## Player Poll 2018: Delightfully Devilish

Elaborate on the merits of specific tournaments or have general theoretical discussion here.
ryanrosenberg
Auron
Posts: 1593
Joined: Thu May 05, 2011 5:48 pm
Location: Palo Alto, California

### Re: Player Poll 2018: Delightfully Devilish

John helped me find a bug in my PATH calculations for CMST; the above sheet is incorrect.

The sheet below has PATH by subject, as a weighted sum of subject PATH (History, Lit, Science, RMPSS x4, Arts x3, Other x1), and a "standard" PATH calculation, as well as the ratio between subject-weighted PATH and standard PATH. I excluded powers from this analysis.

The ratio seems like an interesting quantification of specialization.
Attachments
player_points_xlsx.xlsx
Ryan Rosenberg
Webmaster, ACF | Member, PACE
Yuna
Posts: 880
Joined: Wed Jul 15, 2009 11:36 am
Location: New York, NY

### Re: Player Poll 2018: Delightfully Devilish

Obviously, I am not sympathetic to Andrew Hart's philosophy on what the player poll is meant to achieve. For those who are, though, I would be wary of measuring it using a statistic whose supposed soundness was entirely justified by this paragraph:
Andrew Hart in 2015, when inventing this statistic wrote: My next guess was that squaring both team overall PPG and top team overall PPG would possibly yield more interesting results, since the top teams would be further separated from the weaker teams. (I got the idea of squaring the numerator and denominator from baseball's Pythagorean theorem win percentage formula.) The results of this were indeed interesting, because they jibed very closely with how I might subjectively rank the players. [emphases mine]
If one wished to approach player rankings from a broadly Hartian worldview (but filter this through a less arbitrary statistic), in which all that matters is winning games and contributing to those victories, then two things that we currently account for are irrelevant:
(1) How many tossups a player racks up in a game that they do not win. Except for the effect on PPG in breaking ties, there is no effective difference in losing by 10 vs. losing by 300.
(2) How many tossups one gets past the point at which victory is assured for one's team. Once the game is clinched, no further action performed by anyone matters in the least in terms of the tournament standings.

If one wished to remove these considerations, one could so in one of two ways:

The rawer way is this: for Team X, look at only matches where they win; cut off each match at the tossup where victory was assured for them (this requires the actual scoresheets, or Ophir Stats); and count only the tossups up to that point, and none of the ones after that. This makes no attempt to account for which questions in the round happened to come up before the clinch-point. (e.g. If we clinch the game at Tossup 15, and I get all of my points in the last quarter, those point are wiped from my record, even though I would have gotten them had they been randomized earlier in the packet.) Rather, it assumes that all of these effects of randomization will iron themselves out over time.

Although it goes against the spirit of "all that matters is what literally caused you to win a match," one could adjust for the randomization issue as follows: As before, for Team X, look at only matches where they win. For each match, determine (given the bonus conversion of the team) how many of those tossups they actually needed to get to win (the clinch tossup number, rather than the clinch tossup point). Divide the clinch tossup number by the actual number of tossups the team got, and multiply each player's score for that round by that fraction. This approximates how many points each player contributed towards victory in that round. Sum all of the points you get from matches that were won, and you determine each player's number of "victory points."

As an example of this latter method, let's say that in a particular match, Players A, B, C, and D got 5, 4, 3, and 2 tossups respectively, but they needed only 11 of the 14 tossups to clinch. One could multiply each of these values times 11/14, getting values of 3.93, 3.14, 2.36, and 1.57 respectively. In a match where Players A, B, and C got 5, 4, and 2 tossups respectively, and Player D didn't buzz, those buzzes would count for the full values of 5, 4, and 2, because they were all fully essential in attaining victory.

Please note that this assumes that you can assess how responsible an individual player is for their team winning a match, but not how responsible an individual player is for their team losing a match. In some intuitive sense, this is true: we know who is responsible for what they did, and not who is responsible for what anyone could have done; had any member of your team gotten X tossups more, your team would have won. If this sits badly, it is probably because many of us have the sense that the player who performed most below their mean (or below their expected role) is the one most responsible for a loss. If one wanted to account for that, one would have to add an additional component that calculates this.

