Graduated results in case of abandonment

[souly4]

Hi, due to recent events I am persuaded to suggest an improvement in the results where one player quits and where in total 3 or more players are involved.

As you can imagine, it is very discouraging and annoying to invest much time and effort in a game, maybe even approaching a safe win, only to see it collapse because of an unfair player who is incapable to lose, or because someone's connection is unstable (which doesn't feel much better considering the insecurity).

In my 6 nimmt games since 19.1. alone, there have been six abandoned games. Of these I was in the leading position in four or five. In multiplayer games like 6 nimmt the outcome makes a much more significant difference: in ELO (probably around 40 points for first vs last place) as well as in winning ratio (2.5 points/250 % for one win with the usual 5 players instead of 1 point with 2 players).

The most recent incident is this, the outcome was obvious and 'ydtdmd' left two moves before the end! To me it just looks like he was sore and couldn't stand taking many points in quick succession, and the victim was me:
https://de.boardgamearena.com/#!table?table=49177855

Therefore in my view there really should be a fairer method of calculating the new values after a quit. You could maintain the proven rule that the progression must surpass 50 % in order to be evaluated. If that's not the case, the previous rule should be fine (even better in my opinion: the more progression, the more points you win depending on your place).

At the very least in the case of more than 50 % progression, the final outcome should be taken into consideration for the calculation. Right now the percentages (250 % for five players) are distributed between the 'winners' (250/4 = everyone gets 62.5 %) and they get a very low amount of ELO irrespective of the outcome. I'd like to point out that I was assigned +0 in all these six games combined, as I often was the one with the highest ELO. If we assume I would have finished 1st in three or four of these games, the difference in ELO and wins would be enormous.

A fellow 6 nimmt player today mentioned that quits are even more common than I assumed (namely with more than five players which I don't use) and there the swing what you win/lose is even higher so it gets even more unfair.

I don't play too often anymore and don't attach as much importance to my ratings as in more active times, but I wish the system to be fair at least. If I likely lose, no problem when the result reflects it. If I likely win, I expect my ELO and wins to rise. If you admins already provide so many detailed statistics about ELO scores and thus encourage some users to pay attention to it, it should be totally consistent. So the suggestion is that your place at the time of the quit and the progression determine how many ELO points as well as how much winning percentage you gain.

For 5 players it could be like this.
1st - 1 point (100 %)
2nd - 0.75
3rd - 0.5
4th - 0.25

That's 2.5 points in total too. Maybe this would be for 50 % progression...The longer it goes on, the more likely the places would persist until the end and the bottom places barely have chances to win, so the distances from the top should increase. For example:

90 % progression
1st - 1.5
2nd - 0.7
3rd - 0.2
4th - 0.1

Or if that's too complicated just stick with the first chart as a model for any player count: The person in the middle should get the average percentage and from there it spreads in equal intervals.

I would appreciate when some people who support me with this come forward here (be it because it happened to you often too or you just think the status quo is unfair)

[Lena the Great]

I agree with all above!

[Caffe Latte]

Actually also one of reasons I stopped playing 6 nimmt. The effort vs risk that somebody quits became unbearable.

[seaofgreen]

100% agree

[souly4]

A brief addition about the ELO:

While I comprehend that - the quitter excluded - the player in last place shouldn't be punished because he can still improve in the course of the game, putting him at the same level as the leader is plain unreasonable and absurd. Let's take an example where the 4th place has 250 ELO and the 1st 550. From my experience the former would lose around 10 points, while the winner would gain 10 points. With the current system this is reversed: Due to the quit the leader would now trail by 15 to 20 points (+0 for him, +15/20 for the other). This way he is punished twice just because he has a comparatively high rating.

My proposition for the new ELO calculation would be this: We don't take into account the quitter. From the remaining players the last spot gets +0 and the all the quitter's points are shared between the rest - in relation to place and rating (as ususal) plus this result multiplied by progression. This preserves the proven computing method while nobody can really complain about being disadvantaged...

[Jest Phulin]

Just as a crazy idea, would there be merit in making the abandoning penalty more severe the more players that are affected? IE, use the current system if playing a 2-player game, but count it as 2 abandons if playing in a 3-player, or 4 abandons if playing in a 5-player game?

