At SBPLI many of those really good teams ranked low were also those that weren’t really working. Rage, arguably one of the best teams there, was broken for a good portion of Friday, as were many other good teams.

The point of this system is to try to even out the number of good and bad partners. If you’re highly ranked and you get paired with a team that isn’t working, chances are you’ll drop in the rankings. So most likely next time you’ll be paired with the middle ranked team. If the partner is working then they have a good chance of moving up, if not you’ll move down further and next round you’ll probably be paired with the highest ranked team and thus you should have a good chance of scoring well and moving up. The cycle would continue like that throughout the competition so the advantages and disadvantages would balance each other out. In the current system you could very likely get paired with the lowest ranked partner every time and you’d get screwed even worse. Or you could get paired with the highest ranked every time and you’d be ranked artificially high.

Eh, even though this is an attempt to make things "more fair," it would not limit competitiveness in my mind, rather it would level out the discrepancies of "evenness" among everyone without the consequences of more limiting methods.
About this making the best teams always have to play along side the worst teams...I think that"s a great idea, because doesn"t their status as one of the "best" teams show that they are more readily able to handle a problem like an inoperative alliance partner? As a member on a team last year that was paired a couple of times with a team that was immobile yet still one, I really don't like being stuck with a broken team, yet it was still fun and it made the matches for the broken team fun too because even though their robot didn"t work, they still got some success as a very tangible award for their hard work. This year I was on a different team that was one of those teams that didn't move (well, we moved about 5' average in each of our matches [and yes, that was generally a straight line]). We were initially buoyed up in the standings by our placements with good teams (we still did a bit in the match where most of our points came from, but we couldn"t have one with another robot like ourselves as a partner) but in the end we fell back as the leaders were shown and we were very logically not picked as an alliance member.

Synopsis (for those of you who couldn"t understand my rambling or chose not to read it) This system would more evenly spread the joy of a competition around the teams and it would not prevent the good teams from showing their worth as a partner (because when looking for people to choose as alliance members, scouts will know that the rankings do not tell all [which would also be helpful for those of us teams who may have 5|<1ll2 but aren"t that good at seeding highly]).

(Also, you wouldn"t be that pressed for time to think up strategies, because after everyone scheduled for your next match had finished their previous match, you four wouldn"t exchange spots, you would merely have other teams moving around you in the rankings. And JosephM, it seems KenL disagrees at least somewhat too, so don"t feel too alone.)

Personally, I disagree with that point 100%. I know from personal experience that "good" teams aren't necessarily prepared to go it alone. At Nats in 2002 and 2003, my team was seeded #2 in our respective division. We did it not because we were a do-it-all team that could go out and win matches singlehandedly, but because we were great strategists and team players. When paired with any decent team, we could go out and rack up respectable scores, and the few times we got really good alliances, we'd go out and clean up the match. However, both times, we were hurt and missed out on the #1 seed because we were stuck with an inoperable partner in one of our last matches of the day. While a team that could do it all alone, like, say, 71 in 2002, might be able to survive continuous matchups with broken bots, teams like mine would be hurt because we did well as part of an alliance, not just our team going it alone. This idea is intriguing, but I just don't think it's a good idea. Better randomization of match pairings is the only way to truly give a fair experience.

Good thought...
 

I don't know if I agree with the particulars of the proposal but I think that it would be worth giving some thought to how do you seed teams more effectively.

There are many many cases where teams that do not belong on top end up highly ranked due to lucky partners, lucky opponents or whatever.

I don't have a good answer, but I think it is worth thinking about.

Part of me feels that there is a lot of information in peoples heads about teams.

Is there some way we could allow teams to rate their opponents and their partners performance during the last round and then use this information to help guide the seeding process?

I know there could be a lot of problems with this, but perhaps we could devise a system that could deal with this situation.

Think about it.

Joe J.

I think you missed the point. Only in rare circumstances would a team have to "go it alone." And that circumstance is no different than the random chance that you have to "go it alone" right now.

Consider the case where the number 1 and 2 teams come up against the number 39 and 40 teams. 1+2 versus 39+40 will usually end in a blow out and an uninteresting match. 1 and 2 will become more ensconced in the rankings, while 39 and 40 will be lodged at the bottom more firmly.

