Tournament seeding; A Good Thing?

[sixty_something]

New options exist for selecting the method of seeding a tourney with TourneyBot when the tournament type specified with CREATE is SINGLE as it is for the Bloody Mary tournaments which are going automatic this season. Currently, I have stated the Bloody Marys for Season III will be seeded using RANDOM order as we have during the first two seasons. However, I am considering changing that to seeding by RATING or even some other way.

sixty_something: help SEEDING
TourneyBot: Syntax: TELL TOURNEYBOT SEEDING (rating|registration|random|exp|name|allow) TOURNEY <tn>
TourneyBot: Set the Seeding method for the tourney. Bracket only supports Random (default) and Single supports all of them with Rating as the default

The SEEDING ALLOW command option enables the TD to set a unique preferential seeding order for any tourney. For example, sunray, who won 4 weekly tourneys during Bloody Mary Season II, will be awarded two first round byes using the SEEDING ALLOW command in the BMSC II winners brackets to set him as #1 seed in both brackets. Other winners of two or more tourneys will also be given a seeding priority over one-time winners. One time winners will then be seeded by rating.

The option I like for weekly Bloody Mary tourneys also uses ALLOW. It is option 6 above, which sets first round byes, if any, based first on previous wins during the current season and next on rating. I'm not sure whether this is even possible, but suspect it is. If so, I like the bonus such seeding gives previous winners over rated players in awarding byes. As a second choice, I like seeding by RATING as it awards first round byes, if any, to higher rated players. The Bloody Mary tourney has always drawn a large number of higher rated players and I'd like to continue to encourage that trend - especially since top seeds seldom win. Thus, the Bloody Mary is a good humility lesson for top players. However, I also like RANDOM seeding because lower rated players can get a leg up on a win with a first round bye.

So, vote above to make your opinion known. I may not listen to the vote, being a tourney czar trumps democracy, but I will provide cake for the masses if sent to the guillotine by a peasant insurrection.


"Qu'ils mangent de la brioche" or as popularly translated "Let them eat cake"
-- Marie-Antoinette (1755-93), the Queen consort of Louis XVI

[Tom]

If I remember correctly Seeding ALLOW will let you specify the seeding in the ALLOW list

If the tourney is Private then ONLY the players in allow will be able to register if it is open
then anyone can register but the ALLOWED players will be highest seeded followed by
the other players in REGISTRATION order

We really should test this to be sure... this is just the way I think it will work...

Tom

[sixty_something]

interesting responses to the poll question .. so far, the preferences for seeding order, which determines assignment of byes, are:

• Registration order - 3
• Random - 2
• Previous winners then by rating - 1
• Rating, Experience, Name - 0

as promised, i didn't listen -- after all, czarhood has its privileges

today's tourney used Rating as the seed order .. here are the first and second round pairings:

Round 1 (3 pts):
rockfish (1768) vs adrian (1690) winner plays karmakat (1542),
pilogen (1742) vs supermane (1708) winner plays erdbeere (1661),
stog (1661) vs rollingfool (1538) winner plays jackdaddy (1609),
rossa (1488) vs cassandra (1556) winner plays sweets (1543),
Miccimalou (1607) vs Trilli (1584) winner plays sunray (1811),
WiFi (1507) vs socksey (1644) winner plays sixty_something (1741),

Round 2 (3 pts):
pck (1773) vs johnwayne (1695),
RickrInSF (1750) vs Picasso (1735)
that didn't come out as expected seeding by rating should have assigned the six byes to the top 6 rated players .. it clearly did not, but on closer examination apparently TourneyBot used Random seeding for #2746 .. so, while i, as Bloody Mary czar, didn't listen TourneyBot did .. thus, the random pairing for 22 players look correct .. round 3 pairings with a previous match results further reveal the structure of the pairings

Round 3 (3 pts):
adrian (1690) defeated rockfish and karmakat  vs pck (1773) who defeated johnwayne in Round 2
pilogen (1742) defeated supermane and erdbeere vs jackdaddy (1609) who defeated [winner of stog vs rollingfool],
sweets (1543) defeated [winnner of rossa vs cassandra] vs RickrInSF (1750) who defeated Picasso in Round 2
Miccimalou (1607) defeated Trilli and sunray vs sixty_something (1741) who defeated [winner of WiFi vs socksey]

thus the pairings can be visualized by sketching it out on the back of a napkin, if you need visuals

