I experiment with solving chess puzzles in another site. For every chess puzzle there is an expected time to solve a puzzle. If you solve it under the time limit you get full Elo exchange. But if you solve it after the time limit expires, you can only get a portion of the calculated exchange depending on how much it has taken. You lose points 100%, if you can't solve it no matter how much time it has taken you.
Maybe we can do similar here? For example the saved games must not be taken into account even if you win it. But you lose points anyway if you can't win it?
Or can we create a time limit for the games similarly? I know it requires some storage. But maybe we can have an average time limit for a variant and do adjustments by some calculation depending on the game Elo?
Hey, Hop, you are one of the fastest players here, and I am one of the slowest.
I would be very much against any kind of time constraint!
You'll have to learn to sort the deck faster.
I mentioned something very similar (same idea without the chess reference) about a month ago and it didn't seem to get any traction.
What's a matter Hop, is floccinaucinihilipilification getting too close to you in ELO points. I see that they are a slower player.
There's also the flakiness that occurs. Many a time I've used save in order not to keep retracing my steps.
"What's a matter Hop, is floccinaucinihilipilification getting too close to you in ELO points."
Strange. It didn't even cross my mind.
Come on guys, what's the big deal with not getting full Elo exchange at times? It doesn't even affect your streaks.
I'll tell you the same thing I did months ago, Hop. And that is that most of us aren't hung up on speed. Speed is for you guys who enjoy the tournament milieu. Back well before you got here, before the database got so corrupted.......I was considered a "solver" - and some of us just lean that way. In my case, it's partly from being a science guy - where accuracy is far more important than speed - partly from my history of math tests where wrong 'answers' are penalized (like, imo, they should be here), and partly because to me this is a leisurely, relaxing diversion. Yes, the 'intellect' of someone who can solve game #1234 in 2 minutes could well be greater than someone who solves it in 5 minutes - all things being equal. But they're not; not necessarily at all.
I think you're just barking up the wrong tree here. You speedsters have the tournaments - 24 hrs. a day - to beat games into submission via the brute force of a dozen attempts as fast as you can click. To me that's quite inelegant. Yes, there is a certain skill there, but it obfuscates a calculated, thoughtful approach. EVEN IF, like myself, "streaking" has less importance due to the cheaters.
Methinks you're only looking at things thru your own lens and inclinations. And while that's maybe natural..........there are other valid ways to approach these games. Many good players here have noted - repeatedly - the difficulty of trying to succeed in both "streaking" and tournaments. They're two different worlds.
*Separately from that*, sorta, is the issue of stopping the clock for self-aggrandizement purposes, as joey addresses in the other thread ("anonymous usage"). That's a sticky wicket, but perhaps there's room there for an altered approach. Clearly some players abuse that 'opportunity'.
"Methinks you're only looking at things thru your own lens and inclinations."
If a big chess site employs this idea, that may be something to be considered. Hey, it's just a thought. I don't even know if Denny find it interesting.
But I was expecting that you would support at least the idea not giving Elo points to the saved games. Anyway.
I think that last-mentioned idea has merit. BUT..........as a couple others have noted........there are many *legitimate* reasons for putting a game on hold. Heck, *I* even do it sometimes - but never to try and solve it offline and then come back and enter the solution (like some apparently do). When I do it, it's a phone call, bathroom break, gotta get to a Dr. appt., etc. PLUS...........I often play in the middle of the night, when I should be sleeping. So let's say I'm playing 11x0s at 4 AM. Most of those can be won reasonably quickly - IF they're winnable, and even when I'm way past the point of extreme fatigue. But I draw one of those that can't be quickly discerned whether it can be won or not, and can barely keep my eyes open 🥱. It's not terribly rare for me to pause such a game until I feel I can give it fair attention the next day. But again..........never to somehow use external aids or otherwise gain some kind of advantage to make it look like a faster solve than it was.
YET...................if some sort of system along the lines of what joey laid out could be implemented, then I/we could adapt to that (and I might get a tad more sleep sometimes 😴). But since I'm not an elo chaser, that specific aspect doesn't carry much weight - altho can certainly appreciate that consideration.
