umeu wrote:is there a list of the seeds in proper order?
Absolutely. This entire thing was generated using a program I wrote. Would be happy to show you the seeded outputs as well. We do not have them in a friendly form to share. I can send you a table on skype if your super interested.
Basically groups work like this:
Seeds 1 - 32 are seeded to brackets
Seeds 33 - 160 are seeded to groups. These seeds will be reduced to seeds 33-96 by the end of the group stage (See below)
Seeds 97-128 are Bye seeds given to the seeded players
These groups are then mapped to bracket spots in the Round of 128 based on the seeds in the top two slots. This is done in the same way a normal bracket. So if you finish top of group 1 your seed is 33. That seed is then used to place you into the RO128 brackets.
By this logic only seeds 33-96 are assigned coming out of the group stage (The top 2 seed spots in each group).
Now we have a RO128 Bracket that is seeded in the same way as a traditional bracket
Seeds 1-32 are pre-seeded players Seeds 33-96 are group stage top 2 players Seeds 97-128 are empty for the pre-seeded players
This linking is all done automatically and is updated as the results of group stage roll in.
You can see this projection here in the linked brackets below. Right now since no results are in it is merely sorted by seed. The top 2 players in each group are marked by a ** meaning they are not confirmed. I would also like to mention that the players facing eachother in the RO128 will be from different groups. That is the nature of how this process works. For Group 1, seed 33 will be in a completely different spot than seed 97.
If I'm reading this correctly, does this mean that your program seeds the 64th highest player first in group 32, and the 65th highest player second in group 1? If that's correct, don't you see the gigantic fundamental flaw in your program for creating groups? The disparity between the skill of group 1 and group 32 would be enormous and practically ruin the entire concept of seeding (which is to make sure higher seeded players don't have to play each other that early/not punish good players by making them have to play other good players immediately). Snake order should have been used, and it looks like it wasn't. There's no reason to put the 1st seed with 11th, 2nd with 12th, 10th with 20th, etc------it should be 1st with 20th, 10th with 11th, if you follow.
That said, clearly it's too late to change anything and you ought to just consider this for the next time you use this program.
eh, thats only assuming that the gap between #32 and #64 is enormous, which isnt all that abvious. If seeding is done by elo, its possible that for example #32 has 2300 elo while #64 has 2200 elo. Which isnt all that big of a difference.
umeu wrote:eh, thats only assuming that the gap between #32 and #64 is enormous, which isnt all that abvious. If seeding is done by elo, its possible that for example #32 has 2300 elo while #64 has 2200 elo. Which isnt all that big of a difference.
It's still the wrong way to seed. The argument "well, maybe it won't work out so bad after all if we're lucky about it" is horrible. There's absolutely no reason why the strongest first seed, strongest second seed, strongest third seed, and strongest 4th seed should be in the same group. Same goes for weakest first seed, weakest second, etc.
umeu wrote:is there a list of the seeds in proper order?
Absolutely. This entire thing was generated using a program I wrote. Would be happy to show you the seeded outputs as well. We do not have them in a friendly form to share. I can send you a table on skype if your super interested.
Basically groups work like this:
Seeds 1 - 32 are seeded to brackets
Seeds 33 - 160 are seeded to groups. These seeds will be reduced to seeds 33-96 by the end of the group stage (See below)
Seeds 97-128 are Bye seeds given to the seeded players
These groups are then mapped to bracket spots in the Round of 128 based on the seeds in the top two slots. This is done in the same way a normal bracket. So if you finish top of group 1 your seed is 33. That seed is then used to place you into the RO128 brackets.
By this logic only seeds 33-96 are assigned coming out of the group stage (The top 2 seed spots in each group).
Now we have a RO128 Bracket that is seeded in the same way as a traditional bracket
Seeds 1-32 are pre-seeded players Seeds 33-96 are group stage top 2 players Seeds 97-128 are empty for the pre-seeded players
This linking is all done automatically and is updated as the results of group stage roll in.
You can see this projection here in the linked brackets below. Right now since no results are in it is merely sorted by seed. The top 2 players in each group are marked by a ** meaning they are not confirmed. I would also like to mention that the players facing eachother in the RO128 will be from different groups. That is the nature of how this process works. For Group 1, seed 33 will be in a completely different spot than seed 97.