However, 1+40 versus 2+39 should be an interesting match. It should be more fun for all concerned. A competitive match is more fun to play and gives you a better feeling at the end than a blow out.

The case that many of the nay-sayers are articulating is when 1, 2, 3, and 40 occur. Then, 1+40 will have a tough time against 2+3. They would rather have 1+2 go against 3+40.

This is a whole different problem. However, if everyone does their scouting, if you are one out of 24 working robots, you should get picked for elims and the qualifying rounds are really not important.

Good concept, but...
 

This is a good idea, I just don't think it would be good for FIRST. If you think about it, the best teams float to the top anyway, because what makes a team good is when they can win reguardless of who the partner is. As for the teams who aren't so good that just happen to make it up to the top? Well, they are the cinderella's. Its always fun to root for the teams who are the underdogs. If you try to weed them out, you would have teams year after year getting nixed out of the playoffs, instead of having a chance to experience the energy that the elimination rounds bring.

Good Luck All,
Andy Grady

Re: Good thought...
 

I think that this proposal could be a great thing for FIRST once all the bugs are worked out.

Can you imagine? Indeed, scouting would be a nightmare. However, scouting would become a more important part of competition - instead of relying on seeding position to determine who the best of the best are, you could analyze match play (not that we don't do that now, but now every team would have to be carefully watched instead of just the traditional powerhouse teams and excellent rookies of the year).

I'm very excited. I think this is a great idea... and might just spice things up a bit in the FIRST community.

I agree with Andy Grady here. Over the years I've been in FIRST, I've been on both sides of 2 higher seeded teams just lay out 2 lower seeded teams; 2 middle seeded teams contest closely with the upper and lower seeded team; and 2 lower seeded teams beating 2 higher (in some cases much higher) seeded teams. By the end of an event, the good teams always find their way to the top. The excitement value comes as much from the game as it does from the robots that participate in a particular round, and seeding has little to do with this. I would be OK with going back to only knowing partners 2 minutes before a match, but would not want to get partnered with a team based on seeding.

Steve Shade

Hypothesis / Test theory...
 

I like to think about seeding this way:

Suppose there is some "true" ranking that the gods on Mt. Olympus hand down (via their oracle at Delphi -- notice how I slipped that greek god reference in there smooth as butter ;-), then the ranking we obtain via seeding match information can be viewed as a hypothesis (or guess) at these "true" rankings.

The seeding matches can then be viewed as tests of the hypothesis.

The job of a seeding and ranking is to produce the ranking that is most likey to be the best approximation (which I will define as the ranking with lowest average ranking error).

Thinking about seeding/ranking this way, perhaps there is a method where every team plays a round of seeding. Then we produce a ranking hypothesis. Then we produce another seeding round to test the hypothesis (perhaps teams with ranks near eachother would be have eachother as opponents, as an example). Then we produce another ranking hypothesis, and so on.

I think that this could produce a better ranking in the end (especially if we could figure out a way to include human information in the rank estimates such as what I proposed earlier in this thread).

There are a number of problems, of course. FIRST would be required to do more work -- they are already overworked and need more to do like they need to have 100 lbs strapped to their backs. Also, I am not convinced that FIRST actually WANTS an improved ranking system -- I think FIRST generally believes there are advantages to baking in some randomness (it keeps high end teams from dominating too much and keeps low end teams from hopelessness).

But... ...don't let that stop us. I think the idea of an improved system is worth thinking about. Perhaps we can come up with a system that not only provides an improved ranking method but also provides other benefits (e.g. more exciting competitions) that are worth implementing.


Joe J.

I disagree entirely.
I like knowing your partners ahead of time. I feel that one can better analyze a match ahead of time, and come up with a better strategy when one doesn't need to think about 2 other matches that could possibly occur.

I'm a HUGE fan of pre-match strategy. In 2000/2001 things were too hectic to do this effectively. Just imagine how 2001 would have run if we knew our partners ahead of time. I think we would have seen a few more big scores.

Another reason...
Remember the "screaming coaches" from 2001 that everyone hated so much? Strategy discussions that were more like arguments? If we go back to the "seat of our pants" strategy, we'll see a lot more of this.