"But since I'm not an elo chaser..."
Then why object to not getting Elo points because of saving games? Not a big deal for you.
Do people on the chess site compete for streaks of puzzles won also? If so it seems a strange contradiction to have one set of rules for what counts as a win and another set for what affects Elo rating. Here the Elo ratings are meant to predict how well a player will do at solving streak games. I'm not really interested in a rating system that measures something different from the streaks the site is built around. That's interesting though.
Competition chess has always had a chess timer so time is an understood component of that game.
It you really want to include a time component in Elo you can get a pretty darn close approximation from existing data. We keep average time for each variant and number of plays. Calculate an average for all variants weighted by number of plays and then divide Elo by that value and you'll have a pretty good time-weighted Elo.
MrFixit
Hop: "Then why object to not getting Elo points because of saving games? Not a big deal for you.
Just off the cuff...........I doubt if I would have an objection to that. Seems like that's a pretty trivial issue, but you likely pay more attention to it than I do. And I assume that one could/would still *lose* elo points if the game ended up being lost, right? Otherwise, just an easy way to avoid a loss.
"And I assume that one could/would still *lose* elo points if the game ended up being lost, right?"
Earlier (1st post), I said: "Maybe we can do similar here? For example the saved games must not be taken into account even if you win it. But you lose points anyway if you can't win it?"
Got it. Conceptually, I don't think I'd have an objection to that. But is that currently really an issue? Maybe it is; I wouldn't know.
"But is that currently really an issue?"
No, it isn't. Brainstorming.
Being that ELO has been applied to many other games and competitions, I'd be curious if any of the applications don't equalize ELO for time. You're not getting an accurate head to head comparison to other players if one person uses twice as much time as another. This would seem similar to a chess player beating a chess computer in 1/3 the time of another player, but each would receive an equal ELO boost.
Right, but again you have to define what it is you're rating. Here we're rating the likelihood of players at freecell.net winning the next game they play in streak mode. Who says playing fast is better? Maybe slow is better.
If the end result is a win, my suggestion assumes faster is better.
If you were to put a timer, then it should be per move like in chess. 30 seconds to make a move or devaluation sets in. Multiple violations increase the devaluation. Back and forth movements don't count.
Late to this one.
I guess speed is an element in the playing of these games, but I have the impression that Hop's suggestions would move ELO somewhat closer to an IQ rating(?).
If I'm not mistaken, one suggestion relating to saved games would result in those who pause/save games for 'valid' reasons to be treated in the same way as those who are alleged/assumed to be 'cheating'. Perhaps Mr Fixit could insist upon everyone engaged in streak play to post regularly in an 'assessment' thread, and devise a personality rating to differentiate between those two hypothetical groups.
It doesn't really matter to me how ELO is calculated, or even why, though I do take an interest in my own rating - lets me know how rapidly my mind is degenerating. However, I wonder if a prerequisite would be for all ratings to be reset to their original default values (1500?), given that 'new rules' would apply?
it could be interesting to know the average time to solve a game anyway independently from Elo
why not giving an extra Elo portion if you solve faster than the average. It's about the same, but more positive, slower players don't loose compare to the current calculation
I think this a nice, simple suggestion. I don't think Hop is looking for a short time limit for rewarding players who tend to play faster. Similar to games solved in Hot Streak not impacting Elo ratings. I opened up Hot Streak today winning a 27/138 8x3 deal and thought wouldn't it have been nice to have the Elo points for that. Seems like not giving Elo points for previously saved games is in the same vein.
Two thumbs up here.
There are a myriad of reasons to save your game:
The dog threw up on your foot.
Your lower GI tract is acting up.
Your boss just walked in your office.
The phone made a sound.
Amazon arrived.
The traffic light changed.
Mother spilled her gin on your laptop.
Uncle Vito's gun went off while he was cleaning it.
ICE is at the door.
I would suggest a limit on the save time of around, say five minutes, when you can come back, and everything is as it was before you hit save.
Of, course, if it is ICE, the least of your problems is your saved game. They will provide you with a free cell.
