[FRC Blog] Frank Answers Friday: Championship Slots

[EricH]

Originally Posted by dodar

If 2 teams are on equal standing, why should where they are from set them apart? If a team is from an area that has more "better" teams, why should that hurt their chances of making it?

I'm referring to the original question. For those that have forgotten, the question was, essentially, "Why are REGIONAL areas underrepresented at Championship?" Not about an area having "more better" teams that don't get to go.

So what I'm looking at is that you take the top X that aren't going, and those that are from an UNDERrepresented area get first crack at going, all other things being equal. It's a hybrid system: top X teams that aren't currently going, ranked 1-X, combined in some way with the underrepresentation status of their respective areas to produce the final priority list on who gets asked.

It's not about being from an area with good teams killing your chances. It's about answering the original question (not the somewhat-related but tangential current main focus of the thread, how do we get all of the best teams to the Championship).

[Mr V]

Originally Posted by EricH

That, I think, would be one of the best ideas. Let's expand towards original intent of the question.

1) FIRST polls experts on best 50 teams to not CURRENTLY make championships. (Trust me, there's a reason this number is so high.) This happens sometime around Week 6.
2) FIRST looks at who still has a chance to qualify (playing Week 7, maybe playing Week 6), or is near the top of the waitlist. These teams are set aside.
3) FIRST looks at geography and representation. Teams from underrepresented areas get higher priority than teams that are from fully represented areas.
4) The combined "best teams" and "underrepresented areas' teams" sorts determine the order of offering--but only to the top 25 on the combined list.
5) After Week 7 plays, the list is revisited. The bottom 25 and the teams that were set aside are now eligible for invites.

There is no waitlist to be "near the top" it is a lottery and they already start pulling names out of the hat before week 6 because the teams that won lottery spots in the PNW last season were notified before the end of our week 6 DCMP.

[Gregor]

Originally Posted by EricH

"Why are REGIONAL areas underrepresented at Championship?"

Not all regional areas are underrepresented.

[EricH]

Originally Posted by Mr V

There is no waitlist to be "near the top" it is a lottery and they already start pulling names out of the hat before week 6 because the teams that won lottery spots in the PNW last season were notified before the end of our week 6 DCMP.

Please read the thread before commenting. We aren't discussing the waitlist lottery. We're discussing alternatives that will try to balance out representation of areas to be more proportional.



Gregor, I'm aware of that. But there are a lot of areas that are under-represented, and they're all regional areas. (And there are over-represented areas, too. Not all of those are necessarily regional areas. But that's another discussion altogether.)

[Ian Curtis]

Food for though, given the current HQ direction we probably aren't that many years away from a time where a teams ability or willingness to pay to attend the Championships could become the limiting factor.

In 2018 will ~40% of FRC teams be able to attend a Championship? If that is followed up with 3Champs, will 50%+ of the teams in FRC be able to afford the trip?

If we sit on this issue for a few years, it just might solve itself.

[Brian Maher]

Originally Posted by dodar

If 2 teams are on equal standing, why should where they are from set them apart? If a team is from an area that has more "better" teams, why should that hurt their chances of making it?

Originally Posted by Kevin Leonard

Any sort of equal representation system based on region will do this, and that's (I think) an unavoidable problem, albeit a lesser problem compared to how championship slots are currently allocated to regionals.

These are questions I've wondered about during my time in a district. Should districts be equally represented? Even if it means competitive teams from a more competitive district cannot attend?

I've thought a bit about this system, and have an idea for how to change it:

• The total spots for all district teams is proportional to the total number of district teams
• Half (or some other fraction) of these spots are distributed based on district team count, just like the current system.
• For the previous year's Championship, district points (or an adapted version of district points) are calculated for all the district teams (who didn't qualify via waitlist).
• These points are sums for each district, and sums are added to find a grand total.
• The remaining slots are allocated based on each district's percentage of the total district points.

This system could potentially work well to adjust for variation of skill between districts. If a district's teams don't fare well at champs, indicating a weaker district, it will send fewer the next year. If a district has many successful teams at champs, it will send more (probably also competitive) teams to champs. The system should adjust itself.

However, it may also get out of balance, over-favoring some districts over time. This could probably be fixed by increasing the fraction of spots awarded based on population to more than one half.

I would run the numbers, but there are many waitlist teams and I don't want to have to figure out manually which teams attended from the waitlist and which ones didn't.

[GKrotkov]

Originally Posted by BMSOTM

However, it may also get out of balance, over-favoring some districts over time. This could probably be fixed by increasing the fraction of spots awarded based on population to more than one half.

That's an interesting idea. Not sure if I'm totally behind it, but it's a new way to look at the problem of district representation.

I don't think that that is guaranteed to work in the long term, because it doesn't address the underlying problem that you send fewer teams to Champs, you get fewer Champs points. Could we perhaps solve that by taking the median* of the calculated scores from each district and comparing them, instead of the sum?

* Or mean; I'm not sure if it'd be good to minimize the effect of outliers.

[Doug Frisk]

Originally Posted by Gregor

Not all regional areas are underrepresented.

Big 60+ team regionals (like all five Minnesota regionals) end up underrepresented.

Thirty to forty team regionals tend to have a higher representation at champs. (The largest Canadian regional last year had 48 teams, the smallest couple only had 30. Not picking on Canada, but based on the maple leaf I'm guessing that was what you were referring to.)

Past that, week 1 regionals are over-represented and week 6 regionals are underrepresented.

That's simply because each regional is allocated 6 slots and a maximum of 3 wildcards regardless of size and once a team has won, winning again at another event can prevent another team from going to St. Louis.