[Liallan]

I am totally happy to jump in on this because I've hated how this works since they first implemented it (I even left for 8 months because of it), and hate it even more watching how it actually plays out in real games. So I'll post about this every time it comes up.

I have to say first that I don't know 6 nimmit so I can't follow everything you said, but it's a problem in all games so that's a bit irrelevant. I also only (mostly) play turn-based and that's a bit different situation.

The one thing I have to disagree with is calculating anything at the time of the quit, because you really don't know how things might come out. Maybe that's more obvious in that particular game, but it certainly is not in many games, and some can change drastically. So nothing should be calculated until the final results are in so you know how things actually came out. You gave some different options, but this option does not work at all in my opinion.

I also don't understand the graduated thing, or maybe that's directly related to 6 nimmit. To me it should be as simple as gaining/losing whatever ELO you would have as though the game was completed as normal, except only counting the % of progress where the game was. (i.e. you would have gained 30 ELO, but the quit was at 67% so you only get 20 - but past that adjustment I don't think anything special has to be done. Well, and not comparing with the person who quit.) And really, if it was based on the end-game scores, I'm not sure I see a reason you can't just do 100% of the ELO.

Now, as to what is different in turn-based: In real-time, I suspect most people who quit are doing it because they're losing, and probably do it closer to the end of the game. In turn-based, the vast, vast majority of the time, people leave very early in the game, either taking only one or two turns, or none at all - like they never even show up. That's an entirely different situation.
With that difference, being turn-based, it is often the case that other people start quitting because there's no penalty for that. I'm someone who will tend to quit because that is not the game I agreed to. The best solution is a group abandon, but there are people who simply refuse to do that, even in this situation where everyone else is innocent. So even if other people quit, if there's at least 2 people to go on with the game, everyone still ties, even those who dropped. So if people won't do a group abandon, I'm still stuck with that tie even if I quit too. That makes no sense at all. Granted, because this might be only 5% progress when the quit happened, your ELO is not going to be affected much, but I still don't like the game coming out as a tie, and being included in that, if I never even played the game. Those who decide to leave after a quit should never be included in the final of that game, and the ELO should be determined based on who stayed in it and their actual ending scores.

(I should add that I also find it annoying that this tied game is counted as a win. I'm quite sure many people have no issue with that, but I'm too honest, and I like to win fair and square, not because I "tied" a game in which someone quit. It throws off statistics and that really bothers me too.)

This is one reason I always take a game last, with few exceptions, because I want to know who is on that game and the chances of them quitting. Abandoned games happen far, far less since I started doing that. Also, when I make invitations, I always require 85% reputation, which isn't recommended cause then new players can't take it - well, yeah, that's kind of the point. Everyone is new at some point and I hate to screw them over, but I detest people dropping and this tie thing so much that I'm going to avoid that at all costs. (Except 2 player, and that works differently.) Before I started doing this, quite literally half of my games were abandoned - and I kept track of it so I know. It still happens but not often.

[souly4]

Thanks for your (late) comment

I don't feel like responding to everything and I'm not knowledgable at all in turn-based anyway, so I will briefly dwell on the most relevant points.

I have to say first that I don't know 6 nimmit so I can't follow everything you said, but it's a problem in all games so that's a bit irrelevant. I also only (mostly) play turn-based and that's a bit different situation.

The one thing I have to disagree with is calculating anything at the time of the quit, because you really don't know how things might come out. Maybe that's more obvious in that particular game, but it certainly is not in many games, and some can change drastically. So nothing should be calculated until the final results are in so you know how things actually came out. You gave some different options, but this option does not work at all in my opinion.

What's the alternative to calculating at the time of the quit? People often tend to criticise novelties just because they don't seem perfect, even if they improve the present situation a lot. In psychology there are actual phenomena with names for it, status quo bias or default effect, with similar implications. The ability to improve something continuously shouldnt keep you off taking the first step.

