Paired Tossups and Bonuses

This forum is for discussing tournament formats, question styles, strategy, and such.
Post Reply
User avatar
a named reaction
Rikku
Posts: 345
Joined: Sun Jun 15, 2014 9:16 am

Paired Tossups and Bonuses

Post by a named reaction »

Split from Parsing Answerlines thread - Mgmt.

At the risk both of iconoclasm and derailing this thread, surely the simplest significant quality-of-life improvement to be made in quizbowl packets would be to put bonuses between tossups and have the bonuses associated with dead tossups be skipped; this would both prevent readers from having to switch back and forth between sections of the packet (also meaning that for SCT or other tournaments where teams are given physical packets they may not have to re-staple them), make tossup-bonus feng-shui issues be easier to detect (and eliminate issues where bonus N has issues with tossup N+X that are only apparent if X tossups go dead), and prevent the (possibly merely imagined by me) temptation of putting worse bonuses at the end of packets where they are less likely to be heard by teams. I can imagine objections relating to making it easier for people to study just tossups or just bonuses, or that there are other issues regarding packetization or set production that I, never having done those things, am unaware of, but I don’t think I’ve ever seen a clear argument for why bonuses are not skipped.
Eric Wolfsberg
Bethlehem Central High School 2016
University of Delaware 2020
Stanford 2025 or whatever
User avatar
Mike Bentley
Sin
Posts: 6076
Joined: Fri Mar 31, 2006 11:03 pm
Location: Bellevue, WA
Contact:

Re: Faster (Human) Parsing of Answer Lines

Post by Mike Bentley »

Milhouse wrote: Mon Aug 05, 2019 10:14 pm At the risk both of iconoclasm and derailing this thread, surely the simplest significant quality-of-life improvement to be made in quizbowl packets would be to put bonuses between tossups and have the bonuses associated with dead tossups be skipped; this would both prevent readers from having to switch back and forth between sections of the packet (also meaning that for SCT or other tournaments where teams are given physical packets they may not have to re-staple them), make tossup-bonus feng-shui issues be easier to detect (and eliminate issues where bonus N has issues with tossup N+X that are only apparent if X tossups go dead), and prevent the (possibly merely imagined by me) temptation of putting worse bonuses at the end of packets where they are less likely to be heard by teams. I can imagine objections relating to making it easier for people to study just tossups or just bonuses, or that there are other issues regarding packetization or set production that I, never having done those things, am unaware of, but I don’t think I’ve ever seen a clear argument for why bonuses are not skipped.
If you skip bonuses, you can end up with a round where, say, no science bonuses are read because you just happened to miss the tossups with those bonuses. While I would personally see that as a net positive, science players likely wouldn't be very happy. Yes, the same thing could happen if you randomized the packet such that the last 4 bonuses were all science, but in practice no one does that.

Although it is interesting that modern quizbowl has (apart from a few side events) decoupled the tossup subject from the bonus. In quizbowl-adjacent activities such as game shows and bad local high school formats of yesteryear, bonuses were often a reward to see how deep you can go after getting the initial question. Perhaps the pyramidal tossup makes that redundant. But maybe a tournament should experiment with paired tossup/bonuses categories (specific subjects probably being too much of a pain).
Mike Bentley
VP of Editing, Partnership for Academic Competition Excellence
Adviser, Quizbowl Team at University of Washington
University of Maryland, Class of 2008
User avatar
Irreligion in Bangladesh
Auron
Posts: 2103
Joined: Thu Jul 08, 2004 1:18 am
Location: Winnebago, IL

Re: Faster (Human) Parsing of Answer Lines

Post by Irreligion in Bangladesh »

Milhouse wrote: Mon Aug 05, 2019 10:14 pm At the risk both of iconoclasm and derailing this thread, surely the simplest significant quality-of-life improvement to be made in quizbowl packets would be to put bonuses between tossups and have the bonuses associated with dead tossups be skipped; this would both prevent readers from having to switch back and forth between sections of the packet (also meaning that for SCT or other tournaments where teams are given physical packets they may not have to re-staple them), make tossup-bonus feng-shui issues be easier to detect (and eliminate issues where bonus N has issues with tossup N+X that are only apparent if X tossups go dead), and prevent the (possibly merely imagined by me) temptation of putting worse bonuses at the end of packets where they are less likely to be heard by teams. I can imagine objections relating to making it easier for people to study just tossups or just bonuses, or that there are other issues regarding packetization or set production that I, never having done those things, am unaware of, but I don’t think I’ve ever seen a clear argument for why bonuses are not skipped.
The IHSA does this, and for the reason Mike noted and a few others, I wish they didn't.

When tossups & bonuses are paired, the head editor has to make official decisions as to wish bonuses are tied with which tossups; when they aren't paired, the vagaries of randomly missed questions relieve the head editor of that burden, and any perceived faults get chalked up to chance. If bonuses are paired, you invent the feng shui problem of "am I going to give this history tossup a history bonus on purpose?" If you decide that's a problem - and the IHSA did decide that, and forbids category/category matchups - you then have to A: do the work of looking through your randomized question assignment and checking all your pairs, and B: make the judgment call of what questions to swap. Now, instead of worrying about the temptation to bury bad bonuses at the back, it would be burying them on (the hardest tossups/the tossups you like the least/early tossups so it doesn't feel like the bonuses influence the final outcome of the game/whatever bad justification you like instead). If you don't decide that's a problem, you open yourself up to intentionally bad feng shui & claims of bias (hyperbole, but think "we were doomed, they paired all the science with science and gave literature bonuses to all the history tossups, so we had 10 PPB and they had 30"). Either way, I think it's better to leave them unpaired.

Mike noted the experiment of intentionally pairing TU/B categories - USABB does this, and I don't particularly like it; I think it artificially increases bonus conversion. If you want a "normal"-feeling game of quizbowl, you have to increase bonus difficulty, and at the MS level, that's unwise. Obviously, it's a different story in, say, a side event at a summer weekend event.
Brad Fischer
Head Editor, IHSA State Series
IHSSBCA Ombudsman

Winnebago HS ('06)
Northern Illinois University ('10)
Assistant Coach, IMSA (2010-12)
Coach, Keith Country Day School (2012-16)
User avatar
a named reaction
Rikku
Posts: 345
Joined: Sun Jun 15, 2014 9:16 am

Re: Faster (Human) Parsing of Answer Lines

Post by a named reaction »

Yeah, both of those sets of arguments are reasonable. Thanks!
Eric Wolfsberg
Bethlehem Central High School 2016
University of Delaware 2020
Stanford 2025 or whatever
User avatar
QuestionCactus
Lulu
Posts: 39
Joined: Sat Nov 17, 2018 9:55 pm

Re: Faster (Human) Parsing of Answer Lines

Post by QuestionCactus »

Mike Bentley wrote: Mon Aug 05, 2019 6:45 pm I think it would be an interesting exercise for someone to do a top-down review of the quizbowl packet / scoresheet from a design perspective.
Milhouse wrote: Mon Aug 05, 2019 10:14 pm At the risk both of iconoclasm and derailing this thread, surely the simplest significant quality-of-life improvement to be made in quizbowl packets would be to put bonuses between tossups
When I was writing a packet for a brief side-event-type thing at HSNCT, I divided the layout into two columns with the tossups on the left and bonuses on the right: https://quizbowlpackets.com/2326/Latin-Packet-I.pdf (note that because the bonuses were slightly longer than the tossups, they would have outrun them if I hadn't made some manual adjustments).

