Soooo just some thoughts

[chaoticprout]

Also, top teams with good scouts know who to pick and who complements them the best, even in the loooow rankings (a la regional champions 4 in LA in '06 and '07, 294 in San Diego in '08)

[Frenchie]

Basic economics will tell you that if good teams regularly go unpicked, and poor-er teams get regularly picked, some day some team will boost up its scouting team, put together a killer alliance and win everything. The fact that this hasn't happened yet leads me to think that either the seeding system works so well that the top seeded teams do not really need exceptional partners to win, or that the scouting teams actually do a good job. Either way, this goes against your claim(s).

Also, teams shouldn't design their robot around the alliance. Knowing that a random process will get you your partner, you should design a robot that can win on its own. The fact that some robots go undefeated through the qualification matches either defies statistics or proves that a good robot will not be hurt by those matches. Also, a great robot that hasn't seeded too well (ex. 217 in the 2006 Det. regional) because they missed some matches or ran into technical trouble _will_ get picked in the alliance selection process.

Finally, as far as the whole mentor/student participation thing goes. It really leads nowhere. Many teams are mentor driven that swear they are not, many teams are not who are said to be, and there really is no right or wrong there. It bothers me however to hear people behind the scenes saying "team XXX is 100% mentor built" or "there is no way students did this". Don't assume, you would be surprised. I know for a fact that many of the top tier team see many of their students participate _heavily_ in the design process. I've seen entire systems on a robot put under the responsibility of one student (with the help of other kids to build and as much mentor help as he may ask for obviously...).

That is not to say that FIRST is perfect. Many things are wrong with it and enumerating them will not help because many of you will probably disagree with me.

Francois

[George A.]

I know for a fact (since I helped implement it this year) that most of my team's scouting is entirely statistics based. We have a scout per robot on the field, and it's their job to record the statistics.

For example this years categories are:

Lines in hybrid, balls in hybrid, laps in teleop, herds in teleop, hurdles in teleop/balls placed at end, alliance score, their score. and then a place for notes.

If one team puts up 80 points by themselves in a match, that's really impressive...less so if we find out that none of the opposing robots were moving.

I'm fairly sure that if other teams scouting systems aren't exactly like this, then they're pretty darn close. They look at the bot as a singular unit instead of their alliance.

[benhulett]

After reading jgannon's post about mentor built robots vs student built robots, I do agree with that. I never looked at it that way. I just thought that the majority of the experience gained would be through hands on work rather than watching. I do believe this is a one sided opinion though so I'll take EricH's advice and not go any further into that discussion. But back to the ranking system. Would it not be easier to have a reliable ranking system that brings the top bots forward and reduces the amount of scouting time needed to actually look past the misleading ranks? Because I didn't see much scouting going on to be insanely honest with you all. I'm not saying that our bot deserved this or that, that's not what I'm getting at. I'm getting at the fact that the ranking system shouldn't depend on the outcome of the total match when 1 robot on the entire team did the entirety of the scoring. Although many in this forum have claimed that there are very complex scouting methods out there, I just didn't see it at annapolis this year.

[Ben Mitchell]

Originally Posted by benhulett

After reading jgannon's post about mentor built robots vs student built robots, I do agree with that. I never looked at it that way. I just thought that the majority of the experience gained would be through hands on work rather than watching.

I agree, and this I think is a source of great divergence in philosophy among mentors and teams.

I've been on teams on both ends of the spectrum, and student-built robots are way more fun, and the students have pride and ownership.

Inspiration needs to come from this. Anything else is just a waste of potential. Student's don't learn anything from sending something out to get machined, and learn very little from watching a mentor do something.


On Scouting: I have my team's strategy team do what I did back in 2001: make an excel spreadsheet of all different teams and what they do, including observations. Does it tip? Does it have trouble picking up the ball? Is it slow? Can it hurdle well?

In the end the whole database has a concise matrix of fields on abilities and items of note, and the whole thing can be printed out, then shredded when the next version comes out.

2016 hasn't been in a situation to pick an alliance, but it's good practice, and we use the information in the qualifiers to see what we are up against and who we are working with so that we may adjust our strategy accordingly.

[Joe G.]