Some people may believe that by only allowing a team 1 coach in the booth, this won't happen, but I disagree. In order to keep the peace, FIRST should keep things as they are. Match planning runs so smoothly now. It is also easier for teams to queue without having to discuss strategies "on the fly".

$.02

John

I agree with John,

Too often a 2-minute strategy session becomes a case of "the team that shouts the loudest sets the strategy". The shouting does not have to be physical either. Sometimes a team's standing in the rankings is enough to make things happen their way. Not that this is necessarily their fault. How can they know the little guy had a better idea if he never says anything about it?

Knowing the match pairings well ahead of time allows time for discussion and consideration (in both senses of the word). On the other hand I don't think it is necessary to know every match for the whole weekend on Thursday morning. So adjusting the match schedule based on rankings for Saturday morning could work. An hour's notice should be enough IMO.

Everyone seems to be talking about scouting being harder and the best teams rising to the top. Being on a team that never broke the top 20 in 2 regionals and championships I question the "best" rising to the top. In both regionals our team was chosen in the 1st round by one of the top 6 teams. Even though ranked around 50th at championship we were still chosen. We didn't have good pairing all the time and didn't run up high scores. We were consistant and could do our job well. Other teams noticed and we fit into their strategy. Were we the best? I guess it depends on what best is.
As was stated earlier, nothing is fair. What we have to do is our best. As someone who has been working for one company for 29 years, I know that the best doesn't rise to the top. I also know that satisfaction comes only from hard work and doing your best. You are not always rewarded or recognised, but sleeping is easy knowing that you have put your all into it.
The question I guess is is this a good idea? There are pluses and minuses on both sides. How about a 4 x 4 contest? better chances of getting more balanced teams. More matches per day. Less pit time for repairs. After this season maybe not less pit time.

unecessary pressure...
 

I agree with those who argue against rushed revelations of partners and opponents.

Concerning the game in 2001, while the idea of not knowing your partners and opponents was novel, in practice it was aweful (IMHO).

It put WAY too much stress on the drive team especially with that year's game which had 4 teams of 5 people each (20 people!) that had to negotiate a strategy under stress in very rushed conditions. Top it off with the fact that one task was HUGE in terms of the final goal (which ment that the fate of all depended hugely on which team was selected to perform that huge task) and you had a system that was all but guarranteed to result in humans behaving badly*.

In any case, even if we don't have the added flaws of the 2001 game, I feel that a reasonable amount of time to plan a strategy is good for the game and therefore good for FIRST.

Joe J.

*I am quite serious about the stress on folks being way too high in the 2001 game. Our drive coach was so stressed out that he had chest pains serious enough that he went to see doctor after the New England Regional. We don't need that kind of stress.

Auto Rank Vs. Human Rank
 

Steve's point about ranking low (50th) but being selected high points to my idea that there is a disconnect between ranks as currenlty calculated and some notion of rank that reflects a team's "true" rank. Obviously the humans in the audience (or at least on the teams the did the drafting) thought your rank should have been much higher than your official ranking.

Humans do not rank teams with weighted averages and such, we are able to incorporate a lot of inputs from a variety of source and somehow (by magic?) come up with a relative ranking of robots.

I think that it would be an improvement to the seeding process if we could somehow incorporate this wonderful ability of humans to rank things on complex scales in a fair way that would not allow other less laudable human attributes (selfishness, pettiness, revenge, etc.) to screw it up.

I don't know if it is possible, but it is an interesting problem.

Joe J.

About only having two minutes to prepare for your match with this new partnering method...
For the last couple of years (and probably since before I got into this) FIRST has tried to schedule the matches so that each team has about an hour of pit time between matches. Say you're team A and you just had a match that team B was in as well. The other two teams that would be in your match - teams C and D - were in the previous match, and in about an hour, there would be a match involving teams A, B, C, and D. Once your match with B had concluded, none of the four teams would have another match until the match of ABCD came up around an hour later. Therefore, the rankings of ABC&D would not change in relation to each other until the match you all took part in. Because of this, you would know which teams were the best, worst, and between those extremes an hour before the match began and you would have an hour to strategize.
I tried to say this earlier, but I didn"t get it across. If this is still too cryptic, someone who understands please clarify it.