IMHO, the tourney, our first Bloody Mary run automatically, went really well and ended in 2:06 which is quite fast for a 22 player tourney

i'm sure we'll continue to find surprises and undocumented features of automatic tourneys .. however, from the perspective of this TD, i really liked the way TourneyBot handled opening, registration, and clock warnings .. it made playing a lot easier for me as a TD which is an unexpected benefit .. it also allowed me time to spam Bloody Mary news, facts, and solicitations to play - all a good thing, IMO

so, thanks again, Tom  i'm delighted with TourneyBot's continuing improvements and reliablity 

[dorbel]

The last time that I played in a Bloody Mary tournament, the draw was "seeded". The word seeded can mean several different things, but in this case it meant that players were ordered according to their fibs rating at the time of entry and then the draw was made. The top rated player was paired with the lowest, then the second top with the second lowest and so on. I assume that the draw was also arranged so that the top-rated player couldn't meet the second top-rated player before the final and so on, but I don't know that.
To me, this arrangement defies a basic principle of draws for backgammon tournaments, which is that the draw should be random. It increases the chances of the top-rated players and decreases those of the weakest. Not only that, but in this case the three byes in the draw were regarded as the three weakest players, so the strongest players were awarded the byes! How this is supposed to be fair is beyond me. It must surely discourage entries from lower ranked players and will do nothing to attract the higher ranked. In fact in my case it would actively discourage me from playing.
A case can be made for seeding four players in a sixteen player draw, with seeding in this case meaning that the the seeded players are split into the four quarters of the draw, with the top seed in the first quarter the fourth seed in the second quarter and seeds two and three in the third and fourth quarter.
This arrangement, which is how tennis tournaments are seeded (usually with a larger number of seeds) does slightly increase the chances of the top players, but the effect on the bottom players is partially mitigated by the fact that they can only get one of the strong players in their quarter. Byes should be allocated so that one bye is randomly assigned, two byes so that one is in the top half, one in the bottom and so on.
If fibs tournaments need seeding, a need that is far from clear, then this method is fair and reasonable. The first method described above is not.
The practice of some tournament directors of putting players into the draw in the order that they enter and updating the draw table continuously for people to see is also open to manipulation, creating the possibility of players picking their moment to enter when it will get them a better draw. It doesn't matter much and there is no evidence that anybody does this, but there is a principle involved. The draw should, in my opinion, be open, fair and random.

[sixty_something]

another good an thoughtful reply, dorbel .. thanks for the followup on our discussion recently in shouts .. i merged your post with this thread since it is a current issue with TourneyBot development and the new Bloody Mary seeding order we are experimenting with this season .. your comment below gets to the heart of the issue .. i corrected what i suspect was a typo below

To me, this arrangement defies a basic principle of draws for backgammon tournaments, which is that the draw should be random. It increases the chances of the top-rated players and decreases those of the weakest. Not only that, but in this case the three byes in the draw were regarded as the three weakest [stongest] players, so the strongest players were awarded the byes! How this is supposed to be fair is beyond me. It must surely discourage entries from lower ranked players and will do nothing to attract the higher ranked. In fact in my case it would actively discourage me from playing.

indeed the seed order was set to order seeds by rating .. seeding simply assigns a number from 1 to 32 to all players .. in the tourney you refer to we had 29 players and the top three seeds were indeed correctly awarded the three byes .. further, the pairings in round one and thus the overall bracket structure are defined by the seed order as follows with the byes awarded in a 29 player field in boldface:

• First Bracket: #1 vs #32, #5 vs #28, #9 vs #24, #13 vs #20
• Second Bracket: #3 vs #30, #7 vs #26, #11 vs #22, #14 vs #18
• Third Bracket: #2 vs #31, #6 vs #27, #10 vs #23, #13 vs #19
• Fourth Bracket: #4 vs #29, #8 vs #25, #12 vs #21, #16 vs #17

after the match you referred to dorbel, i did a careful analysis of registered players and the brackets created by TourneyBot which matched the patterns above exactly for Round 1 pairings and highlighted byes .. i have attached a text file showing the brackets for Round 1 from that tourney .. so, when i use the phrase seeded by ranking with TourneyBot this is a template for exactly what that means and Tom has implemented

before proposing these changes to Tom with the Bloody Mary always in mind, i have done careful research on seeding tourneys and maintained extensive records in seed order for past Bloody Marys .. therefore, i am not arbitrarily choosing this option, but i will have to elaborate on my thinking and reasons in a subsequent reply .. however, i repeat that the move toward seeding by rating is experimental for this Bloody Mary season .. your feedback and that of others is most welcome whether in a post here or in the poll above