On Student build vs Mentor build: I have been involved with both extremes, as well as the middle ground. In these cases, I have found that student built, mentor advised workes best. If both voices can be heard, and "hands on" type learners can learn, then it works out. However, this is not to say that the others cannot work. As a student, I found it boring to just sit there, and as a mentor, I found it distressing to have my opinion on matters I felt I knew a lot about thrown out the window. However, I am sure that other teams on both extremes have had better experiences.

On rankings: I feel that the twice the loser's score system helps encourage innovation. A brick on wheels will score very low here, since they keep the losing score so low. I agree that this system isn't perfect. However, it can be compensated for by good scouting. For example, at our bridge battle tournament this weekend, we seeded #2. We were picked by the #1 seed, who then told us that they had no idea who else they wanted. We had been watching every match, and had observed a rookie team, placed in the mid 30s out of 43 teams. They were there mostly through bad luck, but had a very effective robot. We advised the #1 seed to pick them, which they did, and they ended up playing a critical role on our alliance. Good scouting creates good alliances, as it brings out a robot's true performance, rather than wins, losses and ranking points.

There will always be an element of luck, even if we go back to 1v1v1, dependent directly on the other teams on the field.

Finally, when it comes down to it, rankings don't matter. At said vex tournament, we were eliminated narrowly in the semifinals, through a combination of mechanical problems and strong opponents. However, the day was still a great success, as our robot performed phenomenally, scoring 10-16 balls every match. The six days between our FRC regional and the vex tournament of hard work as a team to get our robot from nothing to that is what really counts in FIRST.

[ks_mumupsi]

I have to say I disagree on the alliance note as well. Our team had some not so perfect alliances at both NJ and Chesapeake regional, as far as going that we are the only robot on the field and so on. But our robot was picked to compete in elim matches in both places.

It really does boil down to scouting, I think 1418 did an amazing job of scouting and putting the alliance together at chesapeake and 694 did the same at NJ.

[Ken Streeter]

Originally Posted by benhulett

The top 8 teams were given the choice of alliances, and only 2 bots outside of the top 24 were selected. I saw much better bots than some of those inside the top 24 that were impeded by team performance but were outstanding performers.

When I saw this post, my assumption, having not been at the Chesapeake regional, was that there must have been poor scouting by the top-seeded teams and they ended up relying mostly upon standings rather than performance-based scouting of the individual teams. However, just to be sure about this, I looked at the Chesapeake regional results to see how that came about.

What I actually found is that there was actually quite a bit of selection of teams outside of the top 24. Actually, there were seven alliance picks from outside of the top 24 seeded teams -- indeed, there was only one alliance where all three teams were seeded in the top 24! It seems to me that there was quite a lot of picking outside of the top 24 teams! Seven teams within the top 24 were "passed over" in the alliance picking process. (Indeed, of the #17 to #24 seeds, the only team which was selected for the elimination rounds was the #21 seed, team 11.)

Having not been there, this outside examination makes it look like the alliance captains must have used scouting information to guide their choices, other than simply looking at the official standings. (See below.)

Team     Seeding

#1 Alliance
836	1
1027	4
1748	39

#2 Alliance
614	2
339	14
359	50

#3 Alliance
1111	3
88	16
888	51

#4 Alliance
2377	5
1218	8
1629	12

#5 Alliance
768	6
293	13
134	28

#6 Alliance
224	7
75	15
1980	48

#7 Alliance
2234	9
341	11
449	41

#8 Alliance
1418	10
11	21
2016	43

Teams seeded in the top 24 that were "passed over"
484	17
869	18
1719	19
204	20
1933	22
53	23
2537	24

Originally Posted by Frenchie

Basic economics will tell you that if good teams regularly go unpicked, and poor-er teams get regularly picked, some day some team will boost up its scouting team, put together a killer alliance and win everything.

Actually, this looks like it may be what happened at Chesapeake, at least from the view of the "official rankings." The alliance which emerged as the champions was the #8 alliance, headed by the #10 seed (Team 1418), which selected the #21 seed (Team 11) and the #43 seed (Team 2016) to be their alliance partners. If you calculate the "seeding total" of their alliance, it was 10+21+43=74. This was the highest "seeding total" of any of the elimination-round alliances. Given the "official rankings" as the only information, one would think they would be the weakest alliance of all 8 alliances. Instead, they emerged as the winners! Clearly, they must have been doing some scouting to put together the champion alliance out of the teams that were passed over at least once by each of the other 7 alliance captains!

