What do you think of the current skill calculation system?
SupaDupaNoodle
Join Date: 2003-01-12 Member: 12232Members
Join Date: 2003-01-12 Member: 12232Members
Comments
I think it should take into consideration that 1 match=1 alien + 1 marine round.
But I think Win Lose is pretty much the best and I feel fairly awarded for win streaks, but unfairly penalized for random losses quite often
Only Siths deal in absolutes!!
E.g. "System is broken and produces inaccurate skill ratings which damage gameplay when used for team construction". The system isn't broken. The system is more or less fine (albeit, not perfect). But it DOES produce inaccurate skill ratings. If you put the best, least broken skill system in this game, it would still produce inaccurate skill ratings because of the wealth (or lack thereof) of data it is being fed. There is only so much you can do with so little data. You need to give these things time to converge towards an accurate result. And the time we have had, whether you can argue it is enough or not, there certainly haven't been enough people playing and in enough of a range of skill levels. It would have had far better results if it had been operational since release when there were loads of people playing.
So whether or not the system produces accurate skill ratings does not necessarily have anything to do with whether the system is broken or not. It could just be the data. And likely is.
This is not taking into account the problem that even with a magic skill system that evaluates skill accurately, the player that goes commander after teams are assigned has a large impact on the team. A good/bad commander will affect the team accordingly, especially on marines. Also, unless the skill level on the alien team is even, the choice of lifeform of individual players will be significant. If the one exceptional player goes gorge/commander, the team is probably done for on "balanced teams".
___
Based on personal experience, I find people that have below 200 skill or between 1300-2000 skill to be the most inaccurate? I can reasonably expect players from 200-800 skill to not be useful, 800-1200/1300 to not be horrid, above 2000 to be at least fairly good, but players that have 0-200 and 1300-2000 are all over the place in terms of skill. They seem to be anywhere from below average/average/good/great regardless of whether they are closer to 1300 or 2000.
Its like making a poll "do you like eggs and ham because it cures cancer? OR do you kill nuns in their sleep because you have STIs? You cant have both! You cant have neither! You gotta pick one!"
See Moultano's post above. There HAVE been people making valid suggestions, but there have also been a couple of outspoken dudes who, even after hundreds of posts explaining things to them, still dont understand how it works, nor accept the serious issues with their own proposed changes. This poll is rather transparently loaded with the intention of making anyone who disagrees with your stance on the skill system seem unreasonable.
We don't need to over complicate the issue here. I know some people excel at writing epic theses on the merits and demerits of the system, but the point of the poll was to provide a simple binary choice with which to gauge opinion.
I am questioning the system itself because in my experience it is inconsistent and inaccurate given what happens in each game. When talking about it with others on servers, I find people who feel the same.
Well, the numbers look great 0,2 seconds after a vote, then everything changes or it stays and from time to time it shuffles all rookies in one team together with a frustrated pro that doesn't want to carry anymore and everything goes to hell...
I don't think that the vote system can replace a good admin, there is my answer
Here's my answer: The skill system works well in general but has flaws which are being addressed.
None of the religiously-against-the-skill-system people have seemed to understand how it works and why it works that way, nor have they proposed any feasible changes to it. The real problems and solutions have been proposed by people that can see how it works and are in general agreeance that it works well to an extent.
"System is broken and produces inaccurate skill ratings which damage gameplay when used for team construction"
[System is broken] [inaccurate skill ratings] [damage gameplay when used for team construction] - all the time? some of the time? Wording implies the former, which is why this is loaded. Both answers suffer from loading in this way.
A better set of questions could take the form of
- Skill system is perfect and works every time (essentially this is what you imply with the first option in your post)
- The skill system works well in general but has flaws
- The skill system is broken, but sometimes produces good results
- The skill system is broken and must be replaced entirely or removed
OR
- The skill system statistically produces better games than random vote
- The skill system is no better or worse than a random vote
- The skill system statistically produces worse games than random vote
Reducing complicated issues down to 2 opposite sides is to create conflict and tribalistic thinking.
1. I usually vote for force even teams.
2. I usually vote against force even teams.
Even better, as that supposes nothing. That data be so raw, so raw
Ironically he made it far more complicated than it was to begin with.
Here are some examples of true dichotomies.
- The current skill system is constructive to team balancing.
- The current skill system is not constructive to team balancing.
- The current skill system is destructive to team balancing.
- The current skill system is not destructive to team balancing.
These are true dichotomies, you can not be "in the middle" of any of these positions.
Here's a false dichotomy however:
- The current skill system is constructive to team balancing.
- The current skill system is destructive to team balancing.
It's a false dichotomy because you can also be neutral on the subject.
