In Rapid React, Week 1 comeptetiveness was much higher than in years past, everyone was moving and running much faster than anticipated. Some believe this was because the game was really similar to 2020/21 so most of the core design was the same. Others believe this was because of some reset that happened coming out of covid that allowed teams to regroup and improve dramatically. Do you think 2023 will continue the trend of higher comptetiveness?
As long as COTS assemblies continue to grow in availability and teams continue to purchase them I think we can expect greater levels of competitive performance in earlier events. This of course assumes the COTS stuff is available before the early events
I don’t think 2023 will be quite as competitive. Apart from being similar to 2020, rapid react in general was an easy game without many complexities compared to other games in the past
- The cargo was extremely easy to intake and index and almost never jammed.
- Shooting high wasn’t that difficult either, two rollers and the balls will make it in and stay in, while adding a long distance shooter + a decently tuned limelight resulted in being able to solo the cargo RP. The cargo was also fairly consistent which means not much shooter effort beyond having a top roller is needed to for consistency.
- The endgame, while traversal was the hardest endgame ever in FRC, there has been climbs every game since 2016 which means teams know how to climb. And climbing mid was extremely easy especially with COTS kits.
- Scoring without defense was also easy for drivers. The field was wide open and not much accuracy was needed to intake and score cargo (especially with limelight) meaning that drivers can drive extremely aggressively with little to no consequences.
It is sort of interesting, my general feeling was that the week 1/2 regionals between 2020-2022 weren’t too different in competitiveness. So I was trying to think of a metric to be a proxy for that, and I calculated the top 5 OPR over top 15 OPR from TBA. The results were pretty similar for the regionals we went to those years (MOKC, OKOK) with only a very minimal (0.44 vs 0.43) change between those values.
Anyhow, don’t really know what to expect, or maybe a better measure to use for comparing teams.
Top five OPR seems a reasonable metric but perhaps not.
If you consider MN and WI to be average kinda places I can say that the Week 1/2 Duluth events had a higher level of “working tech” last year. Maybe the Top Five teams were about as amazing as usual. But the number of robots with major systems non functional or even no shows for early matches was down. I’d think the OPR for the middle and bottom quartiles would be better with the elite teams being, as usual, elite.
Again just my perspective from the area but I also thought the level of improvement from start to end of the competition cycle was less impressive than some years. With COTS parts maybe reaching the mechanical level your team is capable of attaining happens sooner? You’d still have driver improvement but those non elite teams that barely got running for Week 1 at least got a bit of drive time in…and perhaps the improvement curve starts out steep - as you debug your robot - and eventually levels off?
These are just my impressions from being at the two Duluth events in week one, the Milwaukee event in iirc 4, and being a ref at Lacrosse. Your mileage may vary.
I think/hope it will, part of the reason I think it will be more competitive is that a lot of teams lost people during COVID and a lot of rookies joined. Now that people have been trained with last year I think we will see more competitive robots.
I am also hoping for a game with very little design convergence unlike Rapid react.
Gonna handle this in parts because I have different opinions about different parts of this post.
What are you using to judge this? “Recent” years I have competed in Week 1 are 15, 16, 17 and 22. However I also competed in Week 2 in 17, 18 and 22. In my opinion teams were more competitive than they were in 15, and 17 but less so than in 16 and 18. I don’t have a good metric to judge this by as “competitiveness” is subjective and it is tough to compare.
This I will agree with, I was pleasantly surprised how few teams had completely nonfunctioning robots at our events this year.
This helps when you look at games where core design elements can be copied it makes life easier, especially when they are so close together. Going from 2012 to 2013 many teams just took their Basketball shooter and turned it sideways to test Frisbee shooting and for many that was basically what they built. 2018 to 2019 many teams either reused the same elevator/arm, had caded a new one for practice in the offseason or copied one of the very good ones from publicly available resources. YMMV but in my opinion it was the correct choice to stick with a “grab ball, shoot ball” type game coming back from the pandemic to give teams this advantage.
