ScatterJoin Date: 2012-09-02Member: 157341Members, Squad Five Blue
A lot of good replies here. I think in terms of what to do with tunnel rushes I think it would be good to take a step back from providing specific solutions and look at a bit more generally. This is because some people don't seem to agree it is an issue in the first place, and I want them to change their mind.
In my experience, people who write "Hive is broken/useless/worse than not using it" have no clue what they are talking about.
Then you must have a limited experience because everyone here who has stated that they have an issue with hive has stated why and provided and argument to support it. Give it a try.
In my experience, people who write "Hive is broken/useless/worse than not using it" have no clue what they are talking about.
Hive shuffle is useless and/or worse than random shuffle on arcade servers. A lot of the more experienced players in that community have very low hive scores.
For very obvious reasons. We aren't talking about arcade servers.
I'm at 2.1k ~ 2.6k Hivescore most of the time and Im typically in the top 1-3 players on the scoreboard unless I am playing support. But I still score high when doing that. I lose all the time however. This is typically due to my team not working well together or playing half assed and not holding res or getting lifeforms. The game is fucked when 6 people have 70 res and we have no lifeforms except gorges....
I feel like hive skill is a fairy accurate representation of players and gives some foresight in how the match will go. I rarely ever have a great, well played, team coordinated match with sub 1200 players. However I feel shuffle doesnt do very well in most cases. The top 2 players always seem to be on the same team after a shuffle and if they are in the 4k zone then the other team is fucked. Not every time but most of the time.
In my experience, people who write "Hive is broken/useless/worse than not using it" have no clue what they are talking about.
Hive shuffle is useless and/or worse than random shuffle on arcade servers. A lot of the more experienced players in that community have very low hive scores.
For very obvious reasons. We aren't talking about arcade servers.
I don't think it's obvious. A lot of people who join arcade servers are confused why hive isn't shown, or why it isn't used for balancing. I've had to explain a lot that we're blacklisted from updating hive scores, so we can't use it for balancing teams. It would be nice if there was something like an "arcade hive score" we could use but no such thing exists.
In my experience, people who write "Hive is broken/useless/worse than not using it" have no clue what they are talking about.
Hive shuffle is useless and/or worse than random shuffle on arcade servers. A lot of the more experienced players in that community have very low hive scores.
For very obvious reasons. We aren't talking about arcade servers.
I don't think it's obvious. A lot of people who join arcade servers are confused why hive isn't shown, or why it isn't used for balancing. I've had to explain a lot that we're blacklisted from updating hive scores, so we can't use it for balancing teams. It would be nice if there was something like an "arcade hive score" we could use but no such thing exists.
IMHO, ns2large is close enough to regular ns2 that it should be included in hive. Filtering arcade servers makes sense for things like The Infested, not things that still use roughly the same skill set as vanilla.
For very obvious reasons. We aren't talking about arcade servers.
That must have been posted while I was writing my post. Mah bad.
I don't think it's obvious. A lot of people who join arcade servers are confused why hive isn't shown, or why it isn't used for balancing. I've had to explain a lot that we're blacklisted from updating hive scores, so we can't use it for balancing teams. It would be nice if there was something like an "arcade hive score" we could use but no such thing exists.
What is the difference between arcade more to other modes? I didnt realize there were different types of servers and Ive been playing for years lol
I feel like a lot of this could be solved with queues into games like CSGO has and player rankings.
For very obvious reasons. We aren't talking about arcade servers.
That must have been posted while I was writing my post. Mah bad.
I don't think it's obvious. A lot of people who join arcade servers are confused why hive isn't shown, or why it isn't used for balancing. I've had to explain a lot that we're blacklisted from updating hive scores, so we can't use it for balancing teams. It would be nice if there was something like an "arcade hive score" we could use but no such thing exists.
What is the difference between arcade more to other modes? I didnt realize there were different types of servers and Ive been playing for years lol
I feel like a lot of this could be solved with queues into games like CSGO has and player rankings.
Arcade servers are currently defined as any server that either has 25+ players on it or declare that their game mode is not "ns2", such as siege or marine vs marine.
Since you asked...As a lowly sub 2k player I think elo is broken for a few reasons...
1) If we win when I'm in last place on marine with 0-3 kills and 10-20 deaths... and my hive score goes up because we won... that's clearly broken
2) Some players have high skill scores because they are good at tunnel rushes, or at commanding... when said 3k+ player gets counter shuffled to balance out a player who is 3k+ due to insanely high kdr... the game isn't going to feel fun or balanced at all.
3) It leads to some people having scores way way below what they should be. (Like those who go to the losing team all the time to balance)
Seems like going by actual skill in kills, score, etc, would provide more accurate data to more accurately balance so teams have an equal number of players who can kill, players who play the objective, and rookie players with no clue....
