Re: Championship Event - Where the "Random" Match Sorting Really "Shines"
 

OK, no problem. That's why I didn't post it.

Thanks,
AB

Re: Championship Event - Where the "Random" Match Sorting Really "Shines"
 

I don't know if the mentors of Chief Delphi would agree to this but here is one suggestion.
Create a thread for this purpose. It can be moderated if need be.
Have some criteria for posting.
Some suggestions off the top of my head:
- pros/cons of match generating program
- impact on the team(s) competition experience(s)
- summary

In business, one learns to deal with less than satisfactory situations. It is something that should be handled professionally by each team or team member that participates in a way that does not detract from the team. The developer will get the message loud and clear, esp. if presented in a manner that expresses the frustration but does not come across as 'hate mail' It could be an opportunity to act as young professionals of FIRST teams expressing serious concerns.

The thread can be forwarded to the developer.
Mike, if this is a no as well, I fully understand.

Jane

Re: Championship Event - Where the "Random" Match Sorting Really "Shines"
 

I would like to thank Dave L for his post. From what I have learned by speaking to the developer I can believe fully Dave's statements. Dave, again thanks for posting and keeping us informed.

Re: Championship Event - Where the "Random" Match Sorting Really "Shines"
 

Jane,
I fully endorse creating a thread where users can discuss ideas and post suggestions, relative to a "better algorithem" (as you have stated).

The thread can be forwarded to FIRST and they can forward it to the supplier/programmer/individual - if they choose to.

Nuttyman,
This site should NOT be used to get this specific software developer direct feedback. If you want to send this person direct feedback, I suggest you do that without involving this website.

CD doesn't need the hassles associated with individuals complaining about having their personal email account bombarded because someone posted their emal addy here.

Sorry - but I suggest you find another way.

Mike Aubry

Re: Championship Event - Where the "Random" Match Sorting Really "Shines"
 

Interesting. I know the developer was very proud of his algorithm. Incidentally, he believed that he had convinced FIRST that the algorithm was a valid way to load-level teams.

He also wrote the PDA/scoring program, which was certainly an achievement, much better than last year. Although to be fair to Hatch, counting "hits" on a PDA is easier than counting balls rolling through a chute. Still, very few scoring failures, and certainly nothing like the fiasco last year when the program couldn't even calculate standings correctly.

Btw, for those that wanted to email him, he does have a CD account. Perhaps he's reading.

Re: Championship Event - Where the "Random" Match Sorting Really "Shines"
 

I'll speak up on this one more time on this matter. We don't need a "match scheduling algorithm" that springs its results on us Thursday morning, only to have a statistical analysis demonstrate that it is totally messed up.

We need a set of match scheduling tables, one for each possible team count at an event. The entries in these tables should be 1 to N where N is the number of teams at the event. These tables should not be held secret and in fact should be published for all to inspect, and possibly improve upon. The table for 47 teams (to pick an example) at an event need never be changed, unless an alternative table is found that is better, replacing the prior one after it is vetted publicly.

On that Thursday that we actually know how many teams actually make it to the event, one can use popsicle sticks labled with team numbers shaken in a jar, or any other high quality random permutation of the team number list (I urge care here, as low quality computer based random number generators abound), to map the indices in the match scheduling table to the team numbers to create the match schedule.

So, the match schedules can be vetted publicly and the mapping of the team numbers to the indices in the schedule can be completely random. The complaints can then stop and we can then spend more time in this forum helping students.

Eugene

Re: Championship Event - Where the "Random" Match Sorting Really "Shines"
 

I do think that a truely random match listing would be unfair, but one idea is to take previous teams placements a point value, and make rookies random.

AC1 - 10pts
AC2 - 9pts
AC3 - 8pts
AC4 - 7pts
AC5 - 6pts
AC6 - 5pts
AC7 - 5pts
AC8 - 5pts

Pick 1 - 15pts
Pick 2 - 12pts
Pick 3 - 10pts
Pick 4 - 8pts
Pick 5 - 8pts
Pick 6 - 5pts
Pick 7 - 5pts
Pick 8 - 5pts

All Second Round Picks - 3pts

DNQ - 1pt

From there, decide upon ties as being random. I think this system would need some work done to it to make it effective, but it is just a concept.

