FRC Blogged - What Do You Think? The ‘Invite to Decline’ Strategy

[Wayne Doenges]

About the only way you could stop the Scorched Earth strategy is to not allow the top 8 team to pick any of the other top 8 teams
Now that would make some interesting alliances.

[dodar]

Originally Posted by Wayne Doenges

About the only way you could stop the Scorched Earth strategy is to not allow the top 8 team to pick any of the other top 8 teams
Now that would make some interesting alliances.

But then you would have teams "fighting" for 9th place.

[nicholsjj]

Originally Posted by Wayne Doenges

About the only way you could stop the Scorched Earth strategy is to not allow the top 8 team to pick any of the other top 8 teams
Now that would make some interesting alliances.

This is what the offseason event in Kansas City does. It makes for an interesting alliance selection in that teams pick based on knowing who they have to play. This year it might be neat with the cttd having two division finalists and another frc top 25 team competing.

[Chris is me]

There's no way to prevent this strategy while still preserving the right to decline.

That said, I haven't the slightest of problems with it. It makes eliminations far more interesting every single time it happens. It provides the top seeds with an important additional incentive that compensates for the disadvantage given to them by the serpentine draft. Perhaps the only thing that bothers me about it is that it's a little awkward.

If you've seeded in the top 8, you should have the right to form your own alliance. Imagine if this wasn't the case. Imagine 6-7 FIRST teams openly lying to the first seed about the state of their robot to try and avoid getting picked. Imagine the off-field deals this would incentivise ("if you don't pick us, we won't pick team xxx that you want on the back half of the draft"). Even more teams would more actively hope to not be on an alliance with someone.

I mean, I guess scorched earth should be avoided, in the sense that we shouldn't have a seeding system and number of qual matches that commonly allows carried teams to seed first / second / third. However, to at least some small extent this is unavoidable. Even in 2010 there were teams with better schedules than others (and, as good at ranking teams as it was, not a ton of people want to go back to 2010 for some reason).

It's also worth noting that not all declines are because the chosen team thinks the top seed sucks or whatever. The serpentine allows for better alliances at lower seeds sometimes. In addition, almost every decline in Canada last year was done in order to play at a seed that faces #1 in the finals, in order to earn wild card slots.

[BrendanB]

Even if you add more matches it will "help" but it won't solve the problem. That's if you would even call it a problem. Over the years we have been given more matches but even then there are still teams who slip into the top 8 who aren't the best and top teams who have rough matches.

Rankings is mainly determined by performance. I'll use our team as an example. Our team had the same record as team 2648 at the Pine Tree Regional (12-1) the difference came down to our autonomous points. Our team had two matches we missed all of our shots due to different reasons. One match we had an air tank pushing down on our plate causing all shots to go low. Another we had a driver station error that prevented us from selecting the proper speed before a match. In the one match we lost we had one partner who was struggling with shooter problems and another who lost all controls to their drivebase.

In the end we would have selected 2648 if we seeded higher but the story holds true in that it didn't matter who had the better robot, it mattered in who could pull it off every match. 2648's autonomous and teleop game were dead on the entire weekend which gave them the number 1 spot. Similarly 125 also had one of the best robots at the event but seeded #20 after a tough match schedule. 1114 was one of the best in the Archimedes division last year but they finished 66th after some issues and were the second pick.

If anything I wished we had less qualification matches. Friday we had 9 matches some with only a 4 match turnaround leaving no time to fix problems.

Unless you have a panel of judges who strictly make the rankings based on robot performance, you will never go into alliance selections with teams ranked where they "should" be.

I like the system the way it is. No 1-8, 1-8, no picking in the top 8, etc.

[Woolly]

Originally Posted by nicholsjj

This is what the offseason event in Kansas City does. It makes for an interesting alliance selection in that teams pick based on knowing who they have to play. This year it might be neat with the cttd having two division finalists and another frc top 25 team competing.

Let's not forget 3928 who has shown to be quite capable at being a game changer for an alliance. If any one of these 4 teams gets a hard schedule or (unfortunately) decides to throw a match or two, you could end up with alliance partners that look like:
3928-1986
3928-1806
3928-3284
3284-1986
3284-1806
Or even the feared and well practiced combination of 1986-1806.

That's not counting other high-scoring bots being in the mix like 1939, 3528, etc.