I proclaim firenze's last sentence just above, concerning ICE, as "quite clever". Considering that his neighbor, MrFixit, has been working for the gub'ment..............I can see why he'd worry about that.
"This is getting confusing. Are the comments suggesting no ELO points for saved games or no time adjustment for saved games? If you get no ELO points at all what about lost saved games. You lose points for a lost save game but gain nothing for a won saved game?"
Practically they are the same. Let's say the time limit is 200 seconds for a variant. You have won a game and the calculated Elo exchange is 1.000 points.
If you solved it less than 201 seconds you get 100% of that point.
But if you solved it say in 400 seconds, you get 50% (0.500) of it. 200/400 = 0.5 (50%)
When you saved a game, probably the times that has passed at solving time is very long, say 10 times, that makes 0.1 (or 10%) and it is very close the zero. Codingwise, it's very easy if you assume it's zero.
Hop:- You have an alternate ratings page on your site already, right? Would it take a lot of coding to just introduce a time element into it, or is it a lot more complicated than that? That way, the 'official' ratings wouldn't be affected but for those interested there'd be yours to look at too.
Uberman,
Yes I can do it when I have time. But it's not possible for me to know if a game is saved. I have only elapsed time. You know, the clock stops when a game is saved.
Ah OK, gotcha. So yeah, it's probably not worth doing if you don't have the most important criteria available.
On the other hand, not giving any Elo points to saved games would be very a easy code change.
so to protect your elo... save, restore, when you know you can't win timewise or gamewise
"so to protect your elo... save, restore, when you know you can't win timewise or gamewise"
No, earlier I said "You lose points 100%, if you can't solve it no matter how much time it has taken you."
"then a save is a loss?"
No. Your streak is not affected. It's just that you do not get any Elo points if you save it (and win it later).
I added "Time Adjusted Elo" to the individual scores page. It's an Elo weighted by average playing time.
First we calculate the average time of all the variants summing the individual times and plays to compute totals and then take an average:
allAvgTime = allTotalTime / allTotalPlays
Then we normalize to 4 minutes as an average game time (with 4 being an arbitrary constant):
allAvgTime = allAvgTime / (4 * 60)
And then we apply that result to scale Elo:
adjustedElo = Elo / allAvgTime
If you want to build a sorted list I believe you can do it by using the User feed to do the same calculations and then generate a list sorted by this time adjusted Elo.
MrFixit
I am hugely in favor of of the time limit: 5 minutes after saving, a win does not count toward a streak or Elo. This addresses a contentious issue or two without punishing the innocent. Even those who save and spend some length of time solving through some other means are not punished -- they are simply not rewarded for gaming the system.
Boy, TT and Ellie drop quite a bit in ranking with the time adjusted elo.
MrFixit, are those avg times and plays Streak only?
if the time adjusted Elo > 5000 does that mean the player is unwinnable?
Hop is over 5000. jimmyp and huffy are both higher than Hop
But, my "Time Adjusted Elo" has gone up by more than 1000! I think I will ignore it...
What I don’t want to see is, any rating influenced by whether someone uses the save button or not. If you had the actual clock time that the game was started, it would be fine to base rating decisions on that: the actual elapsed time. In my opinion if you are seriously trying for ratings maximization, you should pay attention to that and play when you know you’re not gonna be interrupted and play games straight through. Then whatever time you took was the time you took. But using time that can be gamed by the player with use of the save button, not so much.
Wow way to go huffy! I think because the expected play time differs quite widely depending on variant and huffy plays a lot of 10x6 they are scaled up quite a bit by the time adjustment.
OK since everyone cares a lot about whether folks are cheating or not so I'm gonna start turning on folks' camera to verify they'll adhering to the rules and not clicking Save unnecessarily. Or picking their noses.
MrFixit
Thank you! And all the kumquats rejoiced!
I volunteer to monitor the video streams. And can it be interactive?
I know you have been peeking at me, Denny. That's why I play naked.
Maybe time adjusted Elo ratings should be divided by 2 to get them into the range where Elo scores tend to sit anyway.
"Volunteer to monitor video streams"
I'm pretty sure you would end up regretting that offer.