I can see how it would feel that way...
1) Got to remember ELO is working for everyone playing so if some one on your team is new and good they may be very under valued... As a whole it's supposed to even out.
2) yeah, balancing on k/d might be nice for the first 5 minutes but would lead to a higher chances of a tech snowball.
3)your mistaking skill with an ability to win.
ELO doesn't balance accuracy, k/d .... It balances your ability to help your team win. Would you rather have a marine who shoots 35%, but only runs a single lane without bothering to react to dynamic game changes? Or a player that shoots 18% that is always in the right spot at the right time?
If you spend some time to think about it, there really isn't a better metric then winning. If your getting slaughtered with k/d but still manage to win every one is stoked (except the guy who went 70-5 but never helped with an essential push).
In my experience, people who write "Hive is broken/useless/worse than not using it" have no clue what they are talking about.
Hive shuffle is useless and/or worse than random shuffle on arcade servers. A lot of the more experienced players in that community have very low hive scores.
For very obvious reasons. We aren't talking about arcade servers.
I don't think it's obvious. A lot of people who join arcade servers are confused why hive isn't shown, or why it isn't used for balancing. I've had to explain a lot that we're blacklisted from updating hive scores, so we can't use it for balancing teams. It would be nice if there was something like an "arcade hive score" we could use but no such thing exists.
I was trying to say it was obvious for most people in this discussion. It does not surprise me at all that a lot of people don't understand why arcade doesn't use or work well with hive. That is because most people don't understand what it is. The rest of this comment is not in response to you Nin, but more broadly to the topic at hand. Your post is a nice jumping off point for what I have to say.
A lot of people really do misunderstand what the hive system is and what it does. First of all the name "hive skill" is wrong. Hive does not measure skill. There is no measurement for skill. Skill is vague concept that we can't define all the variables for. What we can measure is how well people meet the objective. The only objective in NS2 is winning. Hive measures a players ability to win in the typical environment they play in. This may or may not accurately reflect a players skill, but it doesn't need to. Shuffle does not balance players skill levels, it balances out each teams expected chance of winning.
For example, here is a function that Moultano shared once that shows how likely a team is to win based on the skill differences. For those who don't want to click the link, having about 100 higher average "hive skill" means that your team has about a 70% chance of winning, if I am reading this graph correctly. Shuffle tries to balance each team to have an equal chance of winning. Your hive skill goes up or down based on how much of a suprise your win or loss was. If your team is 100 hive skill higher than the other teams, and you win, then nobodies hive skill will change significantly if it changes at all. If your team is 100 hive skill higher than the other team, then that is a suprise and hive skill will adjust your hive skill value accordingly.
And yes, hive "skill" does actually relate to a players skill level. You may argue the degree it relates, but it does relate. If you look at the table I have shared here you will see that player performance increases on average for each skill rank. What is skill rank? That was something I made up to try and break players into different cohorts based on skill. I have other graphs and tables that show that hive skill does relate to in game performance, but I like this one because it is so general.
SO WHY DOES IT FEEL LIKE HIVE SKILL IS USELESS?
It is because:
We have a small playerbase...
...with a huge skill gap.
To show how wide the skill gap is, first look at these histograms.
The average hive skill of all players is shown to be 689 in that first histogram, and 1018 in the second. Most people participating in this discussion probably have a fairly good idea of how far apart the average 1000 hive skill players is from the average 3k hive skill player. Yes, those low skill players make up most of the playerbase. It is true that high skilled players tend to play more often, but that does not change the fact that low skill players make up a very large portion of the community.
NS2 has about 200 average daily players but they don't play every day. The irony is that NS2 does have that many players trying NS2. If even half of them stuck around NS2 wouldn't have the concurrent playerbase problem it does anymore.
Do you know what makes them leave? Most rookies stick around until they are kicked out of rookie only servers. Then they leave. They are deterred by the sheer skill difference. If NS2 could somehow group players of similar skill into seperate servers, it could improve gameplay quality and dramatically improve player retention.
Imagine the same circumstances NS2 has in basketball (or any sport). Imagine there are 192 basketball players. That is enough for 8 courts, each with large 12 versus 12 games of basketball. 12 players are professional NBA players. 72 players are high school varsity players. 108 players unathletic children. How NS2 currently works is that each of the 8 courts is going to have a random set of players. Any given court might have 2 NBA players, 8 high school varsity players, and 14 unathletic children. How the hell are you supposed to balance games with players like these? NS2 uses the hive skill system to try and make the teams even. NS2 would make two teams of 1 NBA players, 4 high school varsity players, and 7 young children. Sure, teams are "balanced" but the game won't play well.
