[Official 2005 Game Design] Radical Tournament Ideas

[Andrew]

Drawback to Current Tournament Structure
In order to control your destiny, teams have to place into the top 8. Otherwise, you have to rely on a very political alliance selection process to get into the eliminations.

Ranking in qualifying is equally strongly affected by alliance pairings as it is by actual performance.

Teams whose performance (for whatever reason) places them at the bottom of the rankings, have to keep playing, without any real hope of making eliminations. This sometimes incentivizes "bottom of the ranking teams" to engage in wanton brutality.

Tournament Structure Idea
Day One ... Qualifying
This would proceed the same way that qualifying currently proceeds. Random pairings. Ranking proceeds with wins/losses, high score, etc.

At the end of day one, the top 24 seeds are passed on to Round Two. Teams must pass a functional test/reinspection. Failing this would remove them from the seeding.

The remaining seeds are done for the competition. If the bottom teams want to spend day two working on their robot, they can do so without match interruption. If they want to go home and save some hotel money, they can do that as well.

This "cut" is the same as in golf.

Day Two ... Seeding
Scores for the top 24 teams are zeroed; however, rankings are retained.

Teams are not paired randomly for seeding matches.

Round One
Seed 1+2 versus Seed 23+24
Seed 3+4 versus Seed 21+22
Seed 5+6 versus Seed 19+20
Seed 7+8 versus Seed 17+18
Seed 9+10 versus Seed 15+16
Seed 11+12 versus Seed 13+14

Round Two
Seed 1+3 versus Seed 22+24
Seed 2+4 versus Seed 21+23
Seed 5+7 versus Seed 18+20
Seed 6+8 versus Seed 17+19
Seed 9+11 versus Seed 14+16
Seed 10+12 versus Seed 13+15

Round Three
Seed 1+7 versus Seed 18+24
Seed 2+8 versus Seed 17+23
Seed 3+9 versus Seed 16+22
Seed 4+10 versus Seed 15+21
Seed 5+11 versus Seed 14+20
Seed 6+12 versus Seed 13+19

Rankings are determined by win/loss or high score total in these three matches.

Day Two Alliance Selection
Alliance selection proceeds with the top eight after the Seeding matches.

Day Two Eliminations
Eliminations proceed as they do right now.


Advantages to this scheme
Teams that are out of contention can focus on going home or preparing for the next competition.

Teams that are contenders can focus on making the top 24, instead of the top eight. You just have to make the cut to be within striking distance.

Teams that are contenders get more matches. Ie there is an instant reward for making the top 24.

Teams are rewarded for making the top eight in qualifying. They potentially have an easier match schedule at the beginning of day two.

Teams that "luck" into the top eight will have to win against a tougher schedule on day two and will probably not remain in the top eight.

Good teams that are "unlucky" on day one can play into the top eight on day two.

Teams in the top 24 have a smaller field to scout after day one and will make better alliance picks on day two. If you are in the top 24 at the end of day one, you know that you will be playing in eliminations and can prepare accordingly.

Day two is deterministic. It can be scheduled more tightly. FIRST can adhere to an agenda, especially vacating the arena on time.

Disadvantages to this scheme
A good team with exceptionally bad luck can be eliminated without making the cut.

This is a disadvantage to the current scheme as well. It is just not as apparent.

[Andrew]

Another Tournament Structure Idea

Ditch the qualifying, alliance selection, etc entirely.

Maintain the 2v2 format.

Go with a fully deterministic n-loss elimination bracket, where n depends on the number of teams at the tournament.

Start by randomly assigning all teams at the tournament to a position in the opening bracket. Assign byes to each bracket based on the number of teams in the tournament.

The mechanism...
Two teams play against two teams.
The winning alliance moves to the next level in the current bracket, but the alliance members are split.
The losing alliance drops to the next bracket, at the current level of that bracket.

If a team loses n times, it is eliminated.

The winners of the n brackets play into a single elimination tournament to determine the eventual winner. For large numbers of teams, you would have to have fewer than 8 brackets. But, you could terminate the loser's bracket play early. For instance, the winner of the first loser's bracket would be the second seeded alliance and the loser of the first loser's bracket would be the third seeded alliance.