I think this format is visually appealing and easy to read, and much more space-efficient than the current standard design.
Arjun Panickssery [Clarke High School, NY]

Download QuizDroid on Google Play
User avatar
vinteuil
Auron
Posts: 1455
Joined: Sun Oct 23, 2011 12:31 pm

Re: Faster (Human) Parsing of Answer Lines

Post by vinteuil »

I personally find this formatting extremely cramped and unnatural to read. (I have a lot of trouble with double-column text in general)
Jacob Reed (he/him/his)
Chicago, Yale, East Chapel Hill
User avatar
Cheynem
Sin
Posts: 6786
Joined: Tue May 11, 2004 11:19 am
Location: Grand Rapids, Michigan

Re: Faster (Human) Parsing of Answer Lines

Post by Cheynem »

Yeah, I'm not sure on the benefit of that, especially if you're following normal conventions of not linking the tossups to the bonuses.
Mike Cheyne
Formerly U of Minnesota

"You killed HSAPQ"--Matt Bollinger
User avatar
the return of AHAN
Auron
Posts: 1958
Joined: Thu Feb 01, 2007 10:40 pm

Re: Paired Tossups and Bonuses

Post by the return of AHAN »

I've long maintained that pairing TU/bonuses usually has the unintended consequence of having MORE bonuses in categories where the toss-up is missed, since writers strive to ask about something other than the TU category. So, in a 24 TU match, if 4 math TU go dead because both teams are weak at math, math becomes THE most asked about bonus topic.
Jeff Price
Barrington High School Coach
Barrington Station Middle School Coach (2013 MSNCT Champions, 2013 & 2017 Illinois Class AA State Champions)
------------------------------------------------------------------------------
User avatar
btressler
Tidus
Posts: 619
Joined: Fri Mar 26, 2004 7:23 pm
Location: West Chester, PA
Contact:

Re: Paired Tossups and Bonuses

Post by btressler »

One of TJ's tournaments (the one that used to be MLK weekend I believe) used to do this. If you answered a tossup on say Alexander the Great, you got a bonus about the lands he conquered. Or perhaps films about Alexander.

I wouldn't want it every week, but I kinda liked it.
Bill Tressler,
Dickinson ('97) Carnegie Mellon ('99) Delaware ('06)
Seen moderating at various SE Pennsylvania events.
User avatar
Whiter Hydra
Auron
Posts: 1315
Joined: Tue Dec 04, 2007 8:46 pm
Location: Fairfax, VA
Contact:

Re: Paired Tossups and Bonuses

Post by Whiter Hydra »

btressler wrote: Tue Sep 10, 2019 10:00 am One of TJ's tournaments (the one that used to be MLK weekend I believe) used to do this. If you answered a tossup on say Alexander the Great, you got a bonus about the lands he conquered. Or perhaps films about Alexander.

I wouldn't want it every week, but I kinda liked it.
The major problem having tossups and bonuses linked that much was that you had to take away potential clues from the bonus. It was a neat idea in theory, but the execution left a lot to be desired.
Harry White
TJHSST '09, Virginia Tech '13
VP of Technology, PACE
Owner of Tournament Database Search and Quizbowl Schedule Generator
Will run stats for food
User avatar
Carlos Be
Wakka
Posts: 188
Joined: Sun Jun 25, 2017 11:34 pm

Re: Paired Tossups and Bonuses

Post by Carlos Be »

Adding to Eric's points, paired bonuses would make tracking conversion data a lot easier. Currently, moderators often skip bonuses (either by mistake or misunderstanding), making it difficult to tell what bonus was read in a given tossup/bonus cycle, therefore making it difficult to read bonus conversion data from a scoresheet. If bonuses are paired, then this is not an issue. For any set that wants to record conversion data, I think this single advantage greatly outweighs all of the minor issues brought up in this thread.
Mike Bentley wrote: Tue Aug 06, 2019 11:34 am If you skip bonuses, you can end up with a round where, say, no science bonuses are read because you just happened to miss the tossups with those bonuses. While I would personally see that as a net positive, science players likely wouldn't be very happy.
With reasonable conversion numbers, this would happen in less than 1% of games. Also, it already happens that a team doesn't hear a given category (for example, because the other team got all of that category's bonuses), and it isn't a big deal.
Irreligion in Bangladesh wrote: Tue Aug 06, 2019 12:29 pm When tossups & bonuses are paired, the head editor has to make official decisions as to wish bonuses are tied with which tossups; when they aren't paired, the vagaries of randomly missed questions relieve the head editor of that burden, and any perceived faults get chalked up to chance.
The head editor already decides which bonuses are probably tied to which tossups, and they are already blamed when a tossup goes to a bonus of the same category. So shifting part of the blame to chance doesn't help much.
the return of AHAN wrote: Mon Sep 09, 2019 1:58 pm I've long maintained that pairing TU/bonuses usually has the unintended consequence of having MORE bonuses in categories where the toss-up is missed, since writers strive to ask about something other than the TU category. So, in a 24 TU match, if 4 math TU go dead because both teams are weak at math, math becomes THE most asked about bonus topic.
Yes, but not by very much. In this case, math would go from 17% of the bonuses to 20% of the bonuses. This change would only be significant if the math were egregiously bad (e.g. computation), but in that case the real problem is that the math is bad, not that the bonuses are paired.
Justine French
UCLA
Subotai the Valiant, Final Dog of War
Wakka
Posts: 113
Joined: Tue Aug 08, 2017 6:12 pm

Re: Paired Tossups and Bonuses

Post by Subotai the Valiant, Final Dog of War »

Adding to Eric's points, paired bonuses would make tracking conversion data a lot easier. Currently, moderators often skip bonuses (either by mistake or misunderstanding), making it difficult to tell what bonus was read in a given tossup/bonus cycle, therefore making it difficult to read bonus conversion data from a scoresheet. If bonuses are paired, then this is not an issue. For any set that wants to record conversion data, I think this single advantage greatly outweighs all of the minor issues brought up in this thread.
Strong agree.

I also feel like linking bonuses makes the game slightly more fair, since what bonus you are read does not functionally depend on what tossups have been answered. Having the parameters of the entire rest of the game be arbitrarily changed because tossup 1 was hard doesn't seem particularly desirable, if you really think about it that way.
The head editor already decides which bonuses are probably tied to which tossups, and they are already blamed when a tossup goes to a bonus of the same category. So shifting part of the blame to chance doesn't help much.
Also true.

However, deciding that a bonus will never be the same category as the tossup is a rather arbitrary decision that is neither random nor particularly justified (is there any reason one should specifically have to answer a bonus on a topic completely different from the tossup one gets?). If the set is 20% history and 20% science, then on average 20% of bonuses after a history tossup should be history and 20% should be science. Otherwise, teams are hearing substantially different bonus distributions than advertised based on how good they are at various categories and how strictly the editor wants to enforce "no same category bonuses after tossups." Say a team averages 3 history tossups and 6 other tossups a game (something similar to this was definitely true of some of my teams). If there are 0 history bonuses after history tossups, they will hear, instead of 1.8 history bonuses per game, 1.5 history bonuses, which is a 17% decrease in the number of expected history bonuses from random. This strikes me as a highly undesirable change that severely punishes specialism.
Daniel, Hunter College High School '19, Yale '23
User avatar
Mike Bentley
Sin
Posts: 6076
Joined: Fri Mar 31, 2006 11:03 pm
Location: Bellevue, WA
Contact:

Re: Paired Tossups and Bonuses

Post by Mike Bentley »

I don't have very strong opinions about paired bonuses. I don't necessarily see how they would lead to moderators not making errors anymore unless you make them inline after the tossup. And even then you're still going to have some errors that show up because the moderator accidentally skipped a tossup or a tossup went dead and on the next tossup the scorekeeper forgot to skip a tossup. Maybe these errors would be less common than with unpaired bonuses.
Mike Bentley
VP of Editing, Partnership for Academic Competition Excellence
Adviser, Quizbowl Team at University of Washington
University of Maryland, Class of 2008
User avatar
Mike Bentley
Sin
Posts: 6076
Joined: Fri Mar 31, 2006 11:03 pm
Location: Bellevue, WA
Contact:

Re: Paired Tossups and Bonuses

Post by Mike Bentley »

The hotter take is that bonuses end up making a difference in a very small number of games. A large portion of a quizbowl match is devoted to them. As I've argued elsewhere, I'd love to see new approaches to bonuses that preserve the teamwork element but make them more engaging, especially for a team that doesn't answer very many tossups.
Mike Bentley
VP of Editing, Partnership for Academic Competition Excellence
Adviser, Quizbowl Team at University of Washington
University of Maryland, Class of 2008
User avatar
btressler
Tidus
Posts: 619
Joined: Fri Mar 26, 2004 7:23 pm
Location: West Chester, PA
Contact:

Re: Paired Tossups and Bonuses

Post by btressler »

Mike Bentley wrote: Thu May 21, 2020 5:45 pm The hotter take is that bonuses end up making a difference in a very small number of games.
I've often wondered if anyone has analyzed a tossup/bonus tournament to see how the records change if only the tossup points were in play. Does anyone have data to share?
Bill Tressler,
Dickinson ('97) Carnegie Mellon ('99) Delaware ('06)
Seen moderating at various SE Pennsylvania events.
Arras Agility
Wakka
Posts: 105
Joined: Mon May 30, 2016 2:51 pm

Re: Paired Tossups and Bonuses

Post by Arras Agility »

btressler wrote: Fri May 22, 2020 10:58 am
Mike Bentley wrote: Thu May 21, 2020 5:45 pm The hotter take is that bonuses end up making a difference in a very small number of games.
I've often wondered if anyone has analyzed a tossup/bonus tournament to see how the records change if only the tossup points were in play. Does anyone have data to share?
I just did a quick-and-dirty analysis on CALISTO Online; 66/72 games in the Standard division and 77/80 games in Competitive; 143/152 ~= 94% of games were won by the team with more tossup points; 69/72 in Standard and 72/80 in Competitive = 141/152 ~= 93% by the team that got more tossups. This isn't quite seeing "how the records change", but I think it's pretty clear that bonus points don't matter nearly as much as tossups. (That said, of course bonuses are valuable for other reasons, and I personally enjoy them more than tossups)

On another note, I fully endorse everything Justine's said upthread, and would add a few other points.