A week 1 regional will always send 6 teams to worlds regardless of size.

Last year, North Star in Minneapolis (week 6) sent two new teams out of 60 teams attending to worlds.

So, what would I do?

Assuming a goal of 10% of regional participants moving to worlds.

Set a "wildcard threshold" of floor(regional participants /10)

So, if the regional has 30 participants the wildcard threshold is 3, if it's 67 it's 6.

Now, at a regional as many as six teams may go on to worlds. If the winning alliance, Engineering Inspiration, Chairman's, and Rookie All-star are all different teams who aren't already going to worlds they all go.

But if those winners or some of those winners are already going to worlds you'd look into creating wildcards.

Let's say that of those 6 winners there were 4 already going (only generating 2 new invitations) either because of duplicates at the event or previous wins.

So, if this is a 30 team regional it would generate 1 wildcard slot, if it were a 60 team regional it would generate 4 wildcards.

The first 3 wildcard slots are easy and already defined, that's the members of the finalist alliance.

Past those first 3, I'd suggest moving down the seedings established by the qualification matches so that the highest seeded teams who weren't already going get invites.

It would still be advantageous to go to a week 1, or 30 team regional, but not quite as much as it currently is.

[jajabinx124]

Originally Posted by DareDad

Past that, week 1 regionals are over-represented and week 6 regionals are underrepresented.

That's simply because each regional is allocated 6 slots and a maximum of 3 wildcards regardless of size and once a team has won, winning again at another event can prevent another team from going to St. Louis.

A week 1 regional will always send 6 teams to worlds regardless of size.

Last year, North Star in Minneapolis (week 6) sent two new teams out of 60 teams attending to worlds.

The 2015 North Star regional sent 967 and 2531 to champs because of the wildcard system, so North star sent 4 new teams to champs: 2491, 5576, 967, and 2531. 10k sent 6 new teams to champs: 2502, 3184, 5515, 2052, 4536, 4198.

[MechEng83]

Originally Posted by AdamHeard

Agreed. I wasn't criticizing the people, it's just that the process isn't currently setup to identify the best robots (as that's not the current goal of the process). Identifying the best robot would require a large shift, or addition of more judges, to allow substantial match view time by more than a single judge or two.

FYI, when there are enough judges, some are actually designated as "match watchers" to feed back some on-field performance observations to the other judges doing interviews in the pits. All judges are encouraged to watch matches as time allows. Your judge experience mileage may vary.

[Doug Frisk]

Originally Posted by jajabinx124

The 2015 North Star regional sent 967 and 2531 to champs because of the wildcard system, so North star sent 4 new teams to champs: 2491, 5576, 967, and 2531. 10k sent 6 new teams to champs: 2502, 3184, 5515, 2052, 4536, 4198.

You are correct, of the regional winners and 3 major awards, only 2 of those 6 teams weren't already going. Which is what I was thinking about, mea culpa.

That generated a possibility of at most 3 wild cards even though only 2 of a potential 6 teams had received invites.

Of those three wild cards allocated to the finalist alliance, 5576 had already won rookie all star so it didn't use one. That essentially left championship spots on the table.

Under the system I proposed, 2 additional championship spots would have been offered to the highest seeded teams who weren't already advancing. That would have been the number 5 seed 3883 (Data Bits) and number 6 seed 2667 (Knights of the Valley).

[Nemo]

Originally Posted by DareDad

You are correct, of the regional winners and 3 major awards, only 2 of those 6 teams weren't already going. Which is what I was thinking about, mea culpa.

That generated a possibility of at most 3 wild cards even though only 2 of a potential 6 teams had received invites.

Of those three wild cards allocated to the finalist alliance, 5576 had already won rookie all star so it didn't use one. That essentially left championship spots on the table.

Under the system I proposed, 2 additional championship spots would have been offered to the highest seeded teams who weren't already advancing. That would have been the number 5 seed 3883 (Data Bits) and number 6 seed 2667 (Knights of the Valley).

Yeah, we should go deeper than finalist alliances to make sure all six of those champs slots get awarded to somebody at that regional. Personally, I'd be looking at semifinalist captains and first picks, with the alliance that got beat by the winners being higher on the pecking order. I like basing it off elimination rounds. But basing it off qualification seed would still be an improvement over losing those slots. It would also be possible to send another award winner, perhaps Engineering Excellence.

[Caleb Sykes]

To be honest, I don't really think that the under-representation in MN is a problem that HQ should even try to solve. The solution, plain and simple, is that if we want a higher representation at champs, we just have to create more events, either within the regional system or in a new district system.

[AdamHeard]

Originally Posted by MechEng83

FYI, when there are enough judges, some are actually designated as "match watchers" to feed back some on-field performance observations to the other judges doing interviews in the pits. All judges are encouraged to watch matches as time allows. Your judge experience mileage may vary.

Yes, but this is only a few judges. You'd want more discussion and input than that.

[Andrew Schreiber]

Originally Posted by MechEng83

FYI, when there are enough judges, some are actually designated as "match watchers" to feed back some on-field performance observations to the other judges doing interviews in the pits. All judges are encouraged to watch matches as time allows. Your judge experience mileage may vary.

That's entirely up to the Judge Advisor/Judge Coordinator. Most think it's important but this is a relatively new change. And, we ran 2x Match Observers at NEU with only 13 judges, it just meant a little extra work for the other judges.

Edit: Also, the big issue with talking about how judging 'is' is that it varies from event to event because there's a fair bit of freedom in how exactly to get things done. In short, there's no One True Way.