• Welcome to the Speedsolving.com, home of the web's largest puzzle community!
    You are currently viewing our forum as a guest which gives you limited access to join discussions and access our other features.

    Registration is fast, simple and absolutely free so please, join our community of 40,000+ people from around the world today!

    If you are already a member, simply login to hide this message and begin participating in the community!

Proposal: Tie-breaker based on time to submit attempt in FMC single

abunickabhi

Super Moderator
Staff member
Joined
Jan 9, 2014
Messages
9,035
Location
Yo
WCA
2013GHOD01
YouTube
Visit Channel
They should start ranking FMC in a similar way because there are so many ties ranked on moves.
If there is a tie then the attempt should be ranked on time, similar to how MBLD is evaluated if the points are tied.
This was proposed by Aedan about 6 years ago.
Sorry for the delay to bring this proposal to discussion but it does have some merit and should be given serious thought.
 
That would be an interesting change. Including time as a factor (aside from the standard 60 minute time limit) would bring a completely new element to the event; I wonder if the additional pressure would increase average results, as one might allow themself less time to check optimal solutions in order to account for time. I also know people who do think and perform better under more pressure. I'm sure the top FMC solvers will still do excellent officially, but I do wonder how it would shift around typical results.
 
Would this not encourage gambling and rushing to submit a "good enough" solution to tie with everyone else, when you could potentially find a better solution with the remaining time?
The first preference is always given to less move count, so similar to how we do speed FMC in SS weekly, we try to not gamble and look for a better solution investing more time.
 
That would be an interesting change. Including time as a factor (aside from the standard 60 minute time limit) would bring a completely new element to the event; I wonder if the additional pressure would increase average results, as one might allow themself less time to check optimal solutions in order to account for time. I also know people who do think and perform better under more pressure. I'm sure the top FMC solvers will still do excellent officially, but I do wonder how it would shift around typical results.
The results will look more or less like how speed FMC look in SS weekly, the solutions submitted are less optimal than the solver's potential, but there is a smartness attached to the best time to submit a solution without investing more time to shave off a move or two.
 
I don't think the public database contains this information for past solves. I wonder if the WCA Software Team has an internal database with this info

If the info isn’t available for previous results, how would that be handled moving forward? Would it just be a new category like we have for multiblind old style and multiblind new style?
 
What about just making all old results 1:00:00?
it would be super unfair to put all the old results at such a disadvantage

I asked some folks in fmc discord and people are very against the idea of use time for tie breaks

the reasons are:
  • it would involve more luck, so if you find a good solution early you get a better score than someone who found the same solution later
  • it would change the spirit of fmc from "find the fewest moves" to "find few moves as fast as possible"
  • more work for judging and delegates. it would prevent judges from participating in the fmc round
 
I think time is the obvious candidate to break ties, but I also don't think it's the best for reasons people have already said. We also can't really use alphabetical order, because it is very arbitrary. Two solutions come to mind: firstly, lower single in the mean (for example, 20, 22, 24 would beat 21, 22, 23) or another solve. The first one has the problem that a draw in the lowest solve is by no means unlikely, plus you could argue that a lower variance is better. The second one takes an extra hour of competition time, plus some. Basically, I don't think there are any good solutions.

The other thing to consider is that this really isn't a problem. Yeah, it happens a bit, but people don't really mind. So think about solutions, but we only need to think about implementing some when it becomes a problem.
 
you could argue that a lower variance is better.

We might indeed want to reward the ability to produce the best solutions on a regular basis for the duration of the event. With this in mind, it would be interesting - surprising as it may seem - to directly compare the competitors on the worst of the 3 scores, and to break the tie on the second, then the best.
Thus, 19 20 21 beats 18 19 22, and even 17 18 23, because 21 beats 22 and 23.
I'm not saying it's a better system than the existing one, just a different way of looking at things.

Out of curiosity, and assuming that the competitors' current results are also their best according to this approach, I have drawn up a ranking of the world's best. There's no doubt that some people wouldn't like it at all. 😁

Nevertheless…
the-wire-the-king.gif

Code:
 1 ←︎  1        Wong Chong Wen        20 21 19 → 21 20
 2 ←︎  3        Jan Bentlage        21 21 21 → 21 21

 3 ←︎  3        Jayden McNeill        20 21 22 → 22 21 20
 4 ←︎  9        Louis-Marie Ratto    21 22 21 → 22 21 21
 5 ←︎  3        Wojciech Rogoziński    19 22 22 → 22 22 19
 6 ←︎ 13        Jack Love        21 22 22 → 22 22 21

 7 ←︎  2        Firstian Fushada    19 23 20 → 23 20 19
 8 ←︎  3        Radomił Baran        20 23 20 → 23 20 20
 9 ←︎  9        Yunhao Lou        23 21 20 → 23 21
10 ←︎  3        Cale Schoon        23 18 22 → 23 22 18
11 ←︎ 13        Daniel Karnaukh        22 20 23 → 23 22 20
11 ←︎ 13        Max Parris        23 20 22 → 23 22 20
11 ←︎ 13        Krzysztof Pietrusiak    22 23 20 → 23 22 20
14 ←︎ 22        AJ Kamal        22 21 23 → 23 22 21
15 ←︎  9        Jan Riedl        18 23 23 → 23 23 18
16 ←︎ 22        Alexis Le Merrer    23 23 20 → 23 23 20