But even so, throwing a match is harder than it sounds. If it's the last match of the day and you're paired up with someone who's decent at scoring and has a vested interest in making sure you don't fall outside the top 8 and are suddenly allowed to form a super-alliance, you more than likely aren't going to be able to throw it unless you're playing against the team you want to be with in the first place.

[Kevin Leonard]

I honestly love thinking about alliance selections.
"If they pick them, then they need to have a full-court blocker, so they'll pick XXXX, but they also have a ______ autonomous, and XXXX needs to counter that sooo..."
And then:
"But if that match goes that way instead, then XXXX will seed first, and XXXX will decline them, then they can pick 20 later in the draft!"

Not that that scenario happened to us or anything. Nah.

[Caleb Sykes]

I actually think that a 1-8, 1-8 selection would help to discourage scorched earth. 1-8, 1-8 helps captains 1-3, has no effect on captains 4-5, and hurts captains 6-8 when compared against the serpentine draft.

In a serpentine draft, seeds 6-8 are hesitant to accept the invitation of a mediocre #1 seed because the remaining partner will be the 16th pick. (Using seed 7 as an example) If the 7 seed declines, they still get the 7th and 10th picks.
In a 1-8, 1-8 draft however, if the 1 seed invites 7, 7 is much more likely to accept. 1's&7's second pick will be #9. If 7 declines, in contrast, they are stuck with the 7th and 15th pick. Assuming minimal difference between pick 7 and 9, 1 only needs to be marginally better than the 15th pick (could be as low as 23rd overall robot) for them to accept. Not to mention an easier schedule in eliminations.

Since scorched earth strategies rarely (if ever) come from a seed lower than 3, this would definitely limit the occurrences of scorched earth.

[EricH]

Originally Posted by inkling16

I actually think that a 1-8, 1-8 selection would help to discourage scorched earth. 1-8, 1-8 helps captains 1-3, has no effect on captains 4-5, and hurts captains 6-8 when compared against the serpentine draft.

Speaking as someone whose team won a World Championship (from #2) during the last year 1-8, 1-8 was used, there is a definite downside. And that is this:

Blowout city. Think about it. 1st seed gets 1st and 9th picks; 2nd gets #2 and #10 picks; 8th seed gets what's left at the #8 and #16 pick slots, which can get pretty thin at smaller events, particularly the #16 pick (and the third robot can make or break an alliance). When divisional QFs set the world high score, and it's not close even with each alliance having its own set of game pieces...

Now, I don't know about you, but I don't necessarily like a blowout. I'd rather have both sides be able to get the win, just one plays a bit better and wins, like the Einstein semis/finals that year. With the serpentine, it's any alliance's game to win, at least in theory. With "straight", it's just a bit easier for the higher alliance to win.

[SoftwareBug2.0]

I think the people in this thread are generally on the right track: Nobody should be suprised when a team does what's in its best interest. In my mind, the most important condition for the alliance selection method is that teams are incentivized to perform well in the qualification rounds.

The current setup has worked pretty well, but there is at least one case I haven't seen mentioned yet where it would break down. With the recent games, getting to pick 1st and 16th has almost always been better than 2/15 which is better than 3/14 and so on down to picking 8th and 9th. That's true because the performance gap between the very best teams is higher than the performance gap between average teams.

Assume for a minute that teams score a constant number of points per round and do nothing but offense. The top 5 teams might score 70, 60, 52, 46 and 40 pts while the 20th through 24th teams might score 20, 19, 18, 17, and 16 points. In that case, you can see that having the 2nd best robot on your alliance is worth 60-52=8 points per round more than having the 3rd best robot on your alliance. Meanwhile, the choosing last vs. next to last is worth only 17-16=1 point per round. Since where you pick in the first round makes a bigger impact than where you pick in the second round, you want to be a higher alliance captain despite the serpentine.

However, if the top 16 teams had very similar scoring potential while there was a big drop off in performance between the 17th best robot and the 24th best then it would be better to have the 8th and 9th picks than to have the 1st and 16th because picks 1 and 8 are interchangable while pick 9 is much better than pick 16.

I think having a game where this happens isn't outside the realm of possibility. I would have to look at some historical data to know for sure, but I have an inkling that there might have been more differentiation in the second round than the first in 2005. Edit: Though at a quick glance it appears this was not the case

[GaryVoshol]

If this is a problem - and I'm not sure it is - what about this solution? If any team is declined X times (X = 2, 3, whatever) they lose their pick until the end of the round. Now the earth they are scorching may be their own!