This is exactly what NS2 is like. Hive can not make games with 2 NBA players, 8 high school varsity players, and 14 unathletic children work out much better than it already does. That is not the purpose of hive. Hive is not meant to group similar skilled players into different servers. It only tries to make the best teams out of the players it is given. Hive does pretty well at it's purpose, that is making teams with an equal chance of winning. Hive does not make near skilled teams.
SO WHY DOES IT FEEL LIKE SHUFFLE DOES NOTHING OF VALUE?
In my opinion, a lot of it is because of broad misperception of what hive is and what it does. It also appears to be very common for people to F4 the moment a shuffle happens. Sometimes they want to only play aliens. Sometimes they didn't want to play but shuffle forced them to a team. Sometimes they were just afk and shuffle put them on a team. People often mess up the shuffle. It could be from people leaving, or from someone who usually fades going gorge this round, or having a 12 player team with 6 gorges.
I also think that the RTS snowball makes people think games are more stacked than they were. NS2 has a snowball mechanic where the team winning gets multiplicative advantages that makes it hard for the enemy team to come back. As aliens, if you kill every marine player before they build their naturals then you have a huge advantage that is already hard to overcome.
The top 2 players always seem to be on the same team after a shuffle and if they are in the 4k zone then the other team is fucked. Not every time but most of the time.
Shuffle does not do that.
If you were given these 16 players how would you balance the teams? Try to balance these 16 players.
It was almost better in the past when it was based on KDR, because at least then it was only used as a tool to counter stacking, whereas now it's used in literally every single game.
Do you not remember the spawn killing without killing an IP that made? I remember games where aliens would just kill you as you spawned without killing the IP just to get their skill value higher. Pepperidge farm remembers.
I'm at 2.1k ~ 2.6k Hivescore most of the time and Im typically in the top 1-3 players on the scoreboard unless I am playing support. But I still score high when doing that. I lose all the time however. This is typically due to my team not working well together or playing half assed and not holding res or getting lifeforms. The game is fucked when 6 people have 70 res and we have no lifeforms except gorges....
I feel like hive skill is a fairy accurate representation of players and gives some foresight in how the match will go. I rarely ever have a great, well played, team coordinated match with sub 1200 players. However I feel shuffle doesnt do very well in most cases. The top 2 players always seem to be on the same team after a shuffle and if they are in the 4k zone then the other team is fucked. Not every time but most of the time.
I really must agree on your -not using your ressources - that applies to somewhat both sides.
A lot of people don't understand the value of lifeforms - even if they're bad - and think they go no lifeform or gorge - and they can win.
If aliens don't use their ress they probably never will get to the point of having 70 ress because the "snowballing" prevents this.
Once the marines teched up and aliens still have only skulks and gorges it's so easy to kill the skulks because their is no fade or lerk to draw their fire away.
The same works with marines:
- if aliens have no lifeforms - there is no need to actually buy a lot of shotguns - while when there are tons of fades or ultra-aggressive lerks around... shotguns can push them away / make them more defensive in their playstyle
- Marines who don't use their ress could've in general pushed even more - but if they wait, they are likely to loose more and more mapcontrol.... they wait in the faint chance of getting exos (which they do because aliens don't use their ress yay)...
@skav2 why you're still so low in hivescore has mostly reasons like lack of mapawareness, turtleness like slowly reacting / rotating into fights and stuff... I've seen more than enough 3k hiveskill players with 10-15% accuracy just because they know where they're the most useful :-)
Sorry Nordic but shuffle does lump the highest scored players on the same team all the time. It's very common and all it takes is a couple of low scored players.
To make your sample of 16 players more accurate make 2 more of them under 1k... then lump all 4 players under 1k with the 3k, 2k, and 1800 player and move everyone else to "balance" it, because that's what shuffle does constantly.
And people still farm kills to this day... I watched a server die last night because of it.
Hive measures a players ability to win in the typical environment they play in.
This part breaks pubs for me so hard. According to my observatory profile My score per minute is 11.82 which according to your skill ranks would put me in a 3000-3250 ELO. Yet I am still at about 4k ELO. This is probably due to me playing 6v6 competitively, which is an environment where the skills I have helps me to win more than on pubs. The 2 reasons I can think of why the score per minute is so low for my 4k ELO is either because in 6v6 you have less engagements and therefor don't earn points as much or that my ability to win engagements is much lower compared to other 4k players. (probably both applies to some extent)
Anyway I noticed everytime I join pubs the hive shuffle can't judge me correctly because I jumped into a different environment and the result is that 1 side is getting stomped pretty hard. Unfortunately I can't think of a way to use the stats given to the hive to calculate his winning ability in different environments.
Do you know what makes them leave? Most rookies stick around until they are kicked out of rookie only servers. Then they leave. They are deterred by the sheer skill difference. If NS2 could somehow group players of similar skill into seperate servers, it could improve gameplay quality and dramatically improve player retention.