1. Pairing tossups and bonuses allows for better editing/proofreading of finished packets (that is, catching feng shui issues - if your American Lit bonus is a couple questions after your tossup, it's easier to catch it when reading over the packets in the same order that the questions will be read to players) - this probably isn't a problem for sets, but it is a small upside.

2. Whenever I mod with paper packets, the first thing I, and I'm sure many, if not most others, invariably do is *pull the packet apart* - if you step back, it's pretty strange a priori to be saying that the printed form of the questions is not how the questions are to be used. Eliminating this would be more logical, and reduce unnecessary paper-shuffling and place-refinding with paper packets, or tab-switching and place-refinding with online packets. This could also have the minor effect of slightly speeding up games - I usually spend a couple minutes over the course of a game juggling paper packets and a scoresheet and refocusing between them.

3. Contra those arguing that the resulting distributional change would be a bad thing - all-subject quizbowl is already strongly biased against extreme specialism by design. Tossup/bonus categories are different precisely because, I'd argue, part of the ethos of the game is to encourage knowing many genres of thing. It seems that matching tossup/bonus categories, which is pretty much the opposite policy to this, has mostly been used in bad formats or to make the game artificially easier, and I don't see nearly as strong philosophical justification for this direction on the bonus-relatedness spectrum.
Alistair Gray
Homestead '18
UCSD '22
Subotai the Valiant, Final Dog of War
Wakka
Posts: 113
Joined: Tue Aug 08, 2017 6:12 pm

Re: Paired Tossups and Bonuses

Post by Subotai the Valiant, Final Dog of War »

To clarify, I don't think tossup/bonus cycles should be intentionally of the same category (that does indeed help extreme specialism too much, as Alistair has said), but I don't think they should be intentionally different categories either. It's very discouraging if you're a team that's relatively better at one category to know that the more you get from that category, the worse your PPB gets.
Daniel, Hunter College High School '19, Yale '23
User avatar
Carlos Be
Wakka
Posts: 188
Joined: Sun Jun 25, 2017 11:34 pm

Re: Paired Tossups and Bonuses

Post by Carlos Be »

Subotai the Valiant, Final Dog of War wrote: Fri May 22, 2020 2:05 pm To clarify, I don't think tossup/bonus cycles should be intentionally of the same category (that does indeed help extreme specialism too much, as Alistair has said), but I don't think they should be intentionally different categories either. It's very discouraging if you're a team that's relatively better at one category to know that the more you get from that category, the worse your PPB gets.
I think it's desirable not to have specific subcategories go to themselves, like biology --> biology or philosophy --> philosophy. If you mess up on a tossup in a subcategory then you shouldn't forfeit your chance to hear that subcategory's bonus as well. Pairings of broad categories like science --> science or history --> history I think should be allowed for the reasons you said.
Justine French
UCLA
User avatar
CPiGuy
Yuna
Posts: 892
Joined: Wed Nov 16, 2016 8:19 pm
Location: Ames, Iowa

Re: Paired Tossups and Bonuses

Post by CPiGuy »

Carlos Be wrote: Sat May 23, 2020 1:36 pm I think it's desirable not to have specific subcategories go to themselves, like biology --> biology or philosophy --> philosophy. If you mess up on a tossup in a subcategory then you shouldn't forfeit your chance to hear that subcategory's bonus as well.
I think this should happen in approximately one out of every 20 bonuses.
Conor Thompson (he/him)
Bangor High School '16
University of Michigan '20
Iowa State University '25
User avatar
UlyssesInvictus
Yuna
Posts: 827
Joined: Thu Feb 10, 2011 7:38 pm

Re: Paired Tossups and Bonuses

Post by UlyssesInvictus »

CPiGuy wrote: Sat May 23, 2020 5:59 pm
Carlos Be wrote: Sat May 23, 2020 1:36 pm I think it's desirable not to have specific subcategories go to themselves, like biology --> biology or philosophy --> philosophy. If you mess up on a tossup in a subcategory then you shouldn't forfeit your chance to hear that subcategory's bonus as well.
I think this should happen in approximately one out of every 20 bonuses.
I was actually really curious how this would play out in a sub-distribution, so I simulated it (since I'm crap at probability). Using one of the past HFT distributions (so, standard big 3 stuff), it averaged out to about 3.2 questions matching in overall category per packet, and about 1 question matching in sub-category per packet.

The latter number was probably not surprising to someone good at probability, but the former probably also exceeds some % of people's limits for packet feng shui.
Raynor Kuang
quizdb.org
Harvard 2017, TJHSST 2013
I wrote GRAPHIC and FILM
User avatar
Stained Diviner
Auron
Posts: 4830
Joined: Sun Jun 13, 2004 6:08 am
Location: Chicagoland
Contact:

Re: Paired Tossups and Bonuses

Post by Stained Diviner »

My calculations give 1 subcategory and 3.25 category matches per packet if CE/misc is not considered a match, or 3.3 if it is.

If somebody did go to pairing tossups and bonuses, then they could decide ahead of time how many matched categories they want. If they want 0 subcategory and 3 category matches, they can do that. If they want 0 and 2, they can do that. If you want a science tossup paired with a science bonus in exactly half the rounds, that can be arranged. (Science/science would normally happen in 4/5 of rounds by chance. It probably is happening slightly less often now because many editors avoid having the 1st tossup and 1st bonus be in the same category, and possibly continuing that throughout much or all of the packet.)

For timed matches, I as a moderator would prefer that they are paired. I don't track bonus conversion data myself, so I'll defer to others on that point. Other than that, I don't really have an opinion on this topic. Mike Bentley's first post probably is the strongest argument in either direction, and it is against pairing, but I think in timed matches the shuffling papers argument overrides that one. Overall, I don't think the arguments in either direction are that strong, which is why I don't have a strong opinion.

As to Alistair's point about proofreading, all proofreaders should do their proofing by alternating tossups and bonuses in a packet. It doesn't matter how the packet is arranged.
David Reinstein
PACE VP of Outreach, Head Writer and Editor for Scobol Solo and Masonics (Illinois), TD for New Trier Scobol Solo and New Trier Varsity, Writer for NAQT (2011-2017), IHSSBCA Board Member, IHSSBCA Chair (2004-2014), PACE Member, PACE President (2016-2018), New Trier Coach (1994-2011)
User avatar
naan/steak-holding toll
Auron
Posts: 2269
Joined: Mon Feb 28, 2011 11:53 pm
Location: New York, NY

Re: Paired Tossups and Bonuses

Post by naan/steak-holding toll »

My concern with pairing tossups and bonuses based on subjects is that it disincentivizes teams from covering the whole range of the distribution as strongly as having randomized bonuses does. It also means you have to contort your tossup to not reveal what you're asking about in the paired bonus, if you're trying to cover the same material (though perhaps it works if it's just "American Literature" or whatever). Additionally, it doesn't work well if you have less than 1/1 of some categories per round / fractional distributions / etc.