I'm gonna nip this in the butt right away, because a common misconception is that the position "... is not constructive" == "... is destructive". This is not true, you can hold the position that something is not constructive while not thinking it is destructive. In other words; the "not constructive" position contains every position other than "constructive", including neutral and destructive.
Instead they keep join on a team that have higher ranking numbers =='
Ok, let me vote statement-by-statement:
"System calculates skill accurately" - No - I am not convinced by the algorithm used, based on the article, and assuming some implementation details (e.g. 0 SP newbies, 1000 SP after reset). It is proven, that the calculated Skill is clearly wrong for not-so-corner cases. And I am skeptical of its accuracy in the bulk of cases. Yet, I firmly believe it is somewhat more accurate than random shuffle, but I could be wrong.
"can be used as a reflection of player's ability for making even teams" - Yes, it should - You risk unbalanced teams with FET/random. Without it you have pretty much guarantee of unbalanced teams. In reflection of what alternatives there are, it should be used every time. Comparison of the system vs random would be interesting (yet perhaps hard to make).
"System is broken" - Yes - The problem lies more in the user-interface part. You can opt-out as individual and as a team. I would rather have unbalanced teams, than bicker for half hour if FET should be voted, than when it finaly passes people switch and F4 leading to the original unbalanced team (remember the FET only switches as minimally as posible to not upset players. if those players switch back, it's as nothing has happened)
"produces inaccurate skill ratings which damage gameplay when used for team construction" - No - Yes, it produces inaccurate skill rating(viz. above "System calculates skill accurately"). No, that's not what damages gameplay (viz. two answers above).
The question isn't loaded, the answers are (if that is even a thing). I thought I explained it pretty well. The system isn't broken. But it also produces inaccurate ratings. The reason it produces inaccurate ratings has nothing to do with the system itself, but the wealth of data - even if the algorithm was as perfect as realistically possible it would still spit out inaccurate ratings due to the data it has to work with.
Either the poll is about the quality of system itself, or the results it is able to spit out. Because one being good does clearly not necessarily mean the other is good, yet the poll assumes that, hence it is loaded. Two equal and opposite answers would be "The system is broken" and "The system is not broken". Or "Skill values are accurate" and "Skill values are inaccurate".
The way you've worded it is a bit like having a poll on the president where you have to choose between "Obama is a good president, and the country is going well" and "Obama is a bad president, and the country is going poorly". There are some people who might think that Obama is a good president, but the country is going poorly, or likewise that Obama is a bad president, and the country is going well. The quality of the presidency may relate to how the country is doing, but the quality of the presidency does not necessarily entirely reflect how well the country is doing. I.e. A good presidency does not necessarily equal a thriving country and a bad presidency does not necessarily equal a country doing poorly. Just as a good skill system doesn't necessarily equal good results.
It is not a simple binary choice. It could be if you worded the poll differently. As it stands, I can't vote for either answer, because you put too much detail in there such that I can agree with part of the answer, but not the other part. It shouldn't be like that if it were a simple binary choice.
Ah... I think that's the term I was looking for. Not "loaded answer" - that's not even a thing.
In some respects the algorithm is hopelesly frequentist, and dismiss data that could at least accelerate convergence of new players (that is even the evil KDR). On the other hand it does some dangerous assumptions about player time spend in the round - which is good place to cheat the system, if somebody is inclined that way.
Same here... To me it seems like the system works somewhat decently maybe 60% of the time, and is a complete train wreck fail the other 40% of the time, and I think the main problem is lack of different scores for each side.
I play Alien like 90% of the time, so my score mainly reflects my Alien skill level, not my Marine skill level. If there were separate scores for each team my Marine score would easily be half of my Alien score.
A middle option like "the system works some of the time and just needs some improvements" would be my choice.
@moultano you have already mentioned this, but having a separate skill rating for marines and aliens would help alot in balancing teams. Though now you have a matrix to deal with wrt getting the two teams with about even skill numbers, and may have to weight the scores of aliens and marines (ie. a skill 1000 player on aliens might equate to a skill 800 player on marines in early game (skulks vs marines), but if the skill 1000 players is an excellent fade, then it turns that on its head. I think people really over estimate how good ELO / FET is for a complex game like this that rely on team work. We are never going to get perfect, the system is design to TRY and get a sort of balanced game so it isn't a one side stomp.
Games with random teams seem better (= more enjoyable) then not random teams.
This might be due to a motivational effect ('teams be balanced so if i lose i am a pussy') or simply because the system actually work.
Well you constructed the poll in a way that we can not vote. But yea, congrats on your "victory". *roll eyes*
The real bonus is, that we can fill another duplicate thread with 14 pages of titillating discussion!! :P
I feel guilty that I voted for the positive worded part of the poll, even though I did not like either option.
This isn't a great sample size to begin with, but if we look at the people who said they mostly trust the skill system but didn't vote I think the poll would look different right now.