1000 times this. I know that server operators can set minimum and maximum level of allowed skill in the server. I thought this could be used to somewhat acieve that and I saw server that do this. Might be a good idea to get up some more servers that have max 2000 skill or min 1500 and adjust the play now button to put you on servers that have restrictions like that, which you can play on. I think that might be a start but then again I have no idea if the amount of average players online at the same time would allow for a split like that
EDIT: @skav2 sounds for me like you need to play some NSL gathers bro
Sorry Nordic but shuffle does lump the highest scored players on the same team all the time. It's very common and all it takes is a couple of low scored players.
To make your sample of 16 players more accurate make 2 more of them under 1k... then lump all 4 players under 1k with the 3k, 2k, and 1800 player and move everyone else to "balance" it, because that's what shuffle does constantly.
And people still farm kills to this day... I watched a server die last night because of it.
Because that is the fault of the highskill players amirite? Brb, fetching my victimhoodcloak. The amount of times I get shit for "stacking" is ridiculous, when I am as powerless as everyone else on the server to move after a shuffle. I, or other competent players, don't have magical powers that eligible us to move after a shuffle, or switch to another team when it's full.
People farm kills in a PvP game, where kills is one of the available tools to buy yourself more mapcontrol. Gimme a break. If you see me in the hive killing eggs with a shotgun it's obvious to me, I have a purpose. If I am in there alone, I am biding time til more teammates get there and making sure that skulks can't spawn, playing it safe to secure a win.
EDIT: Or pushing a lane to trick bad skulks to rush me down, proceeding to die over and over again(see mezzanine lane, server hive start). Because, their shit descisions is my fault. Assuming responsibility over your own gameplay is just too mean.
Here's a protip. You want less "stacked" games? Stop winning with gorgetunnels. The only issue here is a ton of bad aliens with inflated elos lacking any sort of skill whatsoever playing skulk. If you have 3,5k elo but skulk like a 1k elo, you're going to have a terrible time when hive assumes you can perform a role you can't. Really isn't more to it, you dug your own hole. Deal with it.
Sorry Nordic but shuffle does lump the highest scored players on the same team all the time. It's very common and all it takes is a couple of low scored players.
To make your sample of 16 players more accurate make 2 more of them under 1k... then lump all 4 players under 1k with the 3k, 2k, and 1800 player and move everyone else to "balance" it, because that's what shuffle does constantly.
And people still farm kills to this day... I watched a server die last night because of it.
Because that is the fault of the highskill players amirite?
If you see me in the hive killing eggs with a shotgun it's obvious to me, I have a purpose. If I am in there alone, I am biding time til more teammates get there and making sure that skulks can't spawn, playing it safe to secure a win.
That's not farming kills.. What I'm referring to is when the entire alien team is dead with 4-5+ marines in their main and nobody is shooting the hive!. They're all just camping eggs.
And let's be real, it isn't uncommon to hear "don't end it, let's keep the game going" from some loser who thinks it's "fun" (who is usually in the 1-2k range)
EDIT: @skav2 sounds for me like you need to play some NSL gathers bro
I probably should since most of my games are spent telling the team how to play. Been wanting to play a serious game where people know what to do.
As for the shuffle system, it would probably be most fair if the top 4 players were placed on opposite teams then everyone else placed randomly to even it out. Or at least top 2 players.
Most of my game I hear this "dont go to |map area| because *x* is there. And X gon give it to ya". We literally only track where the best players go lol. Then if they decide to push him I see this
As @Nordic said before the average player skill is low but to stick to the theme of this thread we just need to get people playing quickly and as fair to both teams as possible.
the ideas so far have been
- Fix shuffle to be more fair
- Gorge tunnel changes due to low cost of use and being super effective
- Queue system like CSGO
^ "Probably wont happen"
- Players Git Good
^ "Pshh yea right."
- Ranked matches based on skill.
^ "This one will be hard due to low player base count"
@MoFo1 If you still see intentional spawn camping for minutes on end, and I don't, then that means it is that much less frequent which is a good thing. Just because it happens sometimes doesn't mean it is acceptable to use a system that incentivises that behavior more. In fact, kdr would be easier to manipulate to get a very high or very low hive skill value in NS2 because of the range of skill we have.
@Mephilles Try not to put too much stock in those averages. Not everyone follows the trend. Last I looked (a long time ago) my spm was higher but my kdr was lower than my skill bracket at the time. An average is an average which means it doens't describe everybody. Also, the upper end of the skill ranks there get questionable stats because the sample size drops off to 1% per skill bracket.
The problem with skill restricted servers is that they are really hard to seed.
The problem with skill restricted servers is that they are really hard to seed.