I like bouncebacks and wish they were used more often. They keep teams engaged the whole time. It presumably sucks for a weaker team when a stronger team steals their whole bonus, but on the flip side, they get more chances at those few things they know really deeply (i.e. the person who really likes jazz and gets really excited for that 3/3 jazz per tournament, but the jazz bonus goes to the other team). Bouncebacks also have the effect of helping smooth out variance and decrease the chance of upsets - not sure if people like this or not, I think you can make an argument both ways.
Will Alston
Bethesda Chevy Chase HS '12, Dartmouth '16, Columbia Business School '21
NAQT Writer and Subject Editor
User avatar
Cheynem
Sin
Posts: 6786
Joined: Tue May 11, 2004 11:19 am
Location: Grand Rapids, Michigan

Re: Paired Tossups and Bonuses

Post by Cheynem »

Yeah, I don't support pairing tossups/bonuses by subject, but I'm not opposed to just specifically designating a bonus as bonus 1 or bonus 2 (so bonus 2 is always read after tossup 2)--there's some argument to be made that this can actually result in a more balanced distribution of questions in a packet.

I have some thoughts on bouncebacks, but I wonder if another thread is better for that.
Mike Cheyne
Formerly U of Minnesota

"You killed HSAPQ"--Matt Bollinger
whatamidoinghere
Wakka
Posts: 104
Joined: Wed Apr 25, 2018 10:07 pm

Re: Paired Tossups and Bonuses

Post by whatamidoinghere »

Cheynem wrote: Sat May 30, 2020 7:54 pm Yeah, I don't support pairing tossups/bonuses by subject, but I'm not opposed to just specifically designating a bonus as bonus 1 or bonus 2 (so bonus 2 is always read after tossup 2)--there's some argument to be made that this can actually result in a more balanced distribution of questions in a packet.

I have some thoughts on bouncebacks, but I wonder if another thread is better for that.
I feel like a lot of the time, newer mods will just read bonus 1 for tossup 1, bonus 2 for tossup 2, etc. already, mostly since it seems just in nature (especially for newer mods but also sometimes for experienced mods) to think "hey, we finished tossup 10, I should read bonus 10" even if a tossup went dead. I don't know about the argument for balance, but considering that a lot of mods subconsciously do read bonuses corresponding to tossups, it might make sense to just make bonuses correspond directly to tossups.
Avinash Iyer
Mission San Jose '21
"That was quite wrong"
"You're only as good as your last haircut"
User avatar
Blackboard Monitor Vimes
Forums Staff: Administrator
Posts: 2348
Joined: Sat Aug 18, 2007 5:40 pm
Location: Richmond, VA

Re: Paired Tossups and Bonuses

Post by Blackboard Monitor Vimes »

I've attempted to split discussion of bouncebacks and other more general "why we should or should not have bonuses at all" posts into this thread; apologies if I missed any.
Sam (Angelo) Luongo,
Maggie L. Walker Governor's School 2010 / UVA 2014 / VCU School of Education 2016
PACE
alexdz
Rikku
Posts: 420
Joined: Wed Sep 03, 2008 7:29 pm
Location: Philadelphia, PA

Re: Paired Tossups and Bonuses

Post by alexdz »

Just thought of this... what about designating bonuses by half? So not necessarily matching TU 1 - B1, etc., but saying "These are the first half bonuses, 1-10, and these are the second half bonuses, 11-20." In other words, you read bonuses in order as they come, but once the second half starts, you skip over any remaining first half bonuses and go straight to the second half bonuses. It would allow lower scoring games to hear some of the later bonuses in the packet, without some of the drawbacks of pure pairing. Of course, the paper logistics of this would be much more difficult, too, and that's something to consider.
Alex Dzurick
====
Owner/Editor, SAGES Quizbowl Questions
Coach, Harcum College (PA)
====
Former midwesterner (South Callaway - Mizzou - UIUC) coping with life on the east coast.
User avatar
Ben Dillon
Rikku
Posts: 293
Joined: Thu Apr 28, 2005 10:47 pm
Location: South Bend, IN
Contact:

Re: Paired Tossups and Bonuses

Post by Ben Dillon »

Mike Bentley wrote: Thu May 21, 2020 5:45 pm The hotter take is that bonuses end up making a difference in a very small number of games.
I've seen this kind of statement before, and I believe someone even posted statistics that supported it (i.e. removing the bonus points from both teams didn't change the outcome n% of the time), but I still find it problematic.

Each tossup is worth -5 ≤ x ≤ 15 points and each bonus is worth 0 ≤ y ≤ 30 points. Of course bonuses make a difference; they're worth twice a tossup.

And the closer the two teams match up, the more of a difference it should make:
  • Very Strong versus Very Strong results in a high-scoring match where each team is sweeping their respective bonuses (and stealing bouncebacks). The fact that each bonus is basically double each tossup can easily swing a match.
  • Very Weak versus Very Weak results in a low-scoring match where each team is scrapping to get bonuses that are in their wheelhouse. With these, I've seen the trash bonus swing the match.
I think the statistics would show that most quiz bowl matches aren't between these two teams; most of them are between Very Strong and Strong at national tournaments (or, in my state, a lot of Strong versus Weak). Hence Mike's above statement holds.

But it still makes me uneasy that bonuses are worth so much. So I'd like to throw in a new idea: What if bonuses were 5 points apiece instead of 10? That would mean each bonus would be 0 ≤ y ≤ 15, and x and y would be more on an equal footing in how they impact a final score.

Thoughts?
Ben Dillon, Saint Joseph HS

"Why, sometimes I've believed as many as
six impossible things before breakfast!"
User avatar
Stained Diviner
Auron
Posts: 4830
Joined: Sun Jun 13, 2004 6:08 am
Location: Chicagoland
Contact:

Re: Paired Tossups and Bonuses

Post by Stained Diviner »

Keep in mind that tossups don't just give you the points--they give you the right to the bonus. That's why if there are two evenly matched teams, and one of them gets the tossups 11-9, that team is very likely going to win. If both teams were getting opportunities at 10 bonuses, it wouldn't be unusual for one team to overcome the 20 point difference in tossup scores. However, when you have a 20 point deficit from tossups and you get an opportunity at 2 fewer bonuses, that difference is difficult to overcome. It can happen, but it rarely does.

If a team gets 11 tossups and 20 PPB, then their opponent needs to get 9 tossups and 27 PPB to beat them. (11*30=330, and 9*37=333.) I'm ignoring negs, powers, and bouncebacks to simplify things, and the discrepancy isn't as bad if the tossups go 10-9, but it's still difficult to overcome getting fewer tossups. If a team gets 12 tossups and 17 PPB, then they mathematically clinch the match. (12*27=324, and 8*40=320.)
David Reinstein
PACE VP of Outreach, Head Writer and Editor for Scobol Solo and Masonics (Illinois), TD for New Trier Scobol Solo and New Trier Varsity, Writer for NAQT (2011-2017), IHSSBCA Board Member, IHSSBCA Chair (2004-2014), PACE Member, PACE President (2016-2018), New Trier Coach (1994-2011)
User avatar
Ben Dillon
Rikku
Posts: 293
Joined: Thu Apr 28, 2005 10:47 pm
Location: South Bend, IN
Contact:

Re: Paired Tossups and Bonuses

Post by Ben Dillon »

Stained Diviner wrote: Sun Jun 14, 2020 8:45 am If a team gets 11 tossups and 20 PPB, then their opponent needs to get 9 tossups and 27 PPB to beat them. (11*30=330, and 9*37=333.) I'm ignoring negs, powers, and bouncebacks to simplify things, and the discrepancy isn't as bad if the tossups go 10-9, but it's still difficult to overcome getting fewer tossups. If a team gets 12 tossups and 17 PPB, then they mathematically clinch the match. (12*27=324, and 8*40=320.)
Not sure what you're getting at here. Fewer points per bonus part would make the 11-9 edge MORE insurmountable, right? (Going through the same math with 5 points per bonus part: 11*20=220 for the first team would mean the opponent would have to sweep all 9 bonuses in order to win.)

I'm trying to argue that I think tossups should be worth more to the game as a percentage of the final score. I especially think it should be more tossups deciding a game between two weak teams than my admittedly anecdotal observation that these games are being decided by trash bonuses.
Ben Dillon, Saint Joseph HS

"Why, sometimes I've believed as many as
six impossible things before breakfast!"
User avatar
Stained Diviner
Auron
Posts: 4830
Joined: Sun Jun 13, 2004 6:08 am
Location: Chicagoland
Contact:

Re: Paired Tossups and Bonuses

Post by Stained Diviner »

The rules we currently have makes tossups extremely important and bonuses unimportant, to the point where it's questionable as to why we spend so much time writing and playing bonuses. Your proposal weighs things towards tossups even more, which would mean that bonuses would be a waste of time.

