"Random" match Schedules

[Billfred]

Originally Posted by Richard

As I said earlier, I will add my concurrence to the communication being prepared for FIRST regarding the match generation algorithm; however, to me it seems clear that this is a policy issue and not a defect in the field management system. Let's keep the discussion on this topic focussed on what can/should be done to improve match generation.

How to improve it? A few thoughts:

-Make sure the cycles are flipped around after each period of play (Friday AM, Friday PM, Saturday AM). Being the last match before lunch and the first one after is a little tight, but the Friday-Saturday interlude is the perfect time to shuffle the deck.
-d.courtney's notion of a ten-minute break (a little under two match cycles) may also aid in shuffling things nicely, if doing it at the existing breaks doesn't cut it.
-If you absolutely must stick the same two teams in a second match, make sure they don't end up on the same side again.
-Above all else, remain fully aware that a low number (or lack thereof) is by no means a sure measure of a great robot.

Your mileage, of course, may vary.

[Nuttyman54]

The fact that some teams are always against each other worries me. Forget the fact that its the same general group of teams playing each round, but the fact that 190 has played 4 matches AGAINST 175 and NONE with them is a problem. It shouldn't be that hard to make a way to manually swap two of the teams in a match to make that happen.

I also agree that there is an issue with the grouping effect. In years past it has happened to some degree (an effect of the attempt to space matches evenly), but never to this extreme. I have no problem with teams not getting to play in a match with 2 or 3 of the teams, but more than 10 is a little obscene. I have heard good news that FIRST has gotten the message and that they will fix it for the next regionals.

On that note, week 1 regionals have always been the proving grounds for the game and all the associated software. I'm glad that this is the only issue i've seen so far

[Andrew Blair]

Ok, I did a random selection of teams from different regionals, and just checked how often they played against a particular team. Results may not be complete, but they give a pretty good spread.

St. Louis:

148- 4/7 matches against 45, 3/7 against 217

829- 4/7 matches against 547

1502- 7/7 matches against 1472

2177- 4/7 against 2167, 3/7 against 2219


VCU:

116- 7/7 matches against 122

401- 6/6 matches against 405

900- 3/6 matches against 843

2028- 6/6 matches against 2021


You could go into far greater detail, analyzing alliance partners, ratio of matches with vs. against , whatever, but there really isn't much point. The scoring algorithm, as has been stated, is grouping particular team number together, and is doing so consistently. There wasn't a team I found that wasn't playing with the same team (for or against), at least half the time. To be honest, and nothing bad directed or intended, but it kinda explains why there are so may rookies in the top 8 this year.

St. Louis:

#4- 2177

VCU:

#7- 2068
#8- 2028

PNR:

#1- 2122

NJ:

#5- 2180


There are always really good rookies, but there seems to be a disproportionate amount this year, IMHO. I can't say I'm right here though.

[Jonathan Norris]

We were with or against 663 at BAE for every match today... and seemed to play with the same group of team the whole day. I call not so random...

[Richard Wallace]

One way to argue this is as follows: seeding is meaningless, scouting should determine your alliance selection for eliminations. So the purpose of qualifying matches is not to generate seedings that accurately reflect team strength (major league baseball uses 162 games to do that, and still creates lopsided playoff match-ups); qualifying matches give scouts information about what each team can do, and for that purpose the match-ups are not a significant factor.

[JaneYoung]

Originally Posted by Pavan

I guess its a step in the right direction....Maybe a diagonal step, but forward nonetheless.

Are you on 118?

[Andrew Blair]

Originally Posted by Richard

One way to argue this is as follows: seeding is meaningless, scouting should determine your alliance selection for eliminations. So the purpose of qualifying matches is not to generate seedings that accurately reflect team strength (major league baseball uses 162 games to do that, and still creates lopsided playoff match-ups); qualifying matches give scouts information about what each team can do, and for that purpose the match-ups are not a significant factor.

I'd love to look at it this way, but unfortunately, it doesn't really work out well. If you are constantly playing a powerhouse team, your strategy, in order to try and win, must adapt into, likely, a defensive stance, even though you may be an amazing offensive robot.

Teams scout, but match standing can lend alot of weight to whether or not a team gets picked. Unfortunate, but real. That may be FIRST's argument though, you are right Richard.

[Zach Purser]

Add to your VCU list that 435 played against 510 every round. I could probably double your list of reasons that the current matching system is a bad idea, but if FIRST is already planning a change then I won't bother.

Can anyone confirm that FIRST is changing this for the next round of regionals and for nationals?

[GAMO]

Nobody has posted a similiar situation to the one on my team (2108 in VCU); we face off against 2186 every match, but also, our ally in one match is one of our competitors in the next.

For instance: Match number #11, our allies were 617 and 1413 (I believe). Match #21, we went up against 1413.

It's a cascading cycle where we basically only see three new robots on the field at any given match: 2 new allies, and 2 familiar on the opposition.

Has anybody else seen or have had this happen with other teams?