I've seen two instances:

1. A team wasn't really thrilled with playing with a top seed, but agreed to accept if the seed would first pick others in the top 8 who presumably would decline. This is the true scorched-earth scenario Frank presents.

2. A team ranked #1 based on the strength of its schedule in quals. Other teams thought the #1 alliance had no chance, and so they were declined multiple times. I don't recall for sure, but I think they finally picked someone outside the top 8 who had to accept or risk not playing at all.

[Justin Montois]

I think that the alliance selection process is perfect.

"Scorched-Earth" strategies are just as beautiful to me as making a dark-horse pick that truly pays off or creating a powerhouse alliance.

We all play qualification matches to try and seed #1. The team that does deserves to pick their alliance however they see fit. Likewise, the teams below them have the same right.

Watching that dynamic play out is fascinating and addicting. I think it's the greatest 20 minutes of every event.

[Ian Curtis]

Originally Posted by GaryVoshol

If this is a problem - and I'm not sure it is - what about this solution? If any team is declined X times (X = 2, 3, whatever) they lose their pick until the end of the round. Now the earth they are scorching may be their own!

I've seen two instances:

1. A team wasn't really thrilled with playing with a top seed, but agreed to accept if the seed would first pick others in the top 8 who presumably would decline. This is the true scorched-earth scenario Frank presents.

2. A team ranked #1 based on the strength of its schedule in quals. Other teams thought the #1 alliance had no chance, and so they were declined multiple times. I don't recall for sure, but I think they finally picked someone outside the top 8 who had to accept or risk not playing at all.

I would presume the 2nd scenario is more prevalent, but I don't know that. Does anyone know what the ratio is? I also think Pinecone's suggestion that teams decline to play for a wildcard is an interesting development. It makes perfectly good sense against an overpowered #1 alliance to opt for the other side of bracket, its just another weird strategic twist in how FRC tournaments play out.

[apalrd]

Originally Posted by Wayne Doenges

About the only way you could stop the Scorched Earth strategy is to not allow the top 8 team to pick any of the other top 8 teams
Now that would make some interesting alliances.

There is one expo tournament we've played several times (the IGVC invitational) with exactly 24 teams, where teams are assigned alliances (1-9-17, 2-10-18... 8-16-24). Saturday morning there is sometimes a strong incentive for a team thats 7th or 8th to lose a match to drop to 9th-12th and get a better overall alliance.

If you disallow picking within the top 8, teams will do the same thing to get below 8th to allow themselves to be picked first.

OCCRA does a similar thing (no picking within the top 6). There was a team this year who was 4th going into their last match and planned on throwing it to drop below 6th. Another team (I don't remember their exact placing) who was opposing them in this match knew this and decided to play for the opposing alliance to prevent them from throwing the match and dropping below the cutoff to be pickable (the outcome did not affect this team enough for them to be concerned). In the match, both teams in question scored points for the opposing alliance, so they were literally both trying their hardest to increase their opponents score. The fans were shouting "YOURE BLUE YOUR BLUE" and "NO YOUR RED" and it was really strange. You would see the same exact thing happen in FIRST if picking within the top 8 was forbidden.


I have no issue with scorched earth plays as far as the rules or GP is concerned. The 1st seed has earned their spot by the rules of the game. If we want the best alliances possible (which I think we do), we just need to play more matches so the 1st seed is really the best (or very good) and the lower seeds would have little reason to decline. Changing the rules in any way I can think of would just make everything worse.

Originally Posted by Chris is me

Even in 2010 there were teams with better schedules than others (and, as good at ranking teams as it was, not a ton of people want to go back to 2010 for some reason).

The 2010 seeding system was quite good when everyone played to win IMHO, but I think a lot of people still remember the 6v0 plays and scoring for opponents to increase your seeding and dislike it for that reason. And before the +5 in a team update there was no reason to score if you knew you had no hope of winning (this was hotly debated, but we did play a 6v0 once successfully, and talked it about at least two other times).

[PVCpirate]

I thought it was great that 4 teams declined 1678 on Curie, who responded by winning the division. Alliance selection drama is something I always find myself talking about after an event. I really don't think the rules should be changed, and I think any attempt to do so would make selections needlessly complicated.