It would be easier to get rid of bonuses than to make them fewer points, and the impact on who wins games would be the same either way.
David Reinstein
PACE VP of Outreach, Head Writer and Editor for Scobol Solo and Masonics (Illinois), TD for New Trier Scobol Solo and New Trier Varsity, Writer for NAQT (2011-2017), IHSSBCA Board Member, IHSSBCA Chair (2004-2014), PACE Member, PACE President (2016-2018), New Trier Coach (1994-2011)
User avatar
naan/steak-holding toll
Auron
Posts: 2269
Joined: Mon Feb 28, 2011 11:53 pm
Location: New York, NY

Re: Paired Tossups and Bonuses

Post by naan/steak-holding toll »

David's point about a team mathematically clinching the match with 12 tossups at 17 PPB is quite compelling. This prompts me to ask - at what points would games be mathematically unwinnable for the team with fewer tossups if bouncebacks are implemented, assuming that there are no powers and no negs?

Here are the results of a quick-and-dirty analysis showing the PPB value for your opponent at which the game becomes mathematically unwinnable if your team gets X number of tossups (i.e. you can't win even if you 30 every bonus and steal every free bonus part):
  • 11 tossups - 27 PPB
  • 12 tossups - 24 PPB
  • 13 tossups - 21 PPB
  • 14 tossups - 19 PPB
  • 15 tossups - 17 PPB
  • 16 tossups - 16 PPB
  • 17 tossups - 14 PPB
  • 18 tossups - 13 PPB
  • 19 tossups - 12 PPB
  • 20 tossups - 11 PPB (you ONLY get points by theft here)
Spreadsheet with analysis is attached. Note that I've had the spreadsheet round to the nearest 10, understanding that some PPB numbers are mathematically impossible since all point values must end with 10 with the conditions I've outlined above. This could doubtless be refined slightly, but I think the general point is illustrated.

You can also fiddle with the scenarios by toggling the "Team B PPB" number at the bottom. Note that the formula assumes that your team's points per bounceback will equal the difference between your PPB and the other team's, so you will get no bouncebacks if the other team's PPB is higher, which isn't how actual games work (the spreadsheet only gives truly "accurate" game results when your team is assumed to get 30 PPB). Nonetheless, the results remain intuitively reasonable, i.e. the spreadsheet will tell you a win is "mathematically impossible" if the opposing team has more PPB than you and you each get the same number of tossups, even though the formula will strictly say you get no bouncebacks.

In other words, if you put in 20 PPB, and the opposing team's PPB is 10, the formula assumes that you get bounceback points equal to (your PPB - other team's PPB) x (number of opposing tossups heard), or 10 points per total opposing team bonus. For a scenario in which your team ("Team B") gets 8 tossups and the opponent gets 12, the spreadsheet will say that your team will get:
  • 80 tossup points (8 x 10)
  • 160 bonus points (8 x 20)
  • 120 bounceback points ((20-10) x 12)
And the opposing team will get:
  • 120 tossup points (12 x 10)
  • 120 bonus points (12 x 10)
  • No bounceback points (this assumption is baked in, since we're trying to find mathematically unwinnable scenarios)
So your team wins by 120 points. Again, highly imperfect methodology outside of the "you 30 everything" scenario, but I leave it to others to refine the approach.

For this reason, I've revised my position to favor implementing bouncebacks in more tournaments. This would doubtless require stricter length control, difficulty control, and other factors to make sure games run in a reasonable time and are smooth.
Attachments
bouncebacks.xlsx
(11.14 KiB) Downloaded 6 times
Will Alston
Bethesda Chevy Chase HS '12, Dartmouth '16, Columbia Business School '21
NAQT Writer and Subject Editor
User avatar
naan/steak-holding toll
Auron
Posts: 2269
Joined: Mon Feb 28, 2011 11:53 pm
Location: New York, NY

Re: Paired Tossups and Bonuses

Post by naan/steak-holding toll »

Here's another argument in favor of bouncebacks: let's pretend you're a player who only contributes on 2/2 per game and absolutely nothing else. Let's furthermore assume that you're on a middle team which gets an average of 10 tossups per game. Across a 10 game tournament, this means you are "contributing" on 20 tossups and an average of 10 bonuses (2 bonuses in your categories per game, on average your team hears half the bonuses, so an expected value of one bonus "contribution" per game). So, you contribute on a total of 30 questions.

If you add bouncebacks to the equation, then - assuming that all questions are converted - suddenly you have 10 more questions that you can potentially contribute on if the other team misses a part on bonuses. So, you're contributing on 33% more questions than you were before simply because bouncebacks were added. That's a lot! And even if you relax the conversion assumption to, say, the other teams getting 8 questions on average against your average 10 (with two dead tossups per game), then you're still getting an average of 8 more chances to contribute. I think this is a meaningful difference for specialized players, especially those on weaker teams which aren't getting a large majority of tossups every game - even against a top team, you've still often at least got a shot at the hard part, or a middle part if they slip up.

In other words, this goes much beyond winning, it goes into how many chances a player has to contribute. Obviously this argument also applies to generalists, but they have a ton of chances to contribute no matter what, so the increase is less meaningful.
Will Alston
Bethesda Chevy Chase HS '12, Dartmouth '16, Columbia Business School '21
NAQT Writer and Subject Editor
User avatar
Mike Bentley
Sin
Posts: 6076
Joined: Fri Mar 31, 2006 11:03 pm
Location: Bellevue, WA
Contact:

Re: Paired Tossups and Bonuses

Post by Mike Bentley »

I'm a little conflicted around whether reducing parity by adding bouncebacks that make it more likely for the better team to win is a good idea or not.
Mike Bentley
VP of Editing, Partnership for Academic Competition Excellence
Adviser, Quizbowl Team at University of Washington
University of Maryland, Class of 2008
jdpasspawn
Lulu
Posts: 99
Joined: Fri Oct 23, 2015 7:00 pm

Re: Paired Tossups and Bonuses

Post by jdpasspawn »

naan/steak-holding toll wrote: Mon Jun 15, 2020 6:06 pm Here's another argument in favor of bouncebacks: let's pretend you're a player who only contributes on 2/2 per game and absolutely nothing else. Let's furthermore assume that you're on a middle team which gets an average of 10 tossups per game. Across a 10 game tournament, this means you are "contributing" on 20 tossups and an average of 10 bonuses (2 bonuses in your categories per game, on average your team hears half the bonuses, so an expected value of one bonus "contribution" per game). So, you contribute on a total of 30 questions.

If you add bouncebacks to the equation, then - assuming that all questions are converted - suddenly you have 10 more questions that you can potentially contribute on if the other team misses a part on bonuses. So, you're contributing on 33% more questions than you were before simply because bouncebacks were added. That's a lot! And even if you relax the conversion assumption to, say, the other teams getting 8 questions on average against your average 10 (with two dead tossups per game), then you're still getting an average of 8 more chances to contribute. I think this is a meaningful difference for specialized players, especially those on weaker teams which aren't getting a large majority of tossups every game - even against a top team, you've still often at least got a shot at the hard part, or a middle part if they slip up.

In other words, this goes much beyond winning, it goes into how many chances a player has to contribute. Obviously this argument also applies to generalists, but they have a ton of chances to contribute no matter what, so the increase is less meaningful.
naan/steak-holding toll wrote: Mon Jun 15, 2020 5:45 pm David's point about a team mathematically clinching the match with 12 tossups at 17 PPB is quite compelling. This prompts me to ask - at what points would games be mathematically unwinnable for the team with fewer tossups if bouncebacks are implemented, assuming that there are no powers and no negs?

Here are the results of a quick-and-dirty analysis showing the PPB value for your opponent at which the game becomes mathematically unwinnable if your team gets X number of tossups (i.e. you can't win even if you 30 every bonus and steal every free bonus part):
  • 11 tossups - 27 PPB
  • 12 tossups - 24 PPB
  • 13 tossups - 21 PPB
  • 14 tossups - 19 PPB
  • 15 tossups - 17 PPB
  • 16 tossups - 16 PPB
  • 17 tossups - 14 PPB
  • 18 tossups - 13 PPB
  • 19 tossups - 12 PPB
  • 20 tossups - 11 PPB (you ONLY get points by theft here)
Spreadsheet with analysis is attached. Note that I've had the spreadsheet round to the nearest 10, understanding that some PPB numbers are mathematically impossible since all point values must end with 10 with the conditions I've outlined above. This could doubtless be refined slightly, but I think the general point is illustrated.