[Karthik]

Originally Posted by Richard

One way to argue this is as follows: seeding is meaningless, scouting should determine your alliance selection for eliminations. So the purpose of qualifying matches is not to generate seedings that accurately reflect team strength (major league baseball uses 162 games to do that, and still creates lopsided playoff match-ups); qualifying matches give scouts information about what each team can do, and for that purpose the match-ups are not a significant factor.

Richard,

For once I have to disagree with you. Yes, scouting should determine your alliance section, but seeding is not meaningless. The reason qualifying matches need to reflect team strength is to ensure that the teams who earn the right to select, are deserving to do so. No one wants to see a below average team seed #1, by virtue of constantly playing even weaker teams, or being paired with excellent teams. This unfairly rewards them, and punishes the team they select. (Provided that this below average team scouts well, and picks a top team)

[burkey_turkey]

Originally Posted by Richard

seeding is meaningless, scouting should determine your alliance selection for eliminations.

yes, i agree that of course the seeding doesnt reflect the actual rank of the robots acurately and scouting should determine who goes in the alliances, but the ranking does determine who gets to choose the alliances. Because there are some teams out there, i will not name any specifically but we all know about them, who just build a boxbot and run around doing alot of nothing, they will have a team that always plays them and always wins because they are basically playing 3 v 2. even if the robot isn't that great, it wins every match and gets placed up top where it is made to choose an alliance. this robot might not be one of the best 24 teams and might not be worthy of its spot.

One other thing relating the perpetual opponents to alliance captains is, it is impossible for both teams to be alliance captains. Your record is the inverse of your perpetual opponents record. you cannot possibly both be alliance partners, which is unfair to two really great and worthy teams who just so happen to be numerically close.

my team, we play every match against the same team. the bad thing is that the one downside of our robot is its traction. we have omniwheels and they arent gripping well. all of our opponents mechanisms arent functional so they have to focus on defense. we keep getting alliance partners that cant tube like we can so our opponent just slams us every time we pick up a tube. by the 5th match however, they received a yellow card (if not the first at the regional then definitely the second) for excessive violence. now we play an unagressive/paranoid partner for the rest of the matches which is too much in our advantage in my opinion.

while i see what FIRST is trying to do, i really think that a purely random schedule would be the best, getting paired with whoever you get paired with, assuring the variety you expect.

[Richard Wallace]

Originally Posted by Karthik

For once I have to disagree with you. Yes, scouting should determine your alliance section, but seeding is not meaningless. The reason qualifying matches need to reflect team strength is to ensure that the teams who earn the right to select, are deserving to do so. No one wants to see a below average team seed #1, by virtue of constantly playing even weaker teams, or being paired with excellent teams. This unfairly rewards them, and punishes the team they select. (Provided that this below average team scouts well, and picks a top team)

I don't mind if you disagree with me, Karthik. I'm not certain that I agree with me on this reasoning, either.

[Kevin Sevcik]

I'll have to agree with Karthik. The seeding is very important for ranking atleast the top 8 teams. A veteran #2 seed that has played and beaten all the other veteran teams at the regional with a rookie #1 seed that has barely beaten all the other rookies at the regional... That's just asking for trouble.

In addition, I can't see how a rookie team would prefer to be paired with and against other rookies for an entire regional. Many of these rookies were planning on being enabling robots to leverage the ramps and other features of veteran teams. Basically telling them that they'll never be paired with a veteran team just because is not a nice thing to do.

Mostly, I'm just surprised because the only serious complaints I've ever heard about the alliance pairing algorithms were that teams had to play with or against another team for half a regional or so. I can't really understand why first would turn around and go with a system like this in the face of that feedback.

[Covey41]

All this discussion about Richards idea that this type of seeding rounds was a new approach by FIRST to level the playing, or that this was planned by FIRST, is simply wrong. When this happened at the scrimmage in Suffield, CT 2 weeks ago, it was discussed with 3 high level people from FIRST.They all agreed that what happen that day, and whats happening this weekend, would be corrected by the 1st week events. With all the other fixes that came out of the scrimmage, I guess they ran out of time.

By continuing this line of discussion you are giving FIRST, and the software people an excuse for not fixing something that needs fixing.

Has anyone else notice that Mr Lavery his not added his voice to this discussion. Last year he was all over Hatch for every little issue.

[Jack Jones]

I don’t think they should even try to fix it. Why should the first week teams have to be the guinea pigs? Why should teams attending only the first week events end up having a less rewarding experience than ones who picked week four or five? This is an issue that could have been resolved way before the first event. I’m reminded of last year’s first week debacle with the scoring system and automatic scoring hardware. Not much changed in five weeks. Milwaukee probably had as many or more re-starts than VCU. They were still human counting in Las Vegas. It was not until the Championship that they programmed a break after autonomous - so they could at least get the autonomous winner right! Not much changed, except that we learned to live with it and made the best of it.

But last year had a really great game to make up for the warts. IMO, this year’s game is not as good, not even close. Whether the total experience is enough to make up for [insert pet peeve(s) here] remains to bee seen.