EDIT: Corrected one arithmetic error
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
jonah
Auron
Posts: 2340
Joined: Thu Jul 20, 2006 5:51 pm
Location: Chicago

### Re: Player Poll 2018: Delightfully Devilish

Is this helpful for ICT PATH? Note the "Phase" and "Round" filters. I'm sorry if I did this wrong; I've never computed PATH before.
Jonah Greenthal
theMoMA
Posts: 5790
Joined: Mon Oct 23, 2006 2:00 am

### Re: Player Poll 2018: Delightfully Devilish

I don't really understand how you're getting all that from what I'm saying. My point is not that a team that scores the most tossups necessarily won all of its games, or the tournament, or anything, really. Rather, it's that the team that scores the most tossups stands the best chance of winning a particular game and a particular tournament. This isn't always the case, but historical data show that it's generally a sound assumption. I particularly don't understand the issue of eliminating stats in losses. If a player tends to score about 5 tossups a game on a team that averages 11 tossups a game, and that player's teammates happen to have a bad game despite the player in question scoring the customary 5 tossups, the player has still given the team a chance to win. I also think it would be foolish to eliminate common-opponent data simply because a team didn't happen to win a game, because data in losses is equally valuable to determining how good players are at giving their team a chance to win (or a chance not to lose, when looked at another way).

To me, it makes the most sense to work backward from the goal of quizbowl (winning) to the individual's contribution to winning. Several inferences and assumptions have to be made along the way. The assumptions I make are that winning is mostly based on team tossups converted, and that player contribution to team tossups converted is roughly equal in value, regardless of the kinds of points that players score. These assumptions might be good or bad, and they might work well or collapse under the weight of evidence.

I happen to think that they're good assumptions because they have a foundation in team performance, and because of that, all that the stat cares about is whether a team is elite at getting tossups (which is an objective fact--either your team averages an elite number of tossups per game in the playoffs of a national tournament, or it doesn't). So, as I said before, if you're a team getting 10-12 tossups per game on early buzzes or late buzzes or (most likely) a mixture of both, it's just a fact that your team is elite at getting tossups, and that historically, teams with such elite rates tend to win a lot. You don't have to agonize over whether a player's typical buzz is "good" or "bad" in some subjective way if you simply start with the fact of what happened and work backward from there.

But if you take the opposite approach, and attempt to build up a team's chances of winning from the players' individual talents, you are also forced to make assumptions. You might, for instance, assume that a player's ability to buzz early in a smaller subset of categories is more important than a player's ability to buzz late across wide swaths of the distribution; you might assume the opposite. Regardless, I suspect that this method is not as sound, because it doesn't have the benefit of starting with a measure that says almost everything you need to know about how good a team is, i.e. how many tossup points a team scores relative to the competition. A building-up approach might have more to say about which player would be a particularly good fit on a hypothetical future team, but as I said, I don't really see the point of that when I understand the poll to be asking about "who did best," as opposed to "who is best."
Andrew Hart
Minnesota alum
Sam
Rikku
Posts: 293
Joined: Sat Nov 07, 2009 2:35 am

### Re: Player Poll 2018: Delightfully Devilish

Given quiz bowl's newfound love of cooperative game theory, something that may be close to you what Andrew is describing is the Shapley value. (No link because I'm on a phone, but it's on Wikipedia.) The idea is to consider all the possible coalitions that can be formed without a given player, seeing what the "value" of each of those coalitions is, and then seeing the value of each coalition when you add that player.

For example, suppose there's a two person team that wins 10 games. Suppose player A alone would have won 5, and player B alone would have won 7. A team with no players would presumably win 0. Then, the marginal contribution of A when they join the empty team is 5, and the marginal when they join B is 3. For B, the contributions are 7 and 5, respectively. Player A's value would then be 4, and B's would be 6.

Note that the number of games won for the hypothetical coalitions is itself hypothetical. You could make a crude guess based on total tossups, similar to what John does above, but even if you don't spit out an actual number, the formulation may be a helpful way to think about players' contributions to their team.

EDIT: When I run a crude version of this for the Minnesota team (where each tossup point is multiplied by 1+(total bonus/total tossup), and I assume the other team picks up nothing), I get values of 2.5, 3, 2.583333... and 2.08333... for Jason, me, Shan, and John, so it's even more Hartian than I originally thought.
EDIT2: Nope, just did math wrong.
Sam Bailey
Minnesota 'xx
Chicago '13
Yuna
Posts: 880
Joined: Wed Jul 15, 2009 11:36 am
Location: New York, NY

### Re: Player Poll 2018: Delightfully Devilish

theMoMA wrote:I don't really understand how you're getting all that from what I'm saying.
And I don't really understand who you're arguing against in your most recent post, or how your reasoning holds together. Let's take this piece by piece.
My point is not that a team that scores the most tossups necessarily won all of its games, or the tournament, or anything, really.
I never suggested that this was your point. We agree on this simple fact.
Rather, it's that the team that scores the most tossups stands the best chance of winning a particular game and a particular tournament.
I agree with this, but it is impossible to square this with your claim that we should care only about what players did and not what they hypothetically could have done. If you care about who "stands the best chance of winning a particular game and a particular tournament," you care about who was most likely to win games/tournaments that they didn't actually win. That's a hypothetical.