Final alliances are determined "randomly" by how you end up in your bracket. Ie no alliance selections.

Let's say you have 40 teams at an event.

The winners bracket would be 5 levels
40 teams (10 matches)->20 teams (5 matches)->10 teams (2 matches + 2 byes) -> 6 teams (1 match + 2 byes) -> 4 teams (1 match) -> winner 1

The first loser's bracket would be 7 levels
20 teams (5 matches)->20 teams (5 matches)->14 teams (3 matches + 2 byes)->10 teams (2 matches + 2 byes)->8 teams (2 matches)->6 teams (one match + 2 byes)->4 teams (1 match) -> winner 2

The second loser's bracket would be 8 levels
10 teams (2 matches + 2 byes)->6+10 teams (4 matches)->8+6 teams (3 matches + 2 byes)->6+2+4 teams (3 matches)->6+4 teams (2 matches + 2 byes)->4+2+2 teams (2 matches)->4+2 teams (1 match + 2 byes)->4 teams (1 match) -> winner 3

This can be continued into the third, fourth, fifth, sixth, and seventh loser's bracket.

Although this looks complicated (and I probably made a mistake or two), it is algorithmic and can be programmed without much effort. The result being, with N teams and n losses (where n is ideally 8), you can figure out how many matches are required, what teams get byes at each level, etc.

If you play the first round, then the first loser's round, then the second loser's round, etc. Then resume with the second round, the second first loser's round, etc., you automatically get the desirable "time between matches" match spacing.

This tournament structure would have two corollary benefits.

The teams which are going to "lose out" would be eliminated fairly early. This would give them time to work on their robots for the next competition or enjoy the rest of this competition. In other words, if you go 0-8, you'll be all done by about 2:00 on the first day, having played in the first round of each bracket.

On the other hand, the suspense for this tournament would build as you approach the winners of the brackets. The top seed would be spat out first, followed by the second seed, etc.

You could even have a team go 4-0 early in the first day, only to lose the winner's bracket final. This team (4-1) could potentially play (and lose) in all of the bracket finals.

Another wrinkle, to retain alliance selection, the eight alliances could pick a third partner for final eliminations. Since neither alliance partner would be "captain" this would require considerable cooperation between the two.

[Joe Johnson]

Originally Posted by Rob Colatutto

More than 4 robots on the field at a time. No nessesarily all moving though. An alliance in finals plays all 3 teams at the same time but has switches to disable the robots and only two are allowed on a time. 2003 Ex Team A has a ramp dominator and goes up to the ramp, locks into the mesh and shuts down, enabling team C to move around and stack while team B is destroying the other opponents stacks. 2002 Ex Team A starts the match by dashing forward and grabbing all 3 goals and moves into scoring position. Team A disables themselves and turns on robot C who works with B and both fill the goals with balls.

Sick and twisted... ...how about this wrinkle: Alliance B can somehow influence which robot on Alliance A is disabled (and vice versa). I don't know what it would be that alliances would have to do to toggle the disabled button on the other alliance, but it DOES open up possibilities.

Think about it: Maybe if you knock off the ball during autonomous, you get the right to disable one robot from the competing alliance at some point during the competition for 10 seconds. Maybe if you cap a goal you get to "unfreeze" one of your partners (and freeze yourself or the 3rd alliance partner).

I don't know if it is good idea, but it sure is different.

Joe J.

[Billfred]

Actually, as far as I know, all it would take for, a member of 433 (pure example here) to drive our robot this year would be for them to acquire one of our operator badges. Maybe throw in a shirt to keep the eyebrows from being raised.

Although I think there'd have to be some original team control...lest you have teams driving it like they stole it. And wrecking it accordingly.

[Ali Ahmed]

I think it would be more fun if the random computer alliance selection was a little more random instead of it being where the lower number teams were allied with the higher munber teams.

[Astronouth7303]

Originally Posted by Andrew

Another Tournament Structure Idea

Ditch the qualifying, alliance selection, etc entirely.

Maintain the 2v2 format.

Go with a fully deterministic n-loss elimination bracket, where n depends on the number of teams at the tournament.
...

Why not do a 8-loss and keep the current champ format? I think it's cool.

Also, how about random bot selection? meaning: you don't know what bot your driving!