I have a much more morbid take on this. In Florida at least most of the teams in FRC who did not survive the Pandemic were less competitive overall historically than those who did. There are exceptions (1065, 1251, and 2152 spring to mind) but for the most part we did not lose winner contenders we lost teams who traditionally ranked in the bottom half. When you take away 1/3 of the teams and at least 2/3 of that group happen to be “below average” your average increases.
This is tough. A few teams will be coming back such as my earlier mentioned 2152. These teams will need to shake off the cobwebs the rest of the group did last year. We also most likely are going to have a pick and place game over a shooting game although this is not certain. Pick and place also historically are tougher “rp” game than shooting.
Comparing RP that could be secured whenever via game piece scoring (aka not strictly auto or endgame dependent) in 16 and 22 vs 17 and 19 shows some interesting stuff.
In 2016 you received 1 RP for scoring 8 boulders (you also all had to park under the tower but for this discussion that is not important). This meant 3 good teams could come out of auto with 3/8 of the RP completed if everything went well however it was much more likely only 2 boulders were scored so it was closer to 1/4. While 2 boulder solo autos were possible the vast majority couldn’t do them so 1/4 to 3/8 or 25%-37.5% done seems fair.
In 2022 only 1 RP was available this way and it for for scoring 20 cargo*. A good auto would result in 5 cargo scored. This would mean that you were 1/4 or 20% of the way done however because of the quintet bonus lowering the RP threshold you actually were 5/18 or 27% done with this RP.
With the pick and place games however
In 2017 we are going to ignore the Pressure Auto as it involved shooting and was achieved in less than 3.5% of matches. However the 4 Rotor RP involved scoring 13* gears. A “very good” auto ended with 2 of these gears scored meaning you were 2/13 done with the RP however there was a “free” gear that could be used dropping this down to 2/12 or 17%.
In 2019 you had the Rocket RP. This involved scoring 6 hatch panels and 6 cargo in the rocket. Most “good” autos involving the Rocket would end with 1 hatch panel on it or 1/12 (8.3%) while “very good” autos could get 2 hatch panels this was very rare.
This shows me that FIRST always feels that pick and place games are “easier” so they make how much of the RP you can achieve in auto less than in shooting games since they think most teams will be able to score in Teleop. Well in 2016 8 boulders were scored in around 35% of qualification matches (this data is hard to come by since you can not just use the RP data thanks to the end game requirement) and in 2022 the Cargo RP was achieved in 34% of matches. Compared to the 8% of 4 rotor matches and 4.74% of completed rockets it shows how bad FIRST is at balancing the RP challenges in Pick and Place compared to shooting. This is partially due to how much easier RP shutdown defense is in Pick and Place but once again that comes down to FIRST game design.
So if you judge competitiveness by “how well they accomplish game objectives” I worry from historical precedence that if we have a pick and place game statistically the average robot will be “worse”.
I’m not sure I would assess the quality of robots based on how GDC happens to assign the bonus RPs. If 2022’s cargo bonus had been 40/36 in instead of 20/18, would that mean most of last year’s robots were suddenly “much worse”? Or just that the game was designed so that getting the RP would require at least 2 really good robots?
If we judge competitiveness by which alliances won, then it seems like 2022 was maybe actually less competitive than some previous years (wasn’t it like week 6 before somebody other than the #1 or #2 seed won a regional?).
I think it depends on where you are. In the Ontario District last year, many schools were online only for the first part of January, and many teams were not allowed to meet in person until the end of January. Restrictions may have prevented outside mentors from coming in for a bit also. I think this held back competitiveness quite a bit, especially at early events. The smaller, single-day events also made problem solving between matches difficult. I expect to see a higher level this year, though there were quite a few teams here who didn’t compete at all in 2023, so it may be a bigger challenge for them.
2022 definitely saw an uptick in the “competitiveness” compared to previous years, but I don’t think it’s because of the reasons that OP pointed out (an “easy” game, COVID “team reset”). However, to examine that, we first need to ask “What makes a season competitive?” because I don’t think it’s necessarily obvious.