However, this analysis does confirm your point that the "official rankings" don't do a good job of identifying the best robots. However, it appears that the alliance captains didn't rely solely upon the "official rankings" to make their picks, but instead tried to find the good teams regardless of their "official rankings." Why your team wasn't picked, I don't know as I wasn't there, but it doesn't look like your being passed over was simply because you finished out of the top 24 teams.

[Tom Bottiglieri]

A few years back, our team was not selected for eliminations.
But, two teams who were not attending the regional were selected.
And one team who didn't even exist was selected.

Intelligent scouting is important. Please, for the sake of the other teams at your events, put a little effort in. If you happen to seed in the top 8, a good collection of scouting data will not only help you to win, but also let the best teams out there show their stuff. It doesn't have to be stats, but at least know the teams.

[GaryVoshol]

I was surprised when watching the Midwest webcast that the #9 seed was never picked for an alliance. The team themselves were not surprised, though. They admitted that the #9 ranking was higher than they actually deserved. A great bit of GP shown by a rookie team.

[Bharat Nain]

Originally Posted by Tom Bottiglieri

A few years back, our team was not selected for eliminations.
But, two teams who were not attending the regional were selected.
And one team who didn't even exist was selected.

Intelligent scouting is important. Please, for the sake of the other teams at your events, put a little effort in. If you happen to seed in the top 8, a good collection of scouting data will not only help you to win, but also let the best teams out there show their stuff. It doesn't have to be stats, but at least know the teams.

This is very true. Also, if you get picked first round and your alliance partner does not have a clue on who to pick second, then your scouting data could come in mighty handy. This happen to us in 2003 and we were National Finalist. Besides, if you don't scout, how do you intend to beat your opponents? Luck can only go so far.

[Alex Golec]

Originally Posted by benhulett

I would vote for more matches to let the ranks even out a bit more to be honest.

Indeed, that matter was addressed in 2005 when the alliances were expanded from 2 teams to 3.

There are many flaws in the ranking system, but we don't have an infinite amount of time nor ability to pair every team with every other team against every other team.

Hence, we learn to live with the system, compete our hardest, and roll with the punches luck throws us. If you are an excellent robot, you WILL be in the eliminations, regardless of your rank. In 2002, 469 was in the bottom 10 of the ranks in their division, but was picked early.

Also, if you look at this from a "global" perspective - do the rankings matter at the end of the season? Will you remember every win, every loss, every score you put up there? Or will you take home some hard-earned lessons about teamwork, cooperation, and inspiration?

-Alex

[jgannon]

I don't know that I've actually written this down before, but I figure this is as good a time as any. I just came to realize this sometime last year, and it has helped me come to peace with a lot of things.

There is no mandate as to who is supposed to be building these robots. If you have any question about this, you should review Dave Lavery's comments from the 2008 kickoff. Because of this, there have arisen two competing philosophies with respect to how we are supposed to meet the overarching goal of Inspiration. One school of thought is that students get excited about engineering by doing it, and thus saying, "wow, I want to do this when I grow up". The other is that students get excited about engineering by watching real engineers at work and saying, "wow, I want to do that when I grow up".

I feel pretty strongly about which method works better, but I've accepted that other people have different opinions, and that there may be benefits either way. So long as the students are actually being inspired, then these teams should be applauded for their efforts.

[EricH]

More matches is something we all wish for. Time is our enemy. Just be glad that the "Algorithm of Death" isn't in use this year.

Trust me, I know about the killer matchups. My old team always seems to draw one of the worst schedules at the Los Angeles Regional. (Translation--you're the only good team on your alliance, and you're facing 968 and 980 in the same match. Or 69 and 980. Without the 2007 algorithm. Even 330 can't do that too well normally. One-on-one, yes. Not 2 or 3 on one.) Or something would happen and it would be a one-on-three. Stuff like this happens. The only thing you can really do is play the best game you can. If you do well, some team that's been doing their scouting will notice, even if you don't make the top 8. If teams haven't been doing their scouting, help them.

Strength of schedule is already built into the ranking system--It's the RS column. (QS= Win-Loss-Tie, RS=strength of schedule, and if those can't solve a ranking dispute, the higher scorer wins.)

Simply put, good scouting trumps the ranking system any day. Bad scouting is using the ranking system to do your scouting.

[Nawaid Ladak]

i don't want to get bashed so i won't post my comments on this on CD. I don't think anyone will really care 2 years from now... because there probably will be another thread and im SURE i've spoken on this before...