This erodes your claim that your approach is more grounded in past realities than PATH is. Asking "How much should we expect Player X to score in a given match?" is just as forward-projecting as asking "How much would Player X score if Teammate Y weren't there?".
[...] determining how good players are at giving their team a chance to win
Why is this on the "performance" side of your dichotomy rather than the "talent" side? How good someone is at X is the definition of talent! The only way to reformulate this so that this is an actual measure of "performance" would be to ask: "How much did Player X do to increase their team's chances at winning this specific tournament?" (I'll address again soon why I don't think your approach answers that properly.)
The assumptions I make are that winning is mostly based on team tossups converted, and that player contribution to team tossups converted is roughly equal in value, regardless of the kinds of points that players score.
You don't have to agonize over whether a player's typical buzz is "good" or "bad" in some subjective way if you simply start with the fact of what happened and work backward from there.
You might, for instance, assume that a player's ability to buzz early in a smaller subset of categories is more important than a player's ability to buzz late across wide swaths of the distribution; you might assume the opposite. Regardless, I suspect that this method is not as sound, because it doesn't have the benefit of starting with a measure that says almost everything you need to know about how good a team is, i.e. how many tossup points a team scores relative to the competition.
I not sure that anyone in this thread has made either of these two assumptions, at least not since we've moved beyond anecdotes of teammates' accomplishments to discussing stats. Neither of the assumptions that you're criticizing is entailed by raw PPG, PATH, or any other manipulation of that data that has been proposed thus far. These seem like discussion tropes from a couple of years ago. Your claim that the advocates of so-called "talent" measures must necessarily be relying on these considerations is unwarranted.
To me, it makes the most sense to work backward from the goal of quizbowl (winning) to the individual's contribution to winning.
Yes, and this is what I did in my last post. I took the matches that the players won, and suggested a way to measure how much each player contributed to that win.

Let us say instead, though, that what you meant is not an individual's "contribution to winning" (as in, actual victories) but rather an individual's "contribution to improving their team's chances of winning" (which seems more like what you were trying to say, in which case I took you too literally before). Then indeed, you do not want to remove from consideration the games that they lost.

You still have not engaged with my primary objection, though. (And it is not the first time that I've brought this up in forum discussions with you.) It is a simple fact that after a team has clinched the game, no further tossups continue to increase a team's chances of winning. They don't affect the outcome of the game, the tournament, or anything beyond the buzzing player's own personal stats. So, why do you care about these at all, given your stated position? This is data about something you dismiss caring about.

I think the only way to measure an individual's "contribution to improving the team's chances of winning" would be to employ a modification of what I proposed in my previous post. Take each match (not just winning matches this time), and stop counting the stats for both sides after the point that one team has clinched the game. (Again, this requires scoresheets or Ophir Stats.) Nothing anyone does from that point forward contributes to either team's chance of winning.

As I said before, this might seem unfair to the people who happened to buzz later in the game (e.g. I get no buzzes before we clinch the game, but get three buzzes afterwards, which now get wiped). One might think that that these buzzes should count just as much as the others, even though they didn't actually matter, because they "would have mattered" were the packet order altered. I gave what I think is a way of correcting for that. But I assume that you wouldn't raise that objection in the first place, because you disavow interest in counterfactuals, rather than what actually happened (to quote you: "players actually did get tossups and failed to get others"); and allowing for packet order to have been other than it was is engaging in counterfactuals.

I should add again that I do not endorse this method as a determiner of player poll ranking, at all. I'd much prefer that we take the PATH path. But I do think this is the more logical way to gauge how much a player's performance contributed to victories, if someone cares primarily about that.