If I'm reading this correctly, does this mean that your program seeds the 64th highest player first in group 32, and the 65th highest player second in group 1? If that's correct, don't you see the gigantic fundamental flaw in your program for creating groups? The disparity between the skill of group 1 and group 32 would be enormous and practically ruin the entire concept of seeding (which is to make sure higher seeded players don't have to play each other that early/not punish good players by making them have to play other good players immediately). Snake order should have been used, and it looks like it wasn't. There's no reason to put the 1st seed with 11th, 2nd with 12th, 10th with 20th, etc------it should be 1st with 20th, 10th with 11th, if you follow.
That said, clearly it's too late to change anything and you ought to just consider this for the next time you use this program.
Its funny you said that. After making this post I realize I should start working back up. It should be okay though because the number of outlyers we have in ELO. It will be correct in the bracket stage as well. Next time I am going to change it to go down then up if that makes sense. Either way going to be transparent about how it was done this time.
Cool that you saw it too!
It was quite a bit of work to get everything coded to take from the signup topic and build a linked two stage tournament. Sad that I made such a simple mistake, but Umeu's point will also hold for this event. We have tough groups at high group numbers as well.
umeu wrote:is there a list of the seeds in proper order?
Absolutely. This entire thing was generated using a program I wrote. Would be happy to show you the seeded outputs as well. We do not have them in a friendly form to share. I can send you a table on skype if your super interested.
Basically groups work like this:
Seeds 1 - 32 are seeded to brackets
Seeds 33 - 160 are seeded to groups. These seeds will be reduced to seeds 33-96 by the end of the group stage (See below)
Seeds 97-128 are Bye seeds given to the seeded players
These groups are then mapped to bracket spots in the Round of 128 based on the seeds in the top two slots. This is done in the same way a normal bracket. So if you finish top of group 1 your seed is 33. That seed is then used to place you into the RO128 brackets.
By this logic only seeds 33-96 are assigned coming out of the group stage (The top 2 seed spots in each group).
Now we have a RO128 Bracket that is seeded in the same way as a traditional bracket
Seeds 1-32 are pre-seeded players Seeds 33-96 are group stage top 2 players Seeds 97-128 are empty for the pre-seeded players
This linking is all done automatically and is updated as the results of group stage roll in.
You can see this projection here in the linked brackets below. Right now since no results are in it is merely sorted by seed. The top 2 players in each group are marked by a ** meaning they are not confirmed. I would also like to mention that the players facing eachother in the RO128 will be from different groups. That is the nature of how this process works. For Group 1, seed 33 will be in a completely different spot than seed 97.
If I'm reading this correctly, does this mean that your program seeds the 64th highest player first in group 32, and the 65th highest player second in group 1? If that's correct, don't you see the gigantic fundamental flaw in your program for creating groups? The disparity between the skill of group 1 and group 32 would be enormous and practically ruin the entire concept of seeding (which is to make sure higher seeded players don't have to play each other that early/not punish good players by making them have to play other good players immediately). Snake order should have been used, and it looks like it wasn't. There's no reason to put the 1st seed with 11th, 2nd with 12th, 10th with 20th, etc------it should be 1st with 20th, 10th with 11th, if you follow.
That said, clearly it's too late to change anything and you ought to just consider this for the next time you use this program.
Its funny you said that. After making this post I realize I should start working back up. It should be okay though because the number of outlyers we have in ELO. It will be correct in the bracket stage as well. Next time I am going to change it to go down then up if that makes sense. Either way going to be transparent about how it was done this time.
Cool that you saw it too!
It was quite a bit of work to get everything coded to take from the signup topic and build a linked two stage tournament. Sad that I made such a simple mistake, but Umeu's point will also hold for this event. We have tough groups at high group numbers as well.
Organizing and seeding a large tournament is a difficult and very rarely error free process. I think that the ESO-C team has done a great job bringing seasonal tournaments to the community, and small things like this are frankly nitpicking. However, it is something that I have experience in (I'm on a much larger team hosting a tournament with 2200+ individual entrants that's happening in January)! Feel free to PM me if you ever have a problem that needs solving.
Yeah I concur that this seemed to be primarily nitpicking. Hopefully though you will be able to reuse this process so this will only be a one time thing to set up.
Thanks for spending your time to organize this stuff for us. I appreciate it. . .