[Documentor]

how about this: a seperate division for regional winners.

[miketwalker]

I like Heidi's idea... I've been playing with something similar to what she said in my head for a long time. It's kind've hard to describe without a game scenario so I'm going to make up a completly hypothetical game scenario just to show my idea. This particular scenario uses a 2 vs 2 vs 2 robot match setup in a circular like field (which I know neither the circle field or 3 sets of 2 robots will probably happen, just easiest to use).

Now, at the beginning of the match you'd have whatever the points are and 3 zones (one for each alliance). The matches would be 3 minutes, at the end of every minute your zones change. In other words, you recieve points at each minute park. This would force teams to get points into a zone and then get them the heck out've there and onto their next zone for the next minute part. This allows for many scenarios too occur, and requires teams to have a good offensive and defensive bot... but would involve more complicated scoring, but would result in alliance partners working together (whereas this year some teams would just go for the bar and the other partner would remain to do the rest... which can work, but there's no cooperation). The only way this could work would be drastically redoing the field and robot setup, but would always keep a consistantly upbeat match being both exciting to viewers... as well as a PAIN when it comes to strategy. Tidbits of this could be incorporated into many of the other drastic changes in this thread already.

But DEFINITLY bring back the time-rush scenario we saw in 2001... although I wasn't around, it is much more exciting to have a time crunch in some way. With robots competing against one another it's highly doubtful they'll all shutoff before the end of the 2 minutes, but if you make something in the scoring happening in time increments, there will be a ton of rushing... but you want to make it dynamic so teams don't completly win a match. If you brought back the boxes from 2003 and mixed them with this zones, yea... that'd be fine and dandy to put all the boxes in this zone... but make teams get them back out (like in the 2003 game, make them have to get the boxes back over to the other side after a minute, but make it simpler to do with fewer obstacles to have more robot interaction). This would encourage teams to try many new strategies (In my mind now, I'm visualizing robots like Team 25's 2000 robot... where in that game they could be dynamic... but if you made it where a arm can't just move things around, it'll create a very strategic game). Just my thoughts. It could be incorporated in several ways. I'd love to see dynamic scoring though.

[Andrew]

It seems that the point of these threads is to throw out all of those details of FIRST that are being done "because that's how we've always done things" and re-examine even the fundamental structures.

So, (as I don my fire proof garments), ...

Why not remove the restriction that only pre-college students can drive the robot, participate as human player, etc.?

Most of the pre-college students participating on a team undertake roles other than on-field roles anyway. There are 3-4 students participating on-field whereas most teams have 10-30 students. Since 3/30 is a very small fraction of the interaction, does this rule significantly increase participation by pre-college students?

On the other hand, any restrictions on who can be driver guarantees that the best drivers are not out there for every team. If FIRST wants the competition to look better for media purposes, then it should put the best people behind the sticks. I'm not saying that a pre-college student won't be the best driver for some teams, just not for every team.

[rachakate]

Originally Posted by Andrew

Why not remove the restriction that only pre-college students can drive the robot, participate as human player, etc.?

I'm going to go the exact opposite way and say, let's make the coach have to be a student too.

[TF8]

Originally Posted by ali_rockon22

I think it would be more fun if the random computer alliance selection was a little more random instead of it being where the lower number teams were allied with the higher munber teams.

You have no idea what you are talking about.

[Andrew]

I'm presuming the reason for this thread is to question everything and figure a way to make the FIRST robotics competition more spectator friendly.

So, why continuous 2:00 matches?

FIRST has somewhat broken up the matches (especially in 2003) by having human player time, then autonomous mode, then remote control.

Although the break up for autonomous mode kind of took the flow out of the game, this was not necessarily a bad idea.

So, why not extend this idea.

Have a series of 1:00 plays (say three) with a :30 robot reset between them.

For example, run 1:00 remote control, then stop. Robots are returned to starting position (but the field remains in its current state) in :30 (otherwise you have to time out or take a delay of game penalty).

Run a 1:00 remote control, then stop. Reset the field.

Run the final 1:00 remote control. Count score.

This would be more like the traditional American sports, football, baseball, basketball. Hence, American viewers would be more able to clue into the game. Also the drama between plays would build.