finally, IMHO, regardless of the tournament, the design of match lengths, bracket seeding, and time rules is entirely the decision of the TD .. i have no problems proposing, using, or accepting feedback for any alternative rules i use in any tourneys .. providing they are fairly and consistently enforced, i have no problems with entering and enjoying other tourneys at FIBS and elsewhere which include variations on traditional rules

again, dorbel, your opinion and experience is respected and always welcomed, even when we may respectfully disagree

btw, a related but distinctly different topic is Re: Automatic Tournaments: Bloody Mary "Sunday Special" .. see that for more current information about the mechanics of these new Bloody Mary tourneys .. during this season, i will continue to provide updates and information in both these threads

[playBunny]

i corrected what i suspect was a typo below

Not only that, but in this case the three byes in the draw were regarded as the three weakest players, so the strongest players were awarded the byes!

It was meant as dorbel wrote it. It's a way of explaining the fact that byes are given to the strongest players. The byes are treated as if they were players with zero rating and hence they get paired with the strongest players by the seeded pairs algorithm.

The algorithm is simple and elegant. The order of players in the final list may seem somewhat arbitrary at first glance but the method is very straightforward. For the first round the list of 1-32 players is effectively folded in half so the the bottom meets the top and the middle meets the middle. These pairs are then listed and *that* list folded in the same way so that the top pair. The resulting pairs of pairs are then listed and folded, and so on until there's just one set. That's then the order of matches and puts everyone in the correct bracket. When coding it, it can be written recursively. The following may illustrate the folding.

1         
2         
3          18   
4          27           18,45
-  --->  ---  --->  ------  --->  18, 45, 27, 36
5          36           27,36
6          45
7
8

[playBunny]

The last time that I played in a Bloody Mary tournament, the draw was "seeded". The word seeded can mean several different things, but in this case it meant that players were ordered according to their fibs rating at the time of entry and then the draw was made. The top rated player was paired with the lowest, then the second top with the second lowest and so on. I assume that the draw was also arranged so that the top-rated player couldn't meet the second top-rated player before the final and so on, but I don't know that.

Yes, the purpose of such seeding is to encourage the tournament to follow the course such that the strongest two meet in the final, the strongest four all play in the semi-finals, the best eight in the quarter-finals and so on down.

To me, this arrangement defies a basic principle of draws for backgammon tournaments, which is that the draw should be random.

Seeking randomness out of principle doesn't make sense to me. There has to be a purpose for it.

It [seeding] increases the chances of the top-rated players and decreases those of the weakest. Not only that, but in this case the three byes in the draw were regarded as the three weakest players, so the strongest players were awarded the byes! How this is supposed to be fair is beyond me.

Seeded pairing does discriminate against weaker players but anything other than seeded pairing discriminates against stronger players. Who says that the weaker ones should be favoured? Who says that the strongest should be favoured? There is no truly fair way of pairing players except by using a value system that defines that method as being fair. Egalitarianism isn't necessarily fair.

Is it fair if the two top players meet each other in their first match and one therefore gets knocked out immediately? Not to them, I wouldn't have thought. That can happen with random pairing. That's the extreme case, of course, but with random pairing there may be several pairs of players who wouldn't necessarily meet until a later round. There's a degree of unfairness in each one. Calling it the luck of the draw doesn't make it fair.

It [seeding] must surely discourage entries from lower ranked players and will do nothing to attract the higher ranked. In fact in my case it would actively discourage me from playing.

It may discourage the lower rated players but "must surely" is a claim that needs to be tested. I especially don't see how higher rated players must necessarily find no value in a pairing method that makes it more likely that they will meet stronger opponents in successive rounds rather than arbitrarily. If it would actively discourage you from playing then that says something about your values, in particular it suggests that your feeling for the underdog outweighs your desire to win, or something like that.

A case can be made for seeding four players in a sixteen player draw, ...

A case can be made for every type of seeding.

... with seeding in this case meaning that the the seeded players are split into the four quarters of the draw, with the top seed in the first quarter the fourth seed in the second quarter and seeds two and three in the third and fourth quarter.
This arrangement, which is how tennis tournaments are seeded (usually with a larger number of seeds) does slightly increase the chances of the top players, but the effect on the bottom players is partially mitigated by the fact that they can only get one of the strong players in their quarter.