You can also fiddle with the scenarios by toggling the "Team B PPB" number at the bottom. Note that the formula assumes that your team's points per bounceback will equal the difference between your PPB and the other team's, so you will get no bouncebacks if the other team's PPB is higher, which isn't how actual games work (the spreadsheet only gives truly "accurate" game results when your team is assumed to get 30 PPB). Nonetheless, the results remain intuitively reasonable, i.e. the spreadsheet will tell you a win is "mathematically impossible" if the opposing team has more PPB than you and you each get the same number of tossups, even though the formula will strictly say you get no bouncebacks.

In other words, if you put in 20 PPB, and the opposing team's PPB is 10, the formula assumes that you get bounceback points equal to (your PPB - other team's PPB) x (number of opposing tossups heard), or 10 points per total opposing team bonus. For a scenario in which your team ("Team B") gets 8 tossups and the opponent gets 12, the spreadsheet will say that your team will get:
  • 80 tossup points (8 x 10)
  • 160 bonus points (8 x 20)
  • 120 bounceback points ((20-10) x 12)
And the opposing team will get:
  • 120 tossup points (12 x 10)
  • 120 bonus points (12 x 10)
  • No bounceback points (this assumption is baked in, since we're trying to find mathematically unwinnable scenarios)
So your team wins by 120 points. Again, highly imperfect methodology outside of the "you 30 everything" scenario, but I leave it to others to refine the approach.

For this reason, I've revised my position to favor implementing bouncebacks in more tournaments. This would doubtless require stricter length control, difficulty control, and other factors to make sure games run in a reasonable time and are smooth.
So these arguments for bouncebacks imo forget the most important point about bouncebacks, which is that they (for the college game) suck. For one, tournaments only run as fast as their slowest reader. This means for local tournaments that are not full of nats caliber readers they will (at a bare minimum) add an extra 30 minutes to a tournament and I imagine in most sites will add an hour plus. This likely means that at most tournaments rather than just adding an hour to finishing time, we’re most likely talking about missing multiple games. In addition, there are other aspects of bouncebacks that are p grueling (especially for the “inexperienced” or “new” teams that some people argue these bouncebacks are aimed at). You now have to now pay attention to every bonus part, have teams undergo the demoralizing experience of having all three bonus parts stolen by the other team (which of course is far more likely to happen to new teams), and have to whisper on bonuses to avoid stealing your answers, hampering some of the free flowing dialogue that helps make college qb fun. This all adds up to make an average college quizbowl tournament, already a pretty tiring experience for people like me that greatly enjoy them, signifcantly more grueling. People implementing bouncebacks for their own college tournament and reporting the results back to the community sounds like a solid idea, but I think we really need to consider the unique challenges of the college game when we think about whether this makes sense to implement on a large scale.
Jack Lewis
'18 Battle Ground Academy
'22 University of Maryland
User avatar
naan/steak-holding toll
Auron
Posts: 2269
Joined: Mon Feb 28, 2011 11:53 pm
Location: New York, NY

Re: Paired Tossups and Bonuses

Post by naan/steak-holding toll »

So these arguments for bouncebacks imo forget the most important point about bouncebacks, which is that they (for the college game) suck.
It strikes me as rather bold to make this argument when we have extremely limited empirical evidence about whether they do, in fact, suck for the college game, because we have not run very many collegiate tournaments on bouncebacks! The only one I played that had them was VCUO 2015, and I found that rather enjoyable, as it gave teams lots more opportunities to shine on bonuses that other teams bungled. I think you need very strict length control and bonus timing - for example, I think bouncebacks at ACF Nationals and Penn Bowl are rather infeasible at the moment with how such tournaments are current written (maybe ACF Nationals with strictly 8 line tossups and two-line bonus parts would make it work).

To consider the flip side of the "it sucks to have a good team steal your bonus" argument - one of the most common complaints I've heard from newer teams is that they just are completely locked out of the other team's bonus and they just have to sit there. Most new teams recognize that "good teams are good, they're going to kick our ass" and for the most part they're gonna get drubbed on 80% of the questions and do very little. But for every bonus that gets stolen doesn't matter to the outcome of the game, there are 16 bonuses that this other good team gets where they may have a shot at a hard part they really know, or maybe a slip-up on a middle part. So, you have a lot more opportunities to engage.

If bouncebacks mean you have to whisper and develop more effective communication strategies with your team, well, deal with it. Effective communication is a lot of what being a team is about anyways, and you already have to be quiet during the other team's bonus as-is, otherwise you're being a dick (I'm certainly guilty of this on too many occasions). Bonuses are where the "team" element of the game comes into play the most, and as far as I'm concerned, enhancing that is a good thing. Sure, it's nice to get a mental break and zone out occasionally, but I think there are better ways to reduce the mental tax on players - write your questions in clear language, avoid having players take too many mental leaps to figure clues out, and keep your bonus parts tidy, especially at regular season tournaments.

In any case, I agree that it's probably infeasible to make every tournament use bouncebacks, but I'll certainly consider experimenting with them in the (hopefully limited) set of non-NAQT tournaments I plan to work on in the future.
Will Alston
Bethesda Chevy Chase HS '12, Dartmouth '16, Columbia Business School '21
NAQT Writer and Subject Editor
jdpasspawn
Lulu
Posts: 99
Joined: Fri Oct 23, 2015 7:00 pm

Re: Paired Tossups and Bonuses

Post by jdpasspawn »

I think you need very strict length control and bonus timing
So the latter point is going to be extremely difficult to uniformly enforce across an entire tournament unless every room has a nats caliber reader in it (which most local tournaments definitely do not have!). Bounceback timing is arguably the most difficult aspect of quizbowl reading to do (even for veteran readers) and in my experience there tends to be wildly different enforcings of the rule that lead to seemingly arbitrary rulings and massively extended tournament runtime.
If bouncebacks mean you have to whisper and develop more effective communication strategies with your team, well, deal with it. Effective communication is a lot of what being a team is about anyways, and you already have to be quiet during the other team's bonus as-is, otherwise you're being a dick (I'm certainly guilty of this on too many occasions). Bonuses are where the "team" element of the game comes into play the most, and as far as I'm concerned, enhancing that is a good thing.
So as someone who played the large majority of their HS tournaments with bouncebacks (shoutout TACA) this assertion is largely incorrect. Bouncebacks tended to v significantly disincentivize teammwork and discussion because of the (v reasonable) fear that the other team would overhear your conversation and use it to help steal your bonus. In addition, I really don’t think games should be coming down to how well your team can whisper.

Anyways, like I said in my previous post, I think it’s fine if some tournaments want to experiment with bouncebacks and see how it goes. New TDs however (especially those in burgeoning circuits/ circuits with inexperienced readers) should be v wary of trying to implement bouncebacks for their tournaments for the reasons I’ve laid out in my last two posts.
Jack Lewis
'18 Battle Ground Academy
'22 University of Maryland
User avatar
Cheynem
Sin
Posts: 6786
Joined: Tue May 11, 2004 11:19 am
Location: Grand Rapids, Michigan

Re: Paired Tossups and Bonuses

Post by Cheynem »

I suspect bouncebacks are going to play better at opens and harder tournaments (college and high school).
Mike Cheyne
Formerly U of Minnesota

"You killed HSAPQ"--Matt Bollinger
User avatar
theMoMA
Forums Staff: Administrator
Posts: 5786
Joined: Mon Oct 23, 2006 2:00 am

Re: Paired Tossups and Bonuses

Post by theMoMA »

Bouncebacks add a significant amount of time to gameplay and potentially a significant amount of additional time when readers boot them and it ends up mattering.