Re: Championship Event - Where the "Random" Match Sorting Really "Shines"
 

Actually, this year's program failed to calculate rankings correctly as well. St. Louis was down for over an hour as they did it by hand. They fixed this after week 1, but this was a major defect that should have been caught in testing before it ever got to a regional.

Re: Championship Event - Where the "Random" Match Sorting Really "Shines"
 

I don't think the teams at the championship should be split into different divisions based on team number, as they have all managed to get to the championship, meaning that they must all have done something well.

On the topic of randomness, matches should have been a lot more random. In 7 matches, we were paired with one team 3 times, against another team 3 times, and with the second team another time. It would have been nice to play with/against more teams.

Re: Championship Event - Where the "Random" Match Sorting Really "Shines"
 

for the most part, this is not true. and also, what improvement was made by having a team play another 4 times in one regional that only had 7 matches and playing another team 3 times.
We work so hard all year round and try to attend one of the biggest regionals in NJ, only to find that we play the same teams over again and vice versa as they become our partners also.
WHY ATTEND A REGIONAL OF 59 IF YOU WILL ONLY SEE A FEW!?!

Re: Championship Event - Where the "Random" Match Sorting Really "Shines"
 

It's clear that most of us agree that this year's algorithm was a big problem and that our concerns were communicated through a number of channels starting in week one. However, before lambasting anyone we should recognize a few important facts.

First, and foremost, because of the efforts of FIRST, 4fx and their subcontractors, this year's field control and scoring systems were a tremendous improvement over previous years in terms of ease of setup, reliability and maintainability. One only has to look at how few matches had to be re-run due to field fault this year compared with last to see a clear benefit to the teams and event staff.

Next, according to "the developer", work started on this year's system in Sept. '06 and was completed on time and on what amounts to a shoestring budget, at least in the world of control systems. So, in four months they went from a GDC concept to delivery of the first system, and then replicated it in time to ship to the week one regionals by the end of February.

Finally, with regards to making a major overhaul of the qualification match scheduling algorithm, FIRST probably can't afford to pay for the kind of resources that would have been needed to make significant changes to the Field Management System mid-season. Remember that the people who would make that change were already working full-time fixing bugs that caused ranking problems and FMS database access conflicts - when they weren't running the field at regionals and providing field support. I suspect that the main reason we didn't see a big change mid-season was that it would have increased the risk of FMS failure at events and taken resources away from more critical tasks.

I've debated the team ranking philosophy with "the developer" a few times and at this point I'd say we've agreed to disagree. It's obvious that FIRST has to clearly specify to him what they expect for next year's system and I hope they do it earlier than September. It is not clear to me that everyone at FIRST headquarters agrees that this year's algorithm is all that bad, so if people want their complaints acted on, they should go there, not vent at the guy who wrote it.

Re: Championship Event - Where the "Random" Match Sorting Really "Shines"
 

Well, not to whine and complain, but considering the fact that during the championships, my team was paired with one of the same teams for 3 of our matches in Curie. I don't think it was that random, considering that there were about 80 other teams in the division, and we were never against that team.
Granted one of the matches was during the practice day, but even at Regionals I have found similar things happening.
And to reitterate, many of the same teams do win reoccuringly because of the pairings, and many wonderful teams lose because of the same reason.
However, it does help to have some of these pairings because some of the teams just don't have the experience. Then again, rookie teams win regioanls so, it all depends on the team, I guess.
I am not "hating on F.I.R.S.T." I am just voicing what I think it true.

Re: Championship Event - Where the "Random" Match Sorting Really "Shines"
 

From Team Update #21:

1) The Director of FRC is very aware of the concerns regarding the Alliance Selection
algorithm used throughout the 2007 competition season. Unfortunately, we ran out of time
and the resources necessary to properly implement a change and meanwhile mitigate any
risk to the FIRST Championship.

********************************