It's a reasonable compromise, good for the top players and okay for the bottom ones. But this method therefore discriminates most against the players just below the seeded ones!

Byes should be allocated so that one bye is randomly assigned, two byes so that one is in the top half, one in the bottom and so on.

This also happens with seeded pairing. Just as the top 4 players are spread among the quarters, so are the lowest 4, and thus the byes.

If fibs tournaments need seeding, a need that is far from clear, then this method [seeded top and byes, random otherwise] is fair and reasonable. The first method described above is not.

They're all reasonable and the fairness is debatable. Well, not all. At VogClub the top players get byes and are often then immediately paired with each other in the second round!

But that's usually the effect of particular numbers in a tourney. The general method at Vog is to order by rating and split the list in two but, instead of folding the second half up so that the bottom meets the top, the second half is slid up so that its top player (middle rated) play the topmost player and the first half's lowest player (also middle rated) plays the weakest. This method seeks to minimise the rating difference between pairs of players. That's good for the weakest because they don't play the strongest but that means that the top players get someone stronger than they'd have expected and the lower middle players also get a stronger opponent than they would have otherwise.

The practice of some tournament directors of putting players into the draw in the order that they enter and updating the draw table continuously for people to see is also open to manipulation, creating the possibility of players picking their moment to enter when it will get them a better draw. It doesn't matter much and there is no evidence that anybody does this, but there is a principle involved. The draw should, in my opinion, be open, fair and random.

However you cut it there's unfairness!

[dorbel]

Seeking randomness out of principle doesn't make sense to me. There has to be a purpose for it.

There is a purpose to it. It is to ensure that each player is treated equally when the draw is made.

It may discourage the lower rated players but "must surely" is a claim that needs to be tested.

Let us consider the case of a player at the lower end of the ratings in the tournament, 1400 perhaps. Are we to say to him, "Well from now on, your chances of getting a first round bye are zero and actually, we have arranged the draw so that you will always have to play the strongest player entered in the first round"? That isn't fair at all and must surely discourage any low ranked player. Backgammon thrives on new players constantly coming in and they all start at the bottom.

Speaking as one of the higher ranked players in any fibs tournament draw and the beneficiary of one of the byes in the tournament to which I specifically referred, I have no objection to playing another strong player in the first round. It doesn't seem unfair to me if that is the way that the draw goes. As to getting a bye, I have always been prepared to give up a first round bye to get a latecomer into the tournament and have actually done so on at least two occasions.

If it would actively discourage you from playing then that says something about your values, in particular it suggests that your feeling for the underdog outweighs your desire to win, or something like that.

I have no feeling for the underdog and am indeed delighted to trample him underfoot whenever I can. However, no need to make it harder for the poor bugger.

Seeding makes a lot of sense at a tennis tournament, because it is a commercial enterprise and the organisers naturally wish to make the final as attractive as possible. Nobody wants Federer/Nadal to be in the first round. A tournament on fibs is a bit of fun and everybody who enters should be treated equally. Personally, seeded tournaments and tournaments where the byes are given to the "best" players won't be getting an entry from me, much to the relief of some TDs I'm sure!

[playBunny]

To me, this arrangement defies a basic principle of draws for backgammon tournaments, which is that the draw should be random.

That makes randomness your pricinciple. If you'd said fair instead of random then it would have made sense.

When I said "It may discourage the lower rated players but "must surely" is a claim that needs to be tested." I meant that it needs to be tested by polling real live players, not making up an example of an in-your-face declaration of bias against low rated players. In practice it's not said so bluntly. In fact often enough nothing may be said at all about how byes are handled. A player has to ask or find and read the rules if they want to know.

When it comes to testing the "must surely" claim regarding discouragement I predict that you'll find something surprising. A lot of low rated players don't care two hoots how it's done. They enter, they play, they take their chances and they have fun, win or lose. You'll see this time and time again. Being low rated with high experience means that you have reached an understanding that there's a lot of losing to be done. A low rated player who can't handle that idea doesn't get to become a highly experienced low rated player. A low rated player who's on the up and up might object to higher rated players getting the byes, but they're presumably working towards being one of those higher rated players. It may well act as a mild incentive; a reward for studying hard.

I think the onus is on you to prove that seeding "must surely discourage any low ranked player".