Also, I continue not to understand your reasoning for thinking that your statistic from three years ago yields accurate/useful rather than "interesting" results. Your initial post made it sound like you came to it through guesswork, and then called it a day when it produced results that you liked. (I'm not trying to be uncharitable. I just can't find another way to interpret the paragraph I quoted before.) Why do you think your statistic works?
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
Mnemosyne
Wakka
Posts: 174
Joined: Tue May 31, 2011 12:11 am

### Re: Player Poll 2018: Delightfully Devilish

Geriatric trauma wrote:I put together a rough tier list of generalists and specialists, since I find it very hard to decide between the two groups when making my ballot. In general, I tried to avoid overlap as much as possible, except where it was unavoidable or instructive to prove a point (i.e. Kai/Alston/Jaimie have all demonstrate excellent generalist potential without John/Eric, Clark Smith can develop along either track, Jakob Myers is well on his way to the top). For the "Noteworthy" section, I tried to (on the most part) include people who I feel are underdiscussed or underappreciated, rather than making it a strict fourth tier, since making an exhaustive list of that would stretch too wide for this discussion.

So, Ryan made two good posts in this thread, one with a subjective list of who people think are the best generalists/specialists, and another, objective one, where he posted the top buzz numbers across all sites of CMST. I was interested in checking the subjective view of good specialists with the objective one.

Below, I have taken the names Ryan has mentioned above and some of the names mentioned in recent threads, the CMST raw buzz data across all sites, and I've added two columns for percent of buzzes that were top 3 / top 5. I also left the names of some "famous" people for reference. Obviously this leaves out many people, but it's an easy way to quickly compare the names listed above. Also, of course I sorted by number of Top 3 buzzes, since that's what puts me the highest because it's halfway between 1 and 5. Green indicates that the player is eligible to be voted for (as far as I know).

Maybe this data only indicates early buzz ability, and "specialist" should be defined as having a high percentage of these early buzzes?

Nick Collins
C. E. Byrd '12 (Shreveport, LA)
Louisiana Tech '16, '17
University of Virginia
theMoMA
Posts: 5790
Joined: Mon Oct 23, 2006 2:00 am

### Re: Player Poll 2018: Delightfully Devilish

John, I think we're basically talking past each other. My claim is that, if you look at past Nationals winners, you won't see a ton of similarities except for the fact that they average something like 11 tossups per game in the playoffs. Generally speaking, more of these tossups come against teams that finish toward the bottom of the bracket, as you'd expect, because those teams are less good at converting tossups. This tends to happen every year, where championship teams have very similar tossup conversion numbers, despite various championship teams taking different routes to scoring those tossups. So my assumption is basically that a team scoring around 11 tossups per game across the playoffs, regardless of their method of scoring those tossups (i.e. by very good specialist knowledge, or by solid generalism across the board, etc.), is going to be in contention to win the tournament, because almost every past team that has that kind of tossup scoring prowess has at least been in strong contention.

My position is not that each and every tossup that a player scored contributes equally to that team's chances of winning the tournament, but rather, that a contending team will end up with a certain scoring profile (about 11 tossups a game). And my assumption is that you can take those tossups, divide them out among the team members based on scoring, and arrive at a rough idea of how valuable a player was to the team's efforts. Again, this might not be a good assumption for all sorts of reasons, but I think this approach has a very sound basis, in that it's grounded in a very historically strong indicator of team performance (tossups converted) and assigns individual credit that team performance down intuitively (by simply dividing it up by how much each player contributed).

In other words, assumption #1 is that a team that scores around 11 tossups a game will have a good chance to win the tournament, or most basically, championship teams may look very different in terms of the kinds of buzzes they get or the kinds of questions they specialize in, but they look very similar in terms of one thing, which is how many tossups they get per game. Assumption #2 is that each player should receive credit for tossups converted based on how many tossups they convert. I submit that these are good assumptions because historically, teams that have converted the most tossups are almost always in contention, and this is true regardless of the various ways that teams have scored tossups.
Andrew Hart
Minnesota alum
Yuna
Posts: 880
Joined: Wed Jul 15, 2009 11:36 am
Location: New York, NY

### Re: Player Poll 2018: Delightfully Devilish

I wrote: Whether Jonah wants to do this or someone else wants to instead: If we're looking to restrict ourselves to data involving only common opponents, I think the most logical place to start is to run PATH for the five matches when the six superplayoffs top-bracket teams (Yale A, Berkeley A, Chicago A, Penn A, Columbia, and Michigan A) played each other, and then also run PATH for five matches when the six superplayoffs second-bracket teams (Ohio State, Berkeley B, Maryland A, WUSTL A, McGill A, and NYU A) played each other. To be clear, these would be two separate sets of data, which we would not combine. But it allows us to rank within these brackets.
Okay, probably my last post presenting compiled/calculated stats:

Thanks to Jonah's spreadsheet, I was able to do this. The data was already all there within the spreadsheet, but it was in a lot of different places, where it couldn't be viewed at once. For every team that makes (e.g.) the top-bracket superplayoffs, three of their five matches against common opponents were in rounds 12-14 (the three superplayoff matches). However, the two remaining matches against common opponents could have been in any round of the playoffs. For example, Yale A's matches against Michigan A and Chicago A were rounds 10 and 11, but Columbia's matches against Penn A and Berkeley A were rounds 7 and 9. So, the only way to view common-opponent stats for these teams is to manually turn on the PATHs for just the relevant rounds, compile them, and compare them. This is what I've done in the spreadsheet below.

The results should be particularly interesting to anyone who had the slightest doubt about just how damn good Isaac's performance was at this tournament. (If such people exist.)
Attachments
2018 ICT PATH_Common Opponents.xlsx
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
Wakka
Posts: 168
Joined: Sat Nov 12, 2011 5:09 pm

### Re: Player Poll 2018: Delightfully Devilish

OK, hear me out. Timothy Guice.

Like, he may not be a top-25 talent, but think about how much more it would mean for him than for Eric Mukherjee.
Mayor of quiz bowl at the University of Central Florida (2010-2015)
The club at Reformed Theology Seminary Orlando (2017 - 2055??)
Cheynem
Sin
Posts: 6809
Joined: Tue May 11, 2004 11:19 am
Location: Grand Rapids, Michigan

### Re: Player Poll 2018: Delightfully Devilish

There's a lot of stats talk which is neat, but not many ballots. Let's shoot for a May 7th deadline.
Mike Cheyne
Formerly U of Minnesota

"You killed HSAPQ"--Matt Bollinger
aseem.keyal
Wakka
Posts: 103
Joined: Sun Jan 13, 2013 2:01 pm

### Re: Player Poll 2018: Delightfully Devilish

Really late, but for those who haven't submitted yet, consider voting for Eric Chen in the last couple spots of your ballot.
Aseem Keyal
Westview HS 2014
UC Berkeley 2018
Majin Buu Roi
Wakka
Posts: 141
Joined: Sat May 17, 2014 1:52 pm

### Re: Player Poll 2018: Delightfully Devilish

Like Aseem, I'm also super-late to this, but I definitely thought that this unnecessarily rigorous discussion could use more good old-fashioned "you should vote for my teammates" posts. So if you're still filling out your ballot, you should definitely vote for my teammates. Like, all of them, ideally, but that wouldn't make much realistic sense, sadly, even for me. But you should definitely strongly consider voting for at least one of the people who have regularly played for the Cambridge A squad throughout the year. Presented in no particular order:

Ewan "How Has He Not Been Ranked Already?" MacAulay
A brutal loss to absorb, but played for a significant portion of the year so I think he should be eligible. He's been mentioned as such before, but it bears reiterating: Ewan is a science monster on levels I've only ever seen Eric Mukherjee match/surpass in general and Stephen Eltinge match on a science sub-distro 1/1. On chemistry, there's no point in even trying to buzz while he's playing (unless you want to deliberately neg him out of it, which is a ton of fun), and it's a surprise when he doesn't 30 a chem bonus. On science in general, reference the EFT advanced stats for how thoroughly he outclasses the rest of the British field. Can also buzz (early) on certain literature and history topics and has best-in-quiz-bowl-knowledge of trains. Seriously, the man is obsessed with trains. And drinking. And diet coke. And is just mentally a five year old who enjoys sticking his finger in peoples' mouths when they yawn.

Yanbo "Actually Got a Visa Specifically to Play ACF Nats Because He's Technically a Chinese Citizen" Yin
Only in a club that includes Ewan would it make any remote sense to relegate Yanbo to a "replacement" science player, because Yanbo is ridiculous at science. As in "beat Eltinge to not one but TWO physics-y tossups in our Nats match" good at science, especially physics. Humorously did not get a Nats tossup on something he specifically discussed in his interview to get into Cambridge, though. Also unbeatable on anything written for or transcribed for the viola, a subject unfortunately lacking at Nats. On top of all that, Yanbo is also known to randomly (though most commonly in science and ancient history) buzz too early for anybody to do anything about it, an experience we at Cambridge refer to as "getting Yanboned." At Nats was deeply inspired by Eric Mukherjee. Not to study or learn anything, mind you, but just to expand his already comical full-body buzzing/fulminating technique to further, ludicrous, Mukherjee-an heights.

