"Random" match Schedules

[Bongle]

Originally Posted by Dominicano0519

uhh did u notice at what times they were written?

apparently we started writing at almost the same time, just since my post was longer i took longer to write

The first post in the other thread was started on march 2nd (two days ago), while yours was started today.

[nuggetsyl]

OK we keep talking about the problem. Now lets work to fix this. Programing challenge. Make a system first can use to set matches. I am not a programer but can something like this work?

1 can only play with a team once
2 can only go againts a team once
3 must have 6 matches between rounds (or 5 if it works better)

[The Lucas]

My dad had a good suggestion for how they should generate match schedules. Just generate matrices (size [6] by [total number of matches] filled with numbers 1 to number of teams) that are optimized for number of unique opponents and unique allies, but meet a reasonable rotation rate constraint*. Make these matrices for all possible numbers of teams at a regional (24-100 or so).

Now these optimized matrices can be used at every event. Just randomly assign each team a number 1 to number of teams and drop into the corresponding spots in the matrix.

These optimized matrices will take time to generate, but that is a one time cost for the foreseeable future since the same matrices can be reused. It doesn't seem difficult to write an algorithm to generate these matrices, since it is similar to the classic 8 Queens puzzle.

* When I say rotation rate, I mean the number of a the number of matches until the same team has to go again. Expected rotation rate is (number of teams)/6. I thing 50% of expected rotation should be the minimum constraint of any subset of matches

[Bongle]

Originally Posted by nuggetsyl

OK we keep talking about the problem. Now lets work to fix this. Programing challenge. Make a system first can use to set matches. I am not a programer but can something like this work?

1 can only play with a team once
2 can only go againts a team once
3 must have 6 matches between rounds (or 5 if it works better)

Here's an algorithm

del 2007_algorithm
copy 2006_algorithm to 2007_algorithm

Fixed!

Also, your constraints aren't good. At a small regional like waterloo where there are few robots and many (12+) matches to be played, not being able to play a team twice is impossible. 3 opponents * 12 matches is more than the robots at that regional (30 last year). That's why you have to start using less concrete things like "minimize the # of times a given robot plays another", which makes it more difficult. How do you know a given algorithm minimizes that constraint?

[pufame]

With less than a week before the next set of competitions, there is not enough time to write, test, re-write, distribute, explain, and implement a new system. However there is time to hand out last year's system (which wasn't perfect but a lot better than this) and just use that.

[Steve W]

To satisfy a lot of the issues, bring back the 2006 system with a few changes.

1 - no team can play 2 matches in a row
2 - no team can play more than 2 matches in 2 rounds
- ie: you could play match 6 and 8 but then not play till match 13 if there are 36 teams at the event

This is not perfect and your team could end up playing 4 games in 7 matches but that is the worst scenario. You could, in a 36 team regional, play matches 10, 12, 14, and 16 but then not play till match 34 and 36,

The fact is that the longer that you put between matches played the less random the teams that you play with becomes. I seem to remember the same issues being said last year with Hatch. Part of last years issues was that at the regional, score keepers did not set the variables right. At one venue I was at the practice schedule was the same as the match and almost all of the teams played were in your group of 6. A little tweak in software and we had a much better schedule.

We may not be able to use last years match generation as the software belonged to Hatch and they are gone.

[overlord]

Even with our rookie team, we pretty much faced our sister school, Team 2191, in every match as well as a few times with another rookie team 2140. It didnt really bother me since it led to us being the highest seeded rookie but lets spread the love around and see some other teams.

[ALIBI]

Does anyone know what FIRST had in mind with this years algorithm? My guess is no. I am sure that FIRST has made a major change in how alliances are set by looking at the alliances observed over the last three days and all the anaylizing that has been done. Given that, FIRST will probably be looking over the results to see if what they had intended to happen actually happened. If the results are what FIRST was looking for, I see no reason for them to change (not necessarily my personal opinion, especially if it was to force any team to constantly be opposed to one of the power house teams, that is just plain wrong). If it did not, FIRST may or may not change the algorithm. A change in the algorithm would be considered a major change and FIRST will probably be resistent to make changes this late in the game. The time for major changes like this is during the off season. Pardon me for my naivete, but, shouldn't a change like this be disscused in an open forum before it is instituted? That may have happened and I was simply not aware of it. If FIRST does not change the algorithm, I believe the least they could do is explain to us why the changes were made and what the results they are looking for are. My personal opinion is that you should be allied with and against as many different teams as possible. You learn to work with a lot of different people and stratagies. The teams that were always against eachother were probably getting a little tired of allways facing off. Team number should never be used for anything other than identification. If the alliances continue as they are, let's get over it and plan our statagies accordingly. GAME ON! Good luck to all the teams out there.

[Wedge34]

888 and 868 have been against each other in St. Louis 4 times

[BlondeNerd]

I don't know what the intent of the scheduling was, but I was a bit saddened by it. I love meeting most of the teams, but since I am in the pits I usually only get to meet our neighbors and alliance partners/opponents. Because we only played with a limited number of teams, I didn't get to meet very many people.
Another negative effect was that some teams played against each other every match. One thing I always point out to people who I am explaining FIRST to is that an opponent one match is an ally the next. Teams cannot play with poor behavior towards their opponents because they will have to play with them next. I know that teams should "play nice" regardless, but it is much easier for team ABC to feel animosity towards DEF if they play against them in every round. I am not saying that any specific teams showed ungracious professionalism towards any perpetual opponents. I am just pointing out that this setup made it easier for teams to see each other as opponents and not co-competitors.

[Testament-Doom]

Also, your constraints aren't good. At a small regional like waterloo where there are few robots and many (12+) matches to be played, not being able to play a team twice is impossible. 3 opponents * 12 matches is more than the robots at that regional (30 last year). That's why you have to start using less concrete things like "minimize the # of times a given robot plays another", which makes it more difficult. How do you know a given algorithm minimizes that constraint?

Simple. You could write a code that you can change variables based on those 3 principals that nuggetsyl suggest so it works better for each regional.

[ALIBI]

As a reply to BlondeNerd. What ever will happen to coopetition?

[Brian C]

OK we keep talking about the problem. Now lets work to fix this. Programing challenge. Make a system first can use to set matches. I am not a programer but can something like this work?

1 can only play with a team once
2 can only go againts a team once
3 must have 6 matches between rounds (or 5 if it works better)

One other thing to consider is that FIRST likes to have rookie teams on an alliance with veteran teams so that they can gain experience. You also want to avoid having an alliance made up entirely of rookie teams.

[Nuttyman54]

Another thing that the system affects is alliance selection from the point of view of the teams picking. Sure good scouting can tell you what teams can put up how many points and so on, but you really are in the dark about a lot of team's cooperative abilities.

Anyone with ramps is ineffective if their partners can't get up. Sure, you can look and physically see if their drivetrain can do it, but are the drivers good enough? will they come back to get on? will they listen to specific instructions on how to get up?

Until you play with them, it's difficult to tell if you can actually work well with them, which is especially important in a game that requires as much cooperation and coordination as this.

for the record, I am not complaining about either of our alliance partners (you guys were great). As one of the main scouts however, it was very frustrating when putting our list together of teams we would like to work with, since we had very little information on so many teams.

[Corey Balint]

This was definitely the worst match sheet I've seen. It made so many matches so much easier than others. I heard that it was suppose to average the totals of all the teams in the alliance and then get a similar average against. It made qualifications significantly easier for certain teams, and near impossible for others. It got very boring throughout the day. It made matches very predictable, and again, simple to win for certain teams.