This is why I keep saying that good, realistic behaving AI bots would be an insanely huge improvement to NS2. Especially if they could be controlled by the players in the server somehow. (basically to activate/deactivate them and increase/decrease the difficulty and/or number of bots)
I remember an old server (Tactical Freedom?) that used to have a great seeding mode where all the players got lumped to Marine against a swarm of skulk bots. It was hands down the easiest server to seed that I have ever seen... and that includes other games not just NS2.
If I were in charge at UWE I'd hire an AI programmer to get some really good bot AI.. I think it's an investment that would have a very high chance of increasing the # of people playing. It just solves so many of the problems caused by the game having a small playerbase.
Tactical freedom got the idea for the marines vs alien bots from tgns. The reason skill restricted servers are hard is because it disables quick play players from joining.
When will ppl stop trying to reason with stubborn, unthinking people who outright ignore facts, common sense, and wouldnt recognize a logical thought in front of them if it bit them on the nose and yelled at them for actual years on end.
Good effort, although clearly wasted as ever it has been.
Player numbers are just not high enough and haven't been for like 4 years to sustain some of the other good ideas in this thread.
Think at a certain point, you must accept the fact that you simply cannot have your cake and eat it too. If devs were able to immediately put a new system in place, it would not also magically beckon thousands of players back to the game to really address the issue of skill gap and games being unfun because of it.
The game is old in terms of modern fps games. The fact that it still gets support is astounding. Personally, i can comfortably say i dont play ns2 anymore. It became a chore and almost never fun for all the reasons echoed in this thread and the countless other threads that have existed in this vein year after year after year. Furthermore, it becomes maddening to see the same ppl (or maybe just a select few) spew the same tired bullshit year after year, deaf entirely to reason.
Whether it's incessant adult children, tryhard edgelords immersed in juvenile meme culture, eternal victims shifting blame, or the righteous community warrriors who wouldnt dare speak a word against the game and are there to defend its honor at any given turn or murmur (e.g. having players kicked for expressing dislike of a feature or map - yes there are ppl this petty, I've seen it happen), there is always a source of grating friction which is not enjoyable. This is the sticky, syrupy, aromatic remenant of the reduced wine that is ns2. There arent a whole lot of level-headed players who are terribly vocal. The ones bitching loudly are usually on the extreme end of any spectrum, as with ANY SUBJECT.
You guys are a special herd, and that is both awesome and terrible.
Sorry Nordic but shuffle does lump the highest scored players on the same team all the time. It's very common and all it takes is a couple of low scored players.
The shuffle algorithm does it the best way it can. If you tried to balance the skills on paper by hand, you wouldn't be able to do it any better.
A possible solution would be to kick/ban the outliers from the server (v.high skill and the v.low skill), which isn't fun to said victims.
@Nordic Great post. I've seen the ELO distribution chart before, and I want to re-express my disbelief that even after rookie-status the massive low-skill outlier is still there. Do you have a ELO distribution chart of rookie-only players? Can we blame this outlier to the concurrent player dropout problem? If the outliers are legit, there is no way we can allow rookies to "graduate" until they break out of that outlier.
A future patch should consider firmly placing all lowest-skilled players in the "rookie-only" category until they rise above it. It may go a long way for improving the average pub experience for non-rookies.
Sorry Nordic but shuffle does lump the highest scored players on the same team all the time. It's very common and all it takes is a couple of low scored players.
The shuffle algorithm does it the best way it can. If you tried to balance the skills on paper by hand, you wouldn't be able to do it any better.
A possible solution would be to kick/ban the outliers from the server (v.high skill and the v.low skill), which isn't fun to said victims.
It could do a better job of balancing out so each team has an equal number of the v.high skill and the v.low skill players... Instead of putting both v.high and v.low skill players all on one team and calling it "balanced" because the numbers look even.
I know I'm wasting my time as I lack the mathematical skills to give an example of what I'm talking about... it's just so frustrating to see people go "oh these teams are balanced" because the averages are close, when in reality one team has all the highest skilled players and is almost guaranteed an easy win.
It makes games either frustrating (on the losing team) or boring (on the winning team) which sucks when all you want is to play a balanced match that goes back and forth.
Sorry Nordic but shuffle does lump the highest scored players on the same team all the time. It's very common and all it takes is a couple of low scored players.
The shuffle algorithm does it the best way it can. If you tried to balance the skills on paper by hand, you wouldn't be able to do it any better.
A possible solution would be to kick/ban the outliers from the server (v.high skill and the v.low skill), which isn't fun to said victims.
It could do a better job of balancing out so each team has an equal number of the v.high skill and the v.low skill players... Instead of putting both v.high and v.low skill players all on one team and calling it "balanced" because the numbers look even.