I've noticed a few attempts in this thread and elsewhere to claim that bouncebacks are friendly to new teams because they avoid situations in which they're just sitting there while the other team answers a bonus. I'm dubious of any argument to place oneself in the mind of a new player. In any event, I think it's just as likely that getting a tossup and missing the bonuses while the other team picks off the parts fairly routinely on the bounceback is not a pleasant experience, and that the experience of the moderator asking "bounceback?" in the less-frequent (and more difficult) circumstances when the better team misses a bonus part will not make up for this, and may in fact be bewildering and alienating on its own due to the speed with which the bounceback proceeds. The bounceback adds more opportunities for the better team to get points, and I can't imagine how that would feel like a positive for a new team already getting pummeled.

All that aside, my main issue with bouncebacks is that they add a situation in which teams get points but didn't really do anything to deserve points. It's evident and intuitive why the team that buzzes with the correct answer first gets tossup points. It's intuitive that a thing called a "bonus" is then unlocked and allows the tossup-converting team to confer to get additional ("bonus") points. It is not at all evident and intuitive that the other team should have an opportunity to get "bonus" points in that situation. A bonus for what? There is no game action that the team that failed to convert the tossup has taken that you can trace those points back to. And this leads to absurd situations in which a team could get tossup 20 and the other team, though taking no positive action up to that point of the tossup/bonus cycle, can get bounceback points and win the game.
Andrew Hart
Minnesota alum
User avatar
Auroni
Auron
Posts: 3104
Joined: Thu Nov 15, 2007 6:23 pm
Location: Urbana

Re: Paired Tossups and Bonuses

Post by Auroni »

These are the primary arguments that I would put forth in support of bouncebacks (which, to be clear, I support implementing for a tournament “every now and then” as opposed to establishing as any sort of standard or new norm that college quizbowl should adopt):

1) This is a game produced and played by lovers of knowledge, and bouncebacks are a way to provide more opportunities for engagement with the questions, and therefore increase the number of moments in which one can demonstrate knowledge. The player that I think of is the one who is not necessarily a good generalist or whole-category specialist, but has a few pockets of knowledge in certain areas, or has niche interests that they love. A bonus comes up on those interests, but their team doesn’t control it, and they are left feeling as if one of their limited opportunities to shine/do well has passed them by. Beyond that, it’s a very common experience for people to talk about tournaments after playing them and express regret that a really interesting bonus came up, but the other team got to hear it, presumably missing at least one part the person speaking would have known. Bouncebacks would greatly enhance the experience of those players, which, quite frankly, we all either know or have (once) been ourselves.

2) There have been many discussions about how bonuses are comparatively not very important to quizbowl in a game sense, as the first team to answer 11 tossups wins the game in the vast majority of situations, while taking up a large chunk of the time associated with running (and producing) a tournament. A practical impact of this situation is the general lack of memorability of bonuses, which is lamentable because good writers of bonuses routinely put in a lot of effort in finding interesting connections and thinking about knowledge that would be difficult to test in tossup form. Bouncebacks represent a way to make bonuses more important strictly in a game sense, without tipping the scales in the other direction. The most important thing for a team looking to win a game at a tournament run on bouncebacks is still to answer the tossup.

Apart from those two arguments, which make the case for bouncebacks, here are some responses to common arguments made by detractors, which could serve as subsidiary arguments. I would agree that special accommodations would be needed to sustain a bounceback tournament, such as tight and concise writing, and that it is likely a no-go for an inexperienced tournament host or staffer corps. I fundamentally disagree with Andrew's notion that it is not "evident and intuitive" for an opposing team to have a chance to get the bonus points that the controlling team missed (it reminds me pretty readily of interceptions/steals in sports, a domain that I hear has many parallels to quizbowl) or that even if it were unintuitive, that that alone would make them not worth trying. Nonetheless, I am amenable to changing the name of the term to make more sense within the context of a bonus (perhaps "steal")?
Auroni Gupta
UIUC
ACF
User avatar
Zealots of Stockholm
Rikku
Posts: 477
Joined: Fri Jul 10, 2015 3:28 am
Location: Auburn, AL

Re: Paired Tossups and Bonuses

Post by Zealots of Stockholm »

jdpasspawn wrote: Tue Jun 16, 2020 11:27 am
I think you need very strict length control and bonus timing
So the latter point is going to be extremely difficult to uniformly enforce across an entire tournament unless every room has a nats caliber reader in it (which most local tournaments definitely do not have!). Bounceback timing is arguably the most difficult aspect of quizbowl reading to do (even for veteran readers) and in my experience there tends to be wildly different enforcings of the rule that lead to seemingly arbitrary rulings and massively extended tournament runtime.
If bouncebacks mean you have to whisper and develop more effective communication strategies with your team, well, deal with it. Effective communication is a lot of what being a team is about anyways, and you already have to be quiet during the other team's bonus as-is, otherwise you're being a dick (I'm certainly guilty of this on too many occasions). Bonuses are where the "team" element of the game comes into play the most, and as far as I'm concerned, enhancing that is a good thing.
So as someone who played the large majority of their HS tournaments with bouncebacks (shoutout TACA) this assertion is largely incorrect. Bouncebacks tended to v significantly disincentivize teammwork and discussion because of the (v reasonable) fear that the other team would overhear your conversation and use it to help steal your bonus. In addition, I really don’t think games should be coming down to how well your team can whisper.

Anyways, like I said in my previous post, I think it’s fine if some tournaments want to experiment with bouncebacks and see how it goes. New TDs however (especially those in burgeoning circuits/ circuits with inexperienced readers) should be v wary of trying to implement bouncebacks for their tournaments for the reasons I’ve laid out in my last two posts.
I, like Jack, played a ton of tournaments with bouncebacks in high school, and cosign this post. I personally never found them particularly fun.

Also, upsets are fun and good and decreasing them is bad imo.
Chandler West
Peabody College of Education and Human Development at Vanderbilt University '22
Auburn University '20
Good Hope High School (Cullman, AL) '16
Member, PACE
User avatar
UlyssesInvictus
Yuna
Posts: 827
Joined: Thu Feb 10, 2011 7:38 pm

Re: Paired Tossups and Bonuses

Post by UlyssesInvictus »

I've seen the term "steal" used a lot in local format shows, and it seems pretty intuitive there.

That said, the number of times I've seen a team zone out because they're getting destroyed, then completely not realize there's even a bounceback available when it finally comes around to them (and subsequently groan because they either still don't know it, or missed their opportunity to try getting it), suggests that it still doesn't really help new teams feel like they're participating more.

Honestly, anecdotal evidence, but the only time I've really seen players happy to get a bounceback is:

- They're very good, and just like answering things. (Frankly, in some annoying fashions.)
- It's an extremely close game, and a bounceback can swing the game / close it out.

Only the latter seems like a real pro for bouncebacks in terms of pure game enjoyment, and I don't know how often it happens, or whether the same tension would just be there anyways without bouncebacks.
Raynor Kuang
quizdb.org
Harvard 2017, TJHSST 2013
I wrote GRAPHIC and FILM
User avatar
ThisIsMyUsername
Yuna
Posts: 863
Joined: Wed Jul 15, 2009 11:36 am
Location: New York, NY

Re: Paired Tossups and Bonuses

Post by ThisIsMyUsername »

btressler wrote: Fri May 22, 2020 10:58 am
Mike Bentley wrote: Thu May 21, 2020 5:45 pm The hotter take is that bonuses end up making a difference in a very small number of games.
I've often wondered if anyone has analyzed a tossup/bonus tournament to see how the records change if only the tossup points were in play. Does anyone have data to share?
I'm late to this thread. But in as much as I've always been bothered by the "bonuses don't matter very much" sentiment, I wanted to give a little data to show why this is very much not so. Below is a list of games from the top brackets of the last ten national collegiate championships whose results would have been altered by eliminating bonuses. This list is not short at all. (I did this quickly, so there are almost certainly some mistakes or omissions. Apologies in advance for those.) Furthermore: in two or three cases, this changes what the finals situation would have been; in one case, the result of a tiebreaker could have changed the winner of the title; and in two cases, this automatically changes who would have won the title!

2019 Nats
Berkeley B would have beaten Chicago A in Round 10
Columbia would have beaten Minnesota A in Round 13

Had either of these things happened, Columbia would have cleared the field.

2018 Nats
Northwestern would have beaten Cambridge in Round 10
Chicago A and Penn A would have tied in Round 14

2017 Nats
Michigan and Oxford A would have tied in Round 6
Columbia and Chicago A would have tied in Round 7
Oxford A would have beaten Chicago A in Round 10
Michigan and Yale A would have tied in Round 14

Had Michigan lost either tiebreaker, Maryland would have had the advantage in the final.