17 ←︎  9        Sebastiano Tronto    20 24 20 → 24 20
18 ←︎ 13        Carter Kucala        20 24 21 → 24 21 20
19 ←︎ 22        Alexei Sinyavin        21 24 21 → 24 21 21
20 ←︎  3        Kyeongmin Choi        24 17 22 → 24 22 17
21 ←︎ 13        Krish Shah-Nathwani    19 24 22 → 24 22 19
21 ←︎ 13        Jaye Sloan        22 24 19 → 24 22 19
23 ←︎ 13        Mauro Ortega López    24 18 23 → 24 23

?? ←︎ 13        Yiwei Liu        25 23 17 → 25 23 17
 
Last edited:
So I understand the idea behind this but strongly disagree. This proposal fundamentally changes FMC as an event. Why? Here's a hypothetical =>

There are 2 competitors (1) Jerry and (2) Terry

Jerry and Terry both get a final solution of 24 moves. Both Jerry and Terry find their 24 move solution in 20 minutes. Jerry feeling that he has found the best solution submits it. Terry on the other hand feels that there is a better possible solution. In this case we KNOW that there is a better solution as neither of the solutions are optimal, given God's Number. Terry spends the remainer of his time looking for a better solution but is unable to find one. In this scenario Terry is punished for looking for a more optimal solution.

So why is this unfair? I said earlier, this proposal, within this specific scenario, changes the event from FMC to Speed FMC - these are NOT this same event. Changing the parameters of the event, in a case of a tie, specifically on the terms of time doesn't make sense considering that the entire event is NOT based on time.

A possible better solution would be to measure each solution in both HTM and QTM (or some other turn metric). In this case the winner is rewarded for have an overall more efficient solution - the tie breaker is now based in the parameters of the event - efficiency. This, however, doesn't really work. Here is another hypothetical =>

Again, there are 2 competitors (1) Jerry and (2) Terry

Jerry is a newer competitor in FMC and, as such, uses blockbuilding to reach his solution. Terry on the other is more experienced and uses DR. As in the last scenario both Jerry and Terry reach a final solution of 24 moves. To break the tie between the two we use QTM to find the overall more optimal solution. Here, because Terry uses DR there is a bias towards using double turns and thus Terry loses the tie breaker. This creates a problem where different method/approaches to solving FMC have an advantage - here Terry is punished for using DR which isn't fair to them. The same thing happens if we use other turn metrics to determine the tie breaker. If Terry were to instead use Corners First there now is a bais as the method tends to utilies more slice moves. One way around it could be to compare the solutions with multiple turn metrics, say HTM, QTM, and STM. This helps to smooth things out as to remove some bias but the bias will still exist. The other problem you run into is solutions which are identical - there doesn't exist any metric which breaks the tie (this happens more now than ever with DR as solution tend to be more linear).

Phew, that was a lot but we're not quite done. I am now going to compare this to a different event - 3x3. I will say that this isn't one to one so take this with a grain of salt. That being said I do think that this will at least make my position a bit clearer. Here's the final hypothetical =>

There are 2 competitors (1) Vinnie and (2) Winnie (RIP Jerry and Terry - you will be missed)

Vinnie and Winnie tie with an average or 10 seconds.
Vinnie's times are 8, 12, 10, (7), and (15)
Winnie's times are 9, 10, 11, (8), and (12)

On average Vinnie uses 12 seconds to inspect their cube
On average Winnie uses 8 seconds to inspect their cube

As a parallel to the above proposal, the tie breaker is determined by the time to submit. The lower the average inspection time the lower the time to submit. Winnie wins the tie breaker. Again this funimentally changes the event in the case of a tie. We move from Ao5 with standard 15 second inspection time to Ao5 with 0 seconds of inspection time. These aren't the same event. Again the parameters of the event become fundamentally different. All of a suden something that does not at all counted against you is counted against you, in the case of a tie breaker.

So what is the solution? Well, the more equitable solution would be to extend the parameters of the even which is the best average time. In this case we take the Ao5 without the removal of best and worst times. Vinnie has an average time of 10 seconds and Winnie has a average time of 10.4 seconds. Vinnie loses the tie breaker. Here Winnie is rewarded with being more consistent - which is the entire idea behind the event - best overall average.
Now, there is another possible sotion to a tie breaker - best single. In this case Vinnie would win the tie breaker with a fastest time of 8 seconds. Although is works we again break the rules of the event. An Ao5 shouldn't be determined by singles - it should be determined by average.

Okay, we're at the final stretch. Up until this point, when comairing FMC tie breakers, we have compared ties with SINGLE solutions. But wait, what about averages? Good point. Here out final hypothetical =>

There are 2 competitors (1) Jerry and (2) Terry (welcome back kings/queens)

Both Jerry and Terry have a average of 25
Jerry's solutions are 25, 20, and 30
Terry's solutions are 35, 18, and 22

Just as before we have our 2 possible solutions (1) Average solution length with average solution time and (2) Average solution length in multiple turn metrics. As before I believe sotution (2) is better although it will still run into the same problems. However, if we look back on our 3x3 section we have a secret 3rd solution being that the competitor with the lowest single would win the tiebreaker. Here, Terry would win the tiebreaker. This however still runs into the same problem - the metric we are looking for is the average solution. In this case Jerry is punished even though he is more consistent. Given this, our last and final solution, would be to compare the distance between the best and worst solution. In this case Jerry would win the tie breaker with a difference of 10 moves between their best and worst solutions where Terry would lose with a difference of 17 moves between their best and worst solutions. Here Jerry is rewarded for being more consistent which lines up more with comaring averages. Not perfect but definitely better than the other solutions, imo.

I think that's all of my thoughts on the matter. If there's anything that I can clarify please lmk, as I understand this is a lot to take in all at once and is relatively complicated.
 
Back
Top