Elysia "Have I Mentioned Linguistics Olympiad Today?" Warner
Ellie is arguably the most impressive and easiest-to-underrate player at Cambridge. While basically everyone else on a given Cambridge A iteration has at least one sub-distro on which they're the only person really liable to buzz, Ellie buzzes entirely in categories of which at least one of her other teammates has significant knowledge, thus creating an unusually drastic shadow effect that still results in, oh, second-scoring at Penn Bowl, 20+ ppg at Regionals while playing with me, Ewan, and another lit player, and 20 ppg at Nats in the prelims and even mid-teens scoring with the rest of the Nats squad on the tournament as a whole. Excellent at literature in general (world literature in particular), one of the best other arts players in the game (particularly devastating on dance, and also very good at arts in general), and will not be beaten to linguistics (also very good at RMPSS in general as well). Easily serves as humanities generalist in a pinch. Just actually knows and reads things instead of memorizing clues and performatively pretending to have done the former. Afraid of my high-fives for some reason.

Ephraim "Ephraim Jacob Jacobus Levinson" Levinson
There is nothing more to say than what I posted in the Players' Choice Awards thread:
While I'm posting in this thread I'd also like to nominate the inimitable Ephraim Jacob Jacobus Levinson for 4th Scorer of the year. American quiz bowl has not quite gotten to experience Ephraim's non-sensical poetry/philosophy/literary theory/literary theory-adjacent things dominance first-hand, but let's just say that it makes him the quiz bowl equivalent of the Dos Equis guy: he doesn't always buzz, but when he does, it's on the first bloody clue. His Penn Bowl stat-line of 11/5/1 in ten rounds (it should in fact have been 12/5/0; he negged with an answer that was added to the answer-line in the version of the set uploaded to the database!) might be the most obscene thing I've ever seen in quiz and should by itself strongly suggest him for the 4th scorer title.
Joseph "The Krollercoaster" "The Steamkroller" Krol
Though he can serve as a science player in a pinch, even at CMST level, the Steamkroller mainly brings real math knowledge to the table as well the knowledge of current events and contemporary politics necessary to be one of the remaining active Lib Dems. In addition to just knowing enough random humanities/geopolitical things to serve as a backup humanities generalist, the Krollercoaster is also probably the best visual arts player in the UK (and beat ASEEM to painting in our match with Berkeley A) and absolutely slaughtered the Nats visual arts (along with Ellie; I barely buzzed on any visual arts thanks to how thoroughly they wrecked that content). Has rounds where he just single-handedly steamkrols the likes of Chicago A (at OOT, for example).

Sam "Something About Glaciers And/Or Middle-Earth!" Cook
Redefines "geobeast" to the point where the A-team gets built around him when the distro has at least 1/1 guaranteed geography per packet, as at BSQC. Also the best Byzantine/Medieval Europe player in the UK, except perhaps for Oxford's Alex Peplow (someone you folks should really watch out for next year, incidentally). Also really good at earth science, what with doing a PhD in it.
Jason Golfinos
Trinity School '13 (inexplicably in charge, 2011-13)
Princeton '17 (inexplicably in charge, 2015-16)
Cambridge '18
HLS '22
Cheynem
Sin
Posts: 6809
Joined: Tue May 11, 2004 11:19 am
Location: Grand Rapids, Michigan

### Re: Player Poll 2018: Delightfully Devilish

Results going to be posted sometime this afternoon or evening, so get ballots in ASAP.
Mike Cheyne
Formerly U of Minnesota

"You killed HSAPQ"--Matt Bollinger
Cheynem
Sin
Posts: 6809
Joined: Tue May 11, 2004 11:19 am
Location: Grand Rapids, Michigan

### Re: Player Poll 2018: Delightfully Devilish

Here are the final results. There were 21 ballots and I'll post voter names (but not ballots later):