Every year will see teams of varying skill levels. One part of “competitive” is the absolute skill level of teams across the board. Relative to some benchmark, are teams achieving the game tasks as well as we thought they would? This, generally, is something that will rise over the course of the season.
Another aspect though is the relative skill level of teams in the competition. I think an event with many, equally skilled teams vying for a top rank would be described as competitive, whereas an event with clear domination by a team or two might not be. A competitive game/season/event is some mix of these elements.
The second aspect of relative strength is something I think we saw a marked change in for 2022: the rise of mid-field team strength. In 2022, the 95th percentile team was much stronger when compared to the 99.9th percentile team than that 95th pct. team was in previous seasons. In addition to the concentration of strong teams at a smaller championship this year, I think this was one of the reasons we saw the strongest teams struggling to rank highly: With a stronger mid-field, match schedule becomes more important for locking in the very top seeds, as it becomes more unlikely that you’ll be able to overpower a weak schedule and beat an alliance of strong, mid-tier teams.
The result is my guess for the change we’re seeing: no bag
The best teams were never really limited by the bag. When you’re building two, three, or more robots in a single build season, having to bag one up for competition doesn’t really place a big constraint on what you can achieve on the field, apart from some (sometimes large) logistical challenge with implementing changes at the event.
The teams that were really limited by the bag were the mid-field teams. Teams that can be strong, but are held back by limited funds to purchase hardware for multiple robots, or held back by people/time to implement big changes at an event. These teams wanted to make changes that actually would have helped, but couldn’t. These teams now can continue to work on their one robot throughout the build season as opposed to essentially being (or feeling they were) locked into the robot they bagged. The top teams stopped feeling the bag day limits long ago. The newer teams were (and still are) facing challenges harder to overcome than the bag. The mid-field are who really benefited.
I think it’s easy to forget, but the last full season we played out was 2019. We were bagging after 6 weeks and every event (but the championship) in 2019. 2022 was the first time we really saw “no bag” play out for a full season, and these effects will continue to play out in 2023.
I regret that I only have but one like to give…
As a mid tier team… When we started building a second chassis, the act of building the second chassis took away the bandwidth to actually test a new mechanism on the second chassis before we did the field upgrade.
No-bag is a huge improvement and evolution for the program.
Combination of no-bag and the explosion in quality of open-source tooling makes me think the competition’s going to be wild this year and is only going to get wilder.
There’s still a major issue of program-scale parity that isn’t gonna just go away - teams that don’t have the resources to truly compete will probably remain mostly where they’ve been for the past decade-plus. But from the mid-tier up, it’s going to be bananas.
TBA has an excellent way to look at this. 2022 Insights - The Blue Alliance
I took 2022 Week 1 vs week 5 (last week without championship events mixed in) here is the takeaways that stand out to me. Percentages rounded down for ease.
- Taxi goes up 12% in Quals from 73% to 85%, but only 1% in Playoffs 89% to 90%. I never imagined Playoffs would be only 90%.
-Quintets went up 12% to a max of 17%
-Cargo RP went from 14% to 36%
-Hanger RP went from 27% to 44%
-Average Cargo went from 20 to 32 (Ave count was 10 to 15)
- Average score 38 to 54
Compared to 2019 Week1 vs Week 5:
- Rocket RP 2% to 5%
- HAB RP 26% to 42%
- Average scores went from 41 to 53
Biggest takeaway, Rockets seems like a much harder RP, also was probably more defendable and more congested. If you could score the 12 elements needed for a rocket, 12 elements in 2022 makes an incredible dent in the the Cargo RP.
The climbing RP is nearly dead on the same success.
From week 1 to 5 the difference in points was actually LARGER in 2022, which would make you think Week 1 was further from peak performance. I think part of this may have to do with sandstorm instead of autonomous making more teams active in the first 30 seconds.
Also widely available COTS swerve will make for a lot of people going sideways especially if the game is somewhat open.
Y’all ain’t seen nothin’ yet!