As it's now, in my opinion it's simply insufficient. The duration of the game and the scores (and accordingly the associated positions) are the two large factors that determine who would win most likely, so what else do you intend to include? At present the last spots get rewarded for no accomplishment, robbing the points from the (potentional) winners. Surely there is no guarantee that anyone being ahead would win - the match was interrupted unnaturally so we an need unconventional computation as well. My suggested methods of calculation were my best idea how to solve it, how to appropriately distribute the wins and ELO among everyone. If the fourth place would manage the improbable and rise to first until the end in one of five cases, it's unfair when he gets a lot more than 1/5 of a normal win, because if we repeat five games, only with this ratio the result will be the same (at average). Conversely it's unfair when the 1st place gets about nothing in comparison.

I'm probably negatively affected about the current system by two aspects:

- in 6 nimmt which I mainly referred to the exact score is always visible and there is no sensible strategy to omit something at the start to get a higher score later. Unlike in some worker placement games perhaps. That's why the score at a certain time point is indeed indicative about the outcome; more points -> better chances.
- my streak of being in the first position five times in the last six abandoned games doesn't feel that inspiring. I was denied perhaps over 50 ELO because of it. Think about it, it's totally out of your control (luck) if the abandons happen to be in matches were you are ahead or where you are trailing badly. The difference between both is grave even with a small sample.

I also don't understand the graduated thing, or maybe that's directly related to 6 nimmit. To me it should be as simple as gaining/losing whatever ELO you would have as though the game was completed as normal, except only counting the % of progress where the game was. (i.e. you would have gained 30 ELO, but the quit was at 67% so you only get 20 - but past that adjustment I don't think anything special has to be done. Well, and not comparing with the person who quit.) And really, if it was based on the end-game scores, I'm not sure I see a reason you can't just do 100% of the ELO.

'Graduated' was only referring to the winning percentages like I elaborated. With the ELO we are in the same boat. It is still somewhat graduated because while the 1st place might get less than 2nd or 3rd place (due to large difference in their level), he would receive progression percentage * normal points for 1st and that' very contrary to when everyone counts as equal outcome (all 1st place or 2.5th place with 5 players or whatever the evaluation is).
Your last sentence is a bit mysterious to me in regards to your remaining view. Like this statement earlier: "So nothing should be calculated until the final results are in so you know how things actually came out."
My whole idea was to install a compromise between these two extremes: not rewarding the hard efforts in abandoned games at all (very disadvantageous for the first places) versus allocating all the points/the win only to the leader while people behind him still have a chance to eventually win too (disadvantageous for the last places).

One last remark mainly to the admins: I've recently seen someone being involved in three abandoned 6 nimmt games in a row, and that was only with five players if I'm not mistaken. That only should be enough motive of thinking to overhaul this rule, as for example the differences in ELO would influence all the future calculations with these players too.

[souly4]

Still no reply by the admins. Conveys not too much concern about the player's requests.

I didn't have many abandonments recently, but that doesn't change if there is one, it is as annoying as always. You can guess, I was in first place again. That whole series robbed me so many points and wins, without being influenced by any skill. If we take a sample of 2 games only (for me it was the last 7 or 8) with one abandoned game: One game you get 4th of 5, one game 1st. If the quit is in the first game, it counts as zero losses instead of one, and you keep your win in second game. If the quit is in second game, you lose one and are denied that one win. Bottom line, the results read as +1 compared to -1, which is like two additional wins. Two wins decided by nothing but randomness, namely when someone has the audacity to leave because he loses (like in my newest case) or connection problems.

It heavily decreases the appeal of multiplayer games, and not only in my estimation, but from many people, from which only a fraction answered in this thread.

https://boardgamearena.com/table?table=55694393

[Silene]

Your suggestion would require an individual solution for many games. As you suggest it, it might work for 6nimmt, but many other games have different ways of making points and you cannot say who'd be winning in the end by looking on the score during the game (because i.e. stone age: some players make more points during the game, some go heavily on the cards that give a good end-of-game-scoring).

I think best is to not hand out any points in case of an unfinished game, because it is almost impossible to make a specific solution for so many games. Just don't play again with the players who don't finish.

Also you jump to conclusion very quickly regarding what's admin's concern... Not very kind imo.