I understand what you mean in mathematical terms. "The difference in standard deviation between the two teams," in technical terms. If I understand the shuffle algorithm correctly, it minimizes this variable as much as it can, so no, it really can't do a better job. Nordic's example dataset is a good one here, if you look at the "difference in standard deviation of the two teams" of the shuffle algorithm, it is approximately 625 minus 621: equals 4, which is a good number. Nordic showed an example right after that one with a different shuffle method, and the "difference in standard deviation" is much higher, which is bad.
Whenever you experience a truly bad shuffle, it really is the best the algorithm can do, and there is no better way to do it. If you don't believe this assertation at all, you can prove me wrong by screen capping a terrible post-shuffle result and posting it here in the forums. Ideally you'd have to mathematically prove it yourself, but I would be content with just a screencap and an example of a shuffle that would have been better.
If I understand the shuffle algorithm correctly, it minimizes this variable as much as it can, so no, it really can't do a better job.
If this is the case.. then maybe someone can explain why sometimes you can shuffle and end up with an average that is way off, then people reshuffle and the averages are much closer... (and no, nobody joined or left after the first shuffle)
Does repeated shuffles prioritize the average over the deviation or something?
I've been wondering lately what it would be like to just remove it all... No joining teams, and no ability to shuffle. (just hear me out)
Everyone is put in the ready room at the start.. The doors we currently use could be retextured or whatever to say Khammander/Commander respectively... The first person to walk in each door is the comm for that team, once both sides have a comm teams are autoshuffled and the game starts. (And a reset would put everyone back in the RR)
This would theoretically solve
- repeated shuffles before game start (so.... many.... shuffles!!!)
- High skill players not wanting to command because it will handicap their team (and subsequently having your team handicapped because your best player is comm)
- Games that start when only 1/4th (or less) of the server has joined a team
- Less waiting overall (waiting for commanders, waiting for shuffle, etc)
It would also basically remove the need to have a separate commander score right?
The only downside I see is less of an ability to choose what team you want.. Though with how often I join Alien and get forced over to Marine I doubt it would change much in that regard.
If I understand the shuffle algorithm correctly, it minimizes this variable as much as it can, so no, it really can't do a better job.
If this is the case.. then maybe someone can explain why sometimes you can shuffle and end up with an average that is way off, then people reshuffle and the averages are much closer... (and no, nobody joined or left after the first shuffle)
Does repeated shuffles prioritize the average over the deviation or something?
If this is truly the case, it could be giving too much preference to players' team preferences, causing bad shuffles. At the very least a concrete example would help discover whether it's either an error or a preference for low stddev variance. If there is an error, there is no harm in showing a concrete fail-state that can be improved on.
I have actually been taking screenshots of shuffle results for awhile now. I will screenshot if I think about it when shuffle does something weird, or if there is a certain range of skill playing that I want to save the shuffle results of. I have several dozen screenshots of shuffle results. >85% of the shuffle results are weird because people left after the shuffle started. This is not a representative sample at all, so don't take it for that. I have never actually looked at these until now.
Here is a dump of a handful of teams that people didn't leave after shuffle.
[[[Edit: This math was wrong. So very wrong. Guess what I found? I found one team where the two highest skilled players were on the same team. Look at that. Link to shuffle results. When I first looked at it, the team averages had a difference of about 10, but standard deviation had a difference of nearly 500. My first thought is that the server operator might have set it up so that it had a very loose tolerance for standard deviation. The REALLY weird part is that when I put those numbers into google sheets, the average alien skill calculated to 2653.6 but shuffle is displaying the average alien skill as 2586.9. Shuffle calculated the average incorrectly, which seems to have allowed the two top players to be put on the same team. These teams actually have a difference of 77 hive skill, which gives the alien team about a 60% chance of winning.
I can not begin to explain why shuffle did its math wrong.]]]
There are also two examples in the image dump of a successful shuffle making teams that are not balanced by average team skill at all. As far as I know, this usually happens when the tolerance for standard deviation is too tight, but I can't really explain that either.
If I understand the shuffle algorithm correctly, it minimizes this variable as much as it can, so no, it really can't do a better job.
If this is the case.. then maybe someone can explain why sometimes you can shuffle and end up with an average that is way off, then people reshuffle and the averages are much closer... (and no, nobody joined or left after the first shuffle)
Does repeated shuffles prioritize the average over the deviation or something?
First of all, do you have any idea how many times I have tried to explain exactly what squishpoke just said to you Mofo for you to just not believe me? Gah! I am glad you are finally understanding that point though.
Yes. Shuffle prioritizes skill average over the standard deviation. Standard deviation is handled by a tolerance. I believe average skill and standard deviation both have tolerances that are server configurable. Shuffle tries to find teams that are within those tolerances. By default average skill has a much tighter tolerance than STD. Essentially, shuffle tries to find teams where the average skills are as close to equal as possible within a the set STD tolerance. @SupaFred sets his tolerance a little looser because he thinks it gives players a better experience because it allows them to stay on their preferred team more often.
