Yet Another Pairing Tool... not.

[Enty]

Hi all,

So, here is the first experimental version of my new pairing tool. Right now it is for power users only: it is written in Ruby (therefore it's multiplatform, tested on WinXP and OSX, should work on linux), it takes an input file for its first argument (either an URL or a local file) and outputs the pairings to the console. The format is documented in the code itself.

Why is it special? Because it pairs people as effectively as possible in a way that country and clanmates don't play against each other.

This version is not for the faint hearted but a much more user-friendly version is work in progress and will likely be implemented in WO2008's pairing tool.

You can download it from here (http://pastie.org/253016).

--Enty

P.S: Thanks for FFie for the idea and Kiros for the idea of the algorithm.

[Dario]

I made my own pairing tool with MS excel, and something I took into account (because for me it really matters) is grouping the players on skill levels (obviosuly only good when you more or less know the skills of the different players at that particular scheme). Then the top 4 will be split so that they never face eachother until semifinals, the next 4 the same, then 8 more and so on.
Made that because I was tired of seeing some of the top players lose in the first or second round, (cough, ENTY! :P) just because the fully random matching decided they had to play other top players ^^.

Edit: before anybody says "but how can you know if someone is good but don't know and you match him against another good player"?, well, it is simple: the chances of that happening in this semi-subjective way of pairing are WAY lower than the chances of that sucky-pairing happening with a fully random system :) .

Edit 2: I fully support that country/clan exclusion thingy, just though that adding an "approximate skill level exclusion" thingy is good too :).

[Enty]

I made my own pairing tool with MS excel, and something I took into account (because for me it really matters) is grouping the players on skill levels (obviosuly only good when you more or less know the skills of the different players at that particular scheme). Then the top 4 will be split so that they never face eachother until semifinals, the next 4 the same, then 8 more and so on.
Made that because I was tired of seeing some of the top players lose in the first or second round, (cough, ENTY! :P Note from Enty: <3) just because the fully random matching decided they had to play other top players ^^.

Edit: before anybody says "but how can you know if someone is good but don't know and you match him against another good player"?, well, it is simple: the chances of that happening in this semi-subjective way of pairing are WAY lower than the chances of that sucky-pairing happening with a fully random system :) .

Edit 2: I fully support that country/clan exclusion thingy, just though that adding an "approximate skill level exclusion" thingy is good too :).
Hmm... on a first thought, this is a bit too subjective, however on the second... it may or may not be... I'm gonna go ahead and think out loud a bit.

I don't know if it's fair that top players not playing against each other early, thus having a bigger chance of getting into semis anyways since they play against weaker players. Another idea came to my mind tho.

What if we split players into two groups (and I'm not gonna be PC here): Pros and noobs. Pros plays against each other in the tourney till 1 remains, noobs play against each other till one remains... then these two play the finals.

Then again, who is to decide on pros and noobs? What if we would need three groups, noobs, semi-pros and pros? The only semireliable way would be their WO points, probably in the given scheme, and it's really not that reliable. Also what if we have a newcomer that is apparently really good in the given scheme?

I think skill-based pairing is way too subjective and I don't really know a fair way for that. So... no :)

[franpa]

You should be using the term, newbs. not noobs, because noobs is a insult.

[Enty]

You should be using the term, newbs. not noobs, because noobs is a insult.

Er...sorry, but...
(and I'm not gonna be PC here)

Also i don't think "newbs" would be any more or less insulting than "noobs" :)

And sir you are missing the point of that post.

[GoDxWyvern]

There is no difference whatsoever between the terms "newb" and "noob".

On topic: I'm for complete randomness when it comes to pairing. If you face a mate or someone you loathe - then that's how it is. And who cares if you're assigned a strong opponent? I don't think there is such a thing as "deserving" to make it far in a tourney. If you earned your way to the finals by defeating no matter whom, then you deserve it. =)

[MrBunsy]

There is no difference whatsoever between the terms "newb" and "noob".

Yes there is. Newb is short for Newbie, not a derogatory term in itself. Noob is an insult.

[GoDxWyvern]

Fact is, a beginner would find both terms derogatory. Hence: no difference.

[MrBunsy]

How is newbie in any way derogatory?

[knkn]

Completely offtopic: As a beginner, both terms would/are be derogative. But as an experienced player, we'd know the difference, newbie/newb infact means only someone inexperienced, while n00b/noob is an insult. So, why are we making such a big deal [imagines this is gonna escalate] about this?

Completely ontopic: Great work. Tourneys should play out smoother than the creators random mind/dice.

[DarkOne]

Pairing should never be done on merits, because you simply don't know everyone, especially now that the community is split up like this.

[Random00]

I support Wyv's opinion 100%.
If some totally random pairing tool puts you against some clan mate in round 1, than its just like this.
Dario's system is subjective, but usually you really know who is good and who isnt, so it would make sense. But I think even if you knew the skill level of every player its unfair to lower skilled players, because they have less chances to win.
The best system would probably be something similar to the Footbal World Championship, with some good players picked out, that shouldnt play each other in round 1... But you'd really need to know the skill level of every player for that, and because this isnt the case too often I stay at my full random support :)