2016 Nats
Yale A would have beaten Maryland A in Round 1
Illinois and Oklahoma would have tied in Round 3
Columbia A would have beaten WUSTL in Round 7
Berkeley A would have beaten Michigan A in Round 9
Louisville would have beaten Yale A in Round 10
MIT B would have beaten Yale A in Round 11
Columbia A would have beaten Yale A in Round 13
MIT A and Minnesota A would have tied in Round 13
Illinois would have beaten Oxford in Round 17

Michigan A would have had the disadvantage in the final.

2015 Nats
Chicago A would have beaten Penn A in Round 20
Maryland would have beaten Chicago A in Round 22

2019 ICT
Maryland A and MSU would have tied in Round 7
Berkeley A would have beaten Columbia A in Round 7
Amherst would have beaten OSU in Round 8
Minnesota A and Columbia A would have tied in Round 13
Columbia A and OSU would have tied in Round 14

If Columbia had won both tiebreakers, they would have won the whole tournament, because the first finals match would have been the only finals match.

2018 ICT
Ohio State would have beaten Yale in Round 7
Penn A would have beaten WUSTL Round 8
Chicago A would have beaten Michigan A in Round 7

Berkeley A would have won the tournament, because the first finals match would have been the only finals match.

2016 ICT
North Carolina would have beaten Chicago A in Round 2
Yale A would have beaten Columbia A in Round 9
MIT A would have beaten Yale A in Round 10

Chicago A could not have won the tournament.

2015 ICT
Stanford A would have beaten Chicago A in Round 7

EDIT: Making corrections as I notice errors
Last edited by ThisIsMyUsername on Wed Jun 17, 2020 3:16 pm, edited 2 times in total.
John Lawrence
Yale University '12
King's College London '13
University of Chicago '20

“I am not absentminded. It is the presence of mind that makes me unaware of everything else.” - G.K. Chesterton
User avatar
Mike Bentley
Sin
Posts: 6076
Joined: Fri Mar 31, 2006 11:03 pm
Location: Bellevue, WA
Contact:

Re: Paired Tossups and Bonuses

Post by Mike Bentley »

Doesn't this list just confirm that a very small number of games are affected by bonuses? In some tournaments as few as 1 or 2 games in the top bracket? Yes this affected who won some national titles. But much, much more impactful was pretty much every team converting that 11th or 12th tossup against their opponents.
Mike Bentley
VP of Editing, Partnership for Academic Competition Excellence
Adviser, Quizbowl Team at University of Washington
University of Maryland, Class of 2008
User avatar
shefna21
Lulu
Posts: 12
Joined: Thu Jan 03, 2019 9:42 am

Re: Paired Tossups and Bonuses

Post by shefna21 »

Yeah, I don't think anyone is trying to make the argument that bonuses never make a difference, just that the difference they make is fairly small, and most of the time not enough to swing the game. I'm willing to bet that you could find a similar number of games in nationals top brackets that were determined by who got tossup 7 of the game - and yet if reading tossup 7 singlehandedly took up 50% of the game time, it would be worth reconsidering how the game was structured. I love the communication aspect of bonuses, as well as the ability to ask about things in ways you couldn't in tossups, but as the game is currently structured I'm sure you could get a significantly more reliable measure of which team was better by reading 40 tossups rather than 20 tossups and 20 bonuses.
Nathan Sheffield
[Belmont High School, MA]

If this was real life I'd offer you a peppermint.
User avatar
vinteuil
Auron
Posts: 1455
Joined: Sun Oct 23, 2011 12:31 pm

Re: Paired Tossups and Bonuses

Post by vinteuil »

Mike Bentley wrote: Wed Jun 17, 2020 5:56 pm Doesn't this list just confirm that a very small number of games are affected by bonuses? In some tournaments as few as 1 or 2 games in the top bracket? Yes this affected who won some national titles. But much, much more impactful was pretty much every team converting that 11th or 12th tossup against their opponents.
I don't think John, or anybody, is arguing, that it's a large proportion of games whose outcome is affected by bonuses, but rather that bonuses matter disproportionately in close, impactful games.
Jacob Reed (he/him/his)
Chicago, Yale, East Chapel Hill
cwasims
Lulu
Posts: 70
Joined: Fri Mar 31, 2017 5:16 pm

Re: Paired Tossups and Bonuses

Post by cwasims »

One thing I'll add to this discussion: although PPB is a pretty reliable indicator of a team's skill across an entire tournament, the ~10 bonuses a team hears in a close game have a high variance associated with them that can only imperfectly advantage the team truly better at bonuses as opposed to the team that gets lucky with what categories/subcategories they get asked. Bonuses are quite variable pack-to-pack and bonus-to-bonus (especially at a Nationals level where 30ing a bonus can be a relatively major boost in points) and I am certain there must be cases where bonuses actually have caused upsets (as in a team with a lower PPB winning when they otherwise wouldn't have) as opposed to mitigating them.
Christopher Sims
University of Toronto 2T0
Northwestern University 2020 - ?
User avatar
ValenciaQBowl
Auron
Posts: 2442
Joined: Thu Feb 05, 2004 2:25 pm
Location: Orlando, Florida

Re: Paired Tossups and Bonuses

Post by ValenciaQBowl »

At Valencia practices, I almost always have "steals" as bonus parts in effect, as it's a good way to ensure players stay engaged and learn stuff when the other team is answering a bonus. Or sometimes, if one team is significantly better, I only allow the weaker team to steal. It doesn't take much longer because steals have to be answered pretty much immediately after the other team has missed a part; I don't give them five seconds to deliberate. I don't know if that's an option in actual tournaments, and I've never been involved in an official competition in which "bouncebacks" were allowed, so I don't know if it's standard practice to give the other team a new clock to confer.

One question I'd have regarding the argument that bouncebacks can help keep a team getting whupped to stay engaged is how often will the weaker team know a part that the stronger team doesn't know? I can see that in trash or if the other team has a specialist in the bonus's subject matter, but I imagine that in the vast majority of cases the weaker team won't know it either.
Chris Borglum
Valencia College Grand Poobah
User avatar
UlyssesInvictus
Yuna
Posts: 827
Joined: Thu Feb 10, 2011 7:38 pm

Re: Paired Tossups and Bonuses

Post by UlyssesInvictus »

ValenciaQBowl wrote: Thu Jun 18, 2020 11:31 am It doesn't take much longer because steals have to be answered pretty much immediately after the other team has missed a part; I don't give them five seconds to deliberate. I don't know if that's an option in actual tournaments, and I've never been involved in an official competition in which "bouncebacks" were allowed, so I don't know if it's standard practice to give the other team a new clock to confer.
Last I recall, it's only two seconds to answer a bounceback at NSC, vs. five for the main bonus.

In my experience, usually the team gets more time, though.
Raynor Kuang
quizdb.org
Harvard 2017, TJHSST 2013
I wrote GRAPHIC and FILM
User avatar
Stained Diviner
Auron
Posts: 4830
Joined: Sun Jun 13, 2004 6:08 am
Location: Chicagoland
Contact:

Re: Paired Tossups and Bonuses

Post by Stained Diviner »

Several of John's examples are cases where the teams answered the same number of tossups, and the bonus conversion turned out to be more important than the neg penalties. That list includes 2016 Yale A Maryland A, 2016 Berkeley A Michigan A, 2016 Yale A Columbia A, 2016 Illinois Oxford, 2017 Oxford A Chicago A, 2018 Northwestern Cambridge, and 2019 Berkeley B Chicago A.
In 2016 Yale A Louisville, Yale answered more tossups but was behind on tossup points because of negs.

I think that leaves three cases in the last four ACF Nats involving top bracket teams where the team that answered fewer tossups won the match because of superior bonus conversion. The list basically shows that being good at bonuses can help you overcome a couple of neg penalties, but it can't help you overcome answering fewer tossups except for a very small number of cases.
David Reinstein
PACE VP of Outreach, Head Writer and Editor for Scobol Solo and Masonics (Illinois), TD for New Trier Scobol Solo and New Trier Varsity, Writer for NAQT (2011-2017), IHSSBCA Board Member, IHSSBCA Chair (2004-2014), PACE Member, PACE President (2016-2018), New Trier Coach (1994-2011)
Post Reply