Guess what I found? I found one team where the two highest skilled players were on the same team. Look at that. Link to shuffle results. When I first looked at it, the team averages had a difference of about 10, but standard deviation had a difference of nearly 500. My first thought is that the server operator might have set it up so that it had a very loose tolerance for standard deviation. The REALLY weird part is that when I put those numbers into google sheets, the average alien skill calculated to 2653.6 but shuffle is displaying the average alien skill as 2586.9. Shuffle calculated the average incorrectly, which seems to have allowed the two top players to be put on the same team. These teams actually have a difference of 77 hive skill, which gives the alien team about a 60% chance of winning.
I can not begin to explain why shuffle did its math wrong.
The average that is calculated is correct. Sum the values in a calculator and divide by 9, you'll get 2586.9 rounded to 1 decimal place.
The problem here is likely the very large standard deviation tolerance, which would allow the teams to quickly deteriorate in the name of improving how close the averages are. If you want more balanced teams, the standard deviation tolerance should be low (e.g. the default value of 40 or potentially less if you're feeling optimistic).
Even with the configuration setup, it is still possible for shuffle to get the teams wrong from time to time, simply due to the constraints in play. Not only does it try to respect team preferences and various tolerance values, but it also is limited by execution time and resources.
Currently it uses a hill climbing method to try to find an optimal solution, which is quite cheap in execution and often provides good results. However, this approach can often end up stuck at a local minimum instead of reaching the global minimum, and it will not attempt any further iterations to try to prove that it has found the global minimum. This can mean the final solution may not be the absolute best, and in extreme cases, may in fact be quite far off.
If I had the time and motivation I could implement a smarter algorithm that would either restart the process from a different starting point after finding a solution (though this would throw out part of the team preference bias), allow more solutions to be considered by relaxing the rules (tabu search), or change it to a probabilistic approach such as simulated annealing. However, all of these would increase the time needed to compute the teams and may often result in little improvement, especially when people do not trust the underlying values anyway.
The average that is calculated is correct. Sum the values in a calculator and divide by 9, you'll get 2586.9 rounded to 1 decimal place.
I had google sheets do the math for me with the average command. I triple checked everything was correct.. I also did it in a calculator because you mentioned it, and I still get 2653. Do I keep missing the same number or something? I am reading right from the picture.
The average that is calculated is correct. Sum the values in a calculator and divide by 9, you'll get 2586.9 rounded to 1 decimal place.
I had google sheets do the math for me with the average command. I triple checked everything was correct.. I also did it in a calculator because you mentioned it, and I still get 2653. Do I keep missing the same number or something? I am reading right from the picture.
For sanity's sake, here's exactly what I did (and what the code also did):
4351 + 3989 + 3610 + 2579 + 2447 + 2253 + 1775 + 1179 + 1099 = 23282
23282 / 9 = 2586.88888888888888888888888...
Not sure how you're getting something different, unless one of our calculators/computers is lying to us.
Comments
Then you must have a limited experience because everyone here who has stated that they have an issue with hive has stated why and provided and argument to support it. Give it a try.
I have stated this and my hive skill is 3k so I suppose that isn't the case then is it?
For very obvious reasons. We aren't talking about arcade servers.
I feel like hive skill is a fairy accurate representation of players and gives some foresight in how the match will go. I rarely ever have a great, well played, team coordinated match with sub 1200 players. However I feel shuffle doesnt do very well in most cases. The top 2 players always seem to be on the same team after a shuffle and if they are in the 4k zone then the other team is fucked. Not every time but most of the time.
I don't think it's obvious. A lot of people who join arcade servers are confused why hive isn't shown, or why it isn't used for balancing. I've had to explain a lot that we're blacklisted from updating hive scores, so we can't use it for balancing teams. It would be nice if there was something like an "arcade hive score" we could use but no such thing exists.
IMHO, ns2large is close enough to regular ns2 that it should be included in hive. Filtering arcade servers makes sense for things like The Infested, not things that still use roughly the same skill set as vanilla.
That must have been posted while I was writing my post. Mah bad.
What is the difference between arcade more to other modes? I didnt realize there were different types of servers and Ive been playing for years lol
I feel like a lot of this could be solved with queues into games like CSGO has and player rankings.
Arcade servers are currently defined as any server that either has 25+ players on it or declare that their game mode is not "ns2", such as siege or marine vs marine.
I can see how it would feel that way...
1) Got to remember ELO is working for everyone playing so if some one on your team is new and good they may be very under valued... As a whole it's supposed to even out.
2) yeah, balancing on k/d might be nice for the first 5 minutes but would lead to a higher chances of a tech snowball.
3)your mistaking skill with an ability to win.