This would also keep the problem of robots going "wheels up" or getting disabled early in a match or getting entangled from determining the outcome of the match.

By giving :30 between matches, the coaches of the two alliance partners can plan out the next play, rather than having to adapt on the fly.

It would also allow the refs more time to consult and assess penalties or warn teams so that "play outside the rules" would be less likely to affect the outcome of a match.

It would further allow a more accurate "real time score" to be computed.

By allowing time between plays, the audience would have a chance to see the drama of a match build, play by play. If you think about most modern spectator sports, the "stop action" is as important as the action. Even the "continuous action" sports (soccer, hockey) have a kind of "stop action" as the ball or puck transfers from one side of the field to the other.

[sburro]

Her it goes:
I would like to see something moving or changing on the field. I think that the field should always be changing. Also how about a rocky surface? I would like to see a "crater?" filled field. I relize that this would cause dificulties in teams field setup, but imagine the challenge. Can anyone say drive train? Also I think that the compition time is to short.Watching it from the stands, it seems to be over before it starts.

[Billfred]

Originally Posted by Andrew

I'm presuming the reason for this thread is to question everything and figure a way to make the FIRST robotics competition more spectator friendly.

So, why continuous 2:00 matches?

FIRST has somewhat broken up the matches (especially in 2003) by having human player time, then autonomous mode, then remote control.

Although the break up for autonomous mode kind of took the flow out of the game, this was not necessarily a bad idea.

So, why not extend this idea.

Have a series of 1:00 plays (say three) with a :30 robot reset between them.

For example, run 1:00 remote control, then stop. Robots are returned to starting position (but the field remains in its current state) in :30 (otherwise you have to time out or take a delay of game penalty).

Run a 1:00 remote control, then stop. Reset the field.

Run the final 1:00 remote control. Count score.

This would be more like the traditional American sports, football, baseball, basketball. Hence, American viewers would be more able to clue into the game. Also the drama between plays would build.

This would also keep the problem of robots going "wheels up" or getting disabled early in a match or getting entangled from determining the outcome of the match.

By giving :30 between matches, the coaches of the two alliance partners can plan out the next play, rather than having to adapt on the fly.

It would also allow the refs more time to consult and assess penalties or warn teams so that "play outside the rules" would be less likely to affect the outcome of a match.

It would further allow a more accurate "real time score" to be computed.

By allowing time between plays, the audience would have a chance to see the drama of a match build, play by play. If you think about most modern spectator sports, the "stop action" is as important as the action. Even the "continuous action" sports (soccer, hockey) have a kind of "stop action" as the ball or puck transfers from one side of the field to the other.

I like the idea, although I think real-time scoring is good enough as it is from my experience. I mean, the only thing we can't track is penalties, although we darn well try!

The only problem I see with the idea is that you'd have timing and field issues. I mean, we're looking at four minutes of matches, PLUS field reset (which takes longer than :30, trust me), and reintroducing the next set. And if the field's not reset after each phase, then trust me--the field will be disturbed, especially if it's anything like this year. I mean, there's no way you can walk around a field and not bump into a few balls (unless ComBBAT is on the field, of course (wink wink)).

[Andrew]

Originally Posted by Billfred

(which takes longer than :30, trust me)

By field reset, I meant that the players would return their robot to the starting zone. The rest of the field would remain in the state where it ended.

Originally Posted by Billfred

then trust me--the field will be disturbed, especially if it's anything like this year.

Good point. It would depend on what the field was like. I wasn't thinking about many balls or bins strewn about the field. There could always be a penalty for upsetting the field balance.

If you consider that a team cannot get their bot back to start in :30, then they would receive a delay of game penalty. In other words, you should be thinking about getting your robot back "home" between plays. So, the last :10 of each play would probably be robots heading back "home" similar to 2002's game.

This kind of game may also require more human players. The HP role could be purely getting the robot back into position between plays. You might need 5 players in such a case (driver, manipulator, coach, HP reset 1, HP reset 2).

I realize that such a match structure would mean fewer matches per tournament. However, you would have more match time per match. Overall, you would probably get more match time with a longer match sequence than with more shorter matches, where significant time is consumed by setting up the playing field and getting robots/teams into position.