Backgammon thrives on new players constantly coming in and they all start at the bottom

It sure does and if they're anything like me when I started, such sophistications as who gets the byes aren't really that relevant. There's a ton of more basic stuff to be grasped before worrying about that kind of thing. And even when the basic stuff has been grasped, there are plenty of players who don't concern themselves too deeply with the mechanics of tourneys, they're just happy that someone else is doing the work.

I have no objection to playing another strong player in the first round. It doesn't seem unfair to me if that is the way that the draw goes.

This is a fine sentinment but in fact it is unfair to you. You simply don't mind. I am curious about how much value you give to your strong opponent's convcern about meeting you in the first round. Presumably that player's feelings matter less to you because they have a higher rating? If so then you are being discriminating and I'd wonder how fair that is.

have no feeling for the underdog and am indeed delighted to trample him underfoot whenever I can.

I'm sure that that's the case when you're across the board and you therefore wouldn't give up MWC out of the goodness of your heart but this discussion isn't about that. You're talking about a willingness to give up TWC (tourney winning chances) as a high rated player in order that the lowest rated players have a higher winning chance. As you say, there's no need to make it harder for the poor bugger. And you underscore that with the intent to act: seeded tournaments and tournaments where the byes are given to the "best" players won't be getting an entry from me. You'd effectively resign the tournament? Now that's feeling for the underdog!

A tournament on fibs is a bit of fun and everybody who enters should be treated equally

I agree that it is a bit of fun, though I talk more from experience as a TD of tourneys at DailyGammon. In many instances a player joins to participate, and winning is not that great a priority. That's not how it is for me and I've had to bend my head round the idea that many players don't join the tournament because they want to win it; they join to be a part of it and hopefully they'll do better than their expectations and only maybe, Dice Gods willling, they'll win it.

In conclusion I'm not saying that you're wrong to value the underdog. I've made exactly the same arguments on their behalf myself. When I first learned that seeding paired the top player against the bottom player it seemed daft. I don't make any claims that seeded pairing is the best way. There is no best way. As I said before, there's unfairness in all methods unless you define random pairing as fair.

[dorbel]

It's called Equality of Opportunity and it is at the heart of any democratic system. The high rated player does not need a further advantage, the lower rated player does not deserve to be further handicapped. That just about covers it for me.

[sixty_something]

methinks we often think too much, but this is an excellent dialog fairly presenting two strong points of view

i really enjoy seeing underdogs win a tourney, but, as an underdog, i'd rather win a tourney by fighting my way through a balanced bracket than winning by the fluke of a random draw which paired top seeds in early rounds and sneaking in the backdoor in an easy bracket

i welcome and actively recruit new players, but i have been consciously moving the Bloody Mary toward something a little different -- it is the "Sunday Special" after all .. for now, i am satisfied i am doing everything necessary to improve the tourney, at least as best i can .. so, i will keep on keeping on, always welcoming feedback, but not necessarily always agreeing

the two distinguishing differences with the Bloody Marys have been my gradual move toward setting higher registration limits and seeding by rating, both have stirred dorbel toward strongly expressed responses .. others have questioned the wisdom in my choices, as well, but i have not made the choices willy-nilly .. i have monitored the results carefully and will continue to do so

for me, as TD, the primary measure of success has been whether the tourneys are fun, fair, easy to run, and have an added value for winners .. so far, i am satisfied with all those measures of success .. additionally, the number of players has remained in the mid-twenties since these changes were instituted which is a large turnout for FIBS tourneys, another reasonable measure of success

thanks again to both playBunny and dorbel for freely and actively giving us all something to think about .. that, IMHO, is the essence of democracy in action .. backgammon itself is the ultimate democratic game, so i empathize with dorbel's sentiments and respect his opinion and experience a great deal, but i simply respectfully disagree on the fairness issue .. that's about it for me, too, but i will continue to monitor the changes, report results, and, most of all, enjoy the fun

[playBunny]

As I suggested, I once took dorbel's position and when I adopt my own rooting for the underdog stance, which is more natural to me than otherwise, I take it again. However, having just now been the top rated player in a tourney and paired against the second highest player in our first match, I'm very aware of how that discriminates against the top players. One of us had to go! I don't see how there's equality of opportunity in such a scheme as the benefit goes to everyone rated below the top pair. But if I'm watching a tourney I quite agree with the top players knocking each other out!

I've enjoyed this discussion. I don't think there's any right answer on this one and I think it should be TD's choice. People will vote with their feet but I think that people will join whatever methods are used. I believe that often enough the TD, as a person, a motivator and an organiser, is as much or more of a factor than the tourney or its particulars!