1. Jacob Reed, Yale (518, 16 1st place votes, lowest 3rd)
2. Eric Mukherjee, Penn (496, 5 1st place votes, lowest 4th)
3. John Lawrence, Chicago (493, highest--2nd, lowest--5th)
4. Rafael Krichevsky, Columbia (439, highest--3rd, lowest--10th)
5. Chris Ray, Ohio State (430, highest--4th, lowest--8th)
6. Aseem Keyal, Berkeley (423, highest--2nd, lowest--12th)
7. Itamar Naveh-Benjamin, Missouri (411, highest--4th, lowest--11th)
8. Jason Golfinos, Cambridge (382, highest--6th, lowest--11th)
9. Adam Silverman, Northwestern (341, highest--7th, lowest--14th)
10. Shan Kothari, Minnesota (320, highest--7th, lowest 24th)
11. Jakob Myers, Michigan State (276, highest--7th)
12. TIE--Kenji Shimizu, Michigan (264, highest--9th, lowest--19th) and Derek So, McGill (264 highest--10th, lowest--21st)
14. Caleb Kendrick, Oklahoma (248, highest--11th, lowest--18th)
15. Charles Hang, WUSTL (233, highest--11th)
16. Isaac Kirk-Davidoff, Yale (194, highest--10th)
17. Daoud Jackson, Oxford (146, highest--10th)
18. Bruce Lou, Berkeley (145, highest--13th)
19. Clark Smith, Ohio State (130, highest--14th)
20. Weijia Cheng, Maryland (92, highest--15th)
21. Stephen Eltinge, Yale (90, highest--11th)
22. Nathan Weiser, Stanford (77, highest--13th)
23. Alston Boyd, Chicago (74, highest--16th)
24. Taylor Harvey, Florida (66, highest--15th)
25. Charles Dees, Columbia (59, highest--19th)

Sam Bailey (46), Kai Smith (25), Jaimie Carlson (21), Eric Xu (21), Evan Lynch (20), Matthew Lehmann (18), Rahul Keyal (14), JinAh Kim (10), Nikhil Desai (9), Ewan MacAulay (7) Aidan Mehigan (6), Greg Peterson (3), Elysia Warner (3), Ben Zhang (3), Jason Cheng (2), Jason Zhou (2), Eric Chen (1)

Voters: Jordan Brownstein, Itamar Naveh-Benjamin, Jason Golfinos, Joey Goldman, Aseem Keyal, Eric Mukherjee, Leo Lo, Charles Hang, Ryan Rosenberg, Will Alston, Alston Boyd, Kai Smith, John Lawrence, Taylor Harvey, Jacob Reed, Nick Collins, Joelle Smart, Andrew Hart, Adam Sperber, Jakob Myers, and John Marvin
Mike Cheyne
Formerly U of Minnesota

"You killed HSAPQ"--Matt Bollinger
vinteuil
Auron
Posts: 1455
Joined: Sun Oct 23, 2011 12:31 pm

### Re: Player Poll 2018: Delightfully Devilish

Probably the last year I can reasonably fit these all in one window:

EDIT: Also, I don't think the right side of this will ever be matched:
JR
ryanrosenberg
Auron
Posts: 1593
Joined: Thu May 05, 2011 5:48 pm
Location: Palo Alto, California

### Re: Player Poll 2018: Delightfully Devilish

vinteuil wrote:Probably the last year I can reasonably fit these all in one window:

22/25 players in this year's poll matched or tied their previous best ranking, the highest number ever in a poll.
Ryan Rosenberg
Webmaster, ACF | Member, PACE
The Stately Rhododendron
Rikku
Posts: 470
Joined: Sun Oct 16, 2011 7:18 pm
Location: Heart's in the woods

### Re: Player Poll 2018: Delightfully Devilish

Finally, justice.
IKD
Yale 18
Oakland Mills 14
"I am the NAQT beast I worship."
vinteuil
Auron
Posts: 1455
Joined: Sun Oct 23, 2011 12:31 pm

### Re: Player Poll 2018: Delightfully Devilish

The Stately Rhododendron wrote:Finally, justice.
stephen's still too low you fool
JR
ElysiaJW
Kimahri
Posts: 6
Joined: Sat Dec 05, 2015 6:14 pm

### Re: Player Poll 2018: Delightfully Devilish

Cheynem wrote:Here are the final results... Elysia Warner (3)
I'd like to thank the academy, my parents, and of course God-finos who is definitely behind this
Elysia "Ellie" Warner
Cambridge 2014-18
President Cambridge University Quiz Society 2016-17
theMoMA
Posts: 5790
Joined: Mon Oct 23, 2006 2:00 am

### Re: Player Poll 2018: Delightfully Devilish

This historical poll data is very cool, even if it makes me mad about 2011 again.
Andrew Hart
Minnesota alum
naan/steak-holding toll
Auron
Posts: 2285
Joined: Mon Feb 28, 2011 11:53 pm
Location: New York, NY

