post, I presented the idea of the "marginal utility of not buzzing off a specific clue" as u = dk(1-p), where d is the point differential from a correct buzz, k is the chance of you getting the question off a subsequent clue, and p is the chance of your opponent getting it off a subsequent clue.

At any point before the end of the question, k and p are on the interval (0,1) (unless there is a neg, in which case the relevant variable is 0). That is, there is some positive utility to be gained from not buzzing on that clue (it is an odd fact that the marginal utility for both teams is always > 0, but this does make some sense as d has been normalized such that the team that does not buzz gets 0 points; therefore there is some chance of you getting 0 points and some chance of you getting positive points).

However, consider the end of the question: there are no subsequent clues. Therefore k = 0, so u = 0. That is, there is no utility to not buzz off the last clue. As I stated in that post, with negs, you are wagering 5 points plus the utility of hearing more of the question - in this case, wagering 5 points plus 0 utility. It is still in one's best interest to guess as long as one has greater than a 5/d chance of getting the question right (that minimum bound between ~6% and 25%), but considering that players with small knowledge bases greatly underestimate their chance of getting the question right, many will believe it to their advantage to not buzz in.

Now let us assume that there are no negs. You are only wagering that utility. The marginal utility of not buzzing is still positive at any point in the question, so you are indeed wagering some positive number of points by buzzing in early. At the end of the question, the utility is 0 again. You are therefore getting a free guess at the answer at the end of the question when the neg is not in play.

By removing the neg at the end of the question, the format ensures that there exists a point at which any player can buzz without wagering any real or possible points. At all points during the actual reading of the question, we wish to reward actual knowledge over random guessing (consider that when you know the answer, you are gaining d points with probability 1 and losing 5+u points with probability 0, whereas when you guess, you have some chance of gaining d points and some chance of losing 5+u points depending on how aggressively you play). However, at the end of the question, if actual knowledge was extant, it would have been rewarded by that point. Therefore we are tasked with either rewarding guessing of some kind or rewarding not buzzing.

If we keep the neg in play, there is a good chance that the unconfident player will choose not to buzz. After all there is a good chance of losing 5 points and an artificially deflated chance of gaining d points, so the player's expected buzz value is still negative. However, by eliminating the neg at the end of the question, we remove any chance of losing points. The expected value of a buzz is thus positive (if only minutely so) for any buzz. Thus, it is always in the player's best interest to buzz at the end of the question. Since we want to reward buzzing at the end of the question, it stands that we should eliminate the end-of-question neg.