### Re: Player Poll 2018: Delightfully Devilish

vinteuil wrote: EDIT: Also, I don't think the right side of this will ever be matched:
you fool
you foolish fool
chris ray is still in school
Will Alston
Bethesda Chevy Chase HS '12, Dartmouth '16, Columbia Business School '21
NAQT Writer and Subject Editor
Sen. Estes Kefauver (D-TN)
Chairman of Anti-Music Mafia Committee
Posts: 5640
Joined: Wed Jul 26, 2006 11:46 pm
Location: Columbia, MO

### Re: Player Poll 2018: Delightfully Devilish

4. Rafael Krichevsky, Columbia (lowest--10th)
Is the word "lulz," or is the word "clueless about what actually goes into evaluating the kind of talent that helps people excel at a national championship?"
Last edited by Sen. Estes Kefauver (D-TN) on Mon May 07, 2018 8:17 pm, edited 1 time in total.
Charlie Dees, North Kansas City HS '08
"I won't say more because I know some of you parse everything I say." - Jeremy Gibbs

"At one TJ tournament the neg prize was the Hampshire College ultimate frisbee team (nude) calender featuring one Evan Silberman. In retrospect that could have been a disaster." - Harry White
Sen. Estes Kefauver (D-TN)
Chairman of Anti-Music Mafia Committee
Posts: 5640
Joined: Wed Jul 26, 2006 11:46 pm
Location: Columbia, MO

### Re: Player Poll 2018: Delightfully Devilish

Not everyone needs to vote in the poll if they don't really understand what they're talking about, it's good and healthy to cultivate other interests with your freetime. Also, it's the bottom file that's supposed to be read first, unsure how to embed it in my post.
Attachments
20180507_200145.png
20180507_194522.png
Charlie Dees, North Kansas City HS '08
"I won't say more because I know some of you parse everything I say." - Jeremy Gibbs

"At one TJ tournament the neg prize was the Hampshire College ultimate frisbee team (nude) calender featuring one Evan Silberman. In retrospect that could have been a disaster." - Harry White
Cheynem
Sin
Posts: 6809
Joined: Tue May 11, 2004 11:19 am
Location: Grand Rapids, Michigan

### Re: Player Poll 2018: Delightfully Devilish

I've always found most of the attendant discussion and the voting in these polls to be pretty interesting. I don't know how "useful" this is per se, but I don't know if "useful" is the good metric to use. I enjoy the game of quizbowl a great deal and enjoy talking about the game, and apparently a number of other people do as well.
Mike Cheyne
Formerly U of Minnesota

"You killed HSAPQ"--Matt Bollinger
gimmedatguudsuccrose
Wakka
Posts: 163
Joined: Thu Jun 13, 2013 7:17 pm
Location: Chicago, IL

### Re: Player Poll 2018: Delightfully Devilish

When I set about to construct my ballot for the player poll, the main criteria that I looked at were statistics from the top bracket of ICT and ACF Nationals (at least for those players who did make the bracket). However, it seems that many people were not using this as a basis for their poll, judging by some of the more "eccentric" choices listed above - for instance, I'm referring to decisions like not having Jacob Reed as No.1, putting John Lawrence at No.5, and putting Raphael at No.10. I'm not looking to call anyone out here; rather, I'm simply curious as to what criteria were used for these ranking decisions. Are people putting more weight into statistics from the regular season, or are people weighting certain categories (ie, science) more heavily than others? If anyone was primarily using these alternate methods to rank players, that could generate some interesting discussion.
χ Smith
FHS '15
Chicago '19
Cheynem
Sin
Posts: 6809
Joined: Tue May 11, 2004 11:19 am
Location: Grand Rapids, Michigan

### Re: Player Poll 2018: Delightfully Devilish

This is the last player poll I'm running. Nothing that happened this year or the discussion about dropping the poll were factors in this; as I am less attached to the college game, I had less of an interest in running this. Auroni Gupta will take over the player poll next year and as the time gets closer, he'll explain how he'll run things.

The poll takes a fair share of crap over the years, from people who think the methodology is wrong or others who find that polls create a lot of hurt feelings. I always found the poll interesting as a nothing more, nothing less look at the perceptions of who the best players in the game are in a given year. The top players are honored, skillful years and performances are honored, and good players who at least get a vote get a nod and acknowledgment.

A long time ago, Dr. Jerry Vinokurov (or Bruce Arthur, they're very similar) said the point of any poll is to start discussion. I think these did and I had a good time running them.
Mike Cheyne
Formerly U of Minnesota

"You killed HSAPQ"--Matt Bollinger