ELO doesn't balance accuracy, k/d .... It balances your ability to help your team win. Would you rather have a marine who shoots 35%, but only runs a single lane without bothering to react to dynamic game changes? Or a player that shoots 18% that is always in the right spot at the right time?
If you spend some time to think about it, there really isn't a better metric then winning. If your getting slaughtered with k/d but still manage to win every one is stoked (except the guy who went 70-5 but never helped with an essential push).
I was trying to say it was obvious for most people in this discussion. It does not surprise me at all that a lot of people don't understand why arcade doesn't use or work well with hive. That is because most people don't understand what it is. The rest of this comment is not in response to you Nin, but more broadly to the topic at hand. Your post is a nice jumping off point for what I have to say.
A lot of people really do misunderstand what the hive system is and what it does. First of all the name "hive skill" is wrong. Hive does not measure skill. There is no measurement for skill. Skill is vague concept that we can't define all the variables for. What we can measure is how well people meet the objective. The only objective in NS2 is winning. Hive measures a players ability to win in the typical environment they play in. This may or may not accurately reflect a players skill, but it doesn't need to. Shuffle does not balance players skill levels, it balances out each teams expected chance of winning.
For example, here is a function that Moultano shared once that shows how likely a team is to win based on the skill differences. For those who don't want to click the link, having about 100 higher average "hive skill" means that your team has about a 70% chance of winning, if I am reading this graph correctly. Shuffle tries to balance each team to have an equal chance of winning. Your hive skill goes up or down based on how much of a suprise your win or loss was. If your team is 100 hive skill higher than the other teams, and you win, then nobodies hive skill will change significantly if it changes at all. If your team is 100 hive skill higher than the other team, then that is a suprise and hive skill will adjust your hive skill value accordingly.
And yes, hive "skill" does actually relate to a players skill level. You may argue the degree it relates, but it does relate. If you look at the table I have shared here you will see that player performance increases on average for each skill rank. What is skill rank? That was something I made up to try and break players into different cohorts based on skill. I have other graphs and tables that show that hive skill does relate to in game performance, but I like this one because it is so general.
SO WHY DOES IT FEEL LIKE HIVE SKILL IS USELESS?
It is because:
The average hive skill of all players is shown to be 689 in that first histogram, and 1018 in the second. Most people participating in this discussion probably have a fairly good idea of how far apart the average 1000 hive skill players is from the average 3k hive skill player. Yes, those low skill players make up most of the playerbase. It is true that high skilled players tend to play more often, but that does not change the fact that low skill players make up a very large portion of the community.
NS2 has about 200 average daily players but they don't play every day. The irony is that NS2 does have that many players trying NS2. If even half of them stuck around NS2 wouldn't have the concurrent playerbase problem it does anymore.
Do you know what makes them leave? Most rookies stick around until they are kicked out of rookie only servers. Then they leave. They are deterred by the sheer skill difference. If NS2 could somehow group players of similar skill into seperate servers, it could improve gameplay quality and dramatically improve player retention.
Imagine the same circumstances NS2 has in basketball (or any sport). Imagine there are 192 basketball players. That is enough for 8 courts, each with large 12 versus 12 games of basketball. 12 players are professional NBA players. 72 players are high school varsity players. 108 players unathletic children. How NS2 currently works is that each of the 8 courts is going to have a random set of players. Any given court might have 2 NBA players, 8 high school varsity players, and 14 unathletic children. How the hell are you supposed to balance games with players like these? NS2 uses the hive skill system to try and make the teams even. NS2 would make two teams of 1 NBA players, 4 high school varsity players, and 7 young children. Sure, teams are "balanced" but the game won't play well.
This is exactly what NS2 is like. Hive can not make games with 2 NBA players, 8 high school varsity players, and 14 unathletic children work out much better than it already does. That is not the purpose of hive. Hive is not meant to group similar skilled players into different servers. It only tries to make the best teams out of the players it is given. Hive does pretty well at it's purpose, that is making teams with an equal chance of winning. Hive does not make near skilled teams.
SO WHY DOES IT FEEL LIKE SHUFFLE DOES NOTHING OF VALUE?
In my opinion, a lot of it is because of broad misperception of what hive is and what it does. It also appears to be very common for people to F4 the moment a shuffle happens. Sometimes they want to only play aliens. Sometimes they didn't want to play but shuffle forced them to a team. Sometimes they were just afk and shuffle put them on a team. People often mess up the shuffle. It could be from people leaving, or from someone who usually fades going gorge this round, or having a 12 player team with 6 gorges.
I also think that the RTS snowball makes people think games are more stacked than they were. NS2 has a snowball mechanic where the team winning gets multiplicative advantages that makes it hard for the enemy team to come back. As aliens, if you kill every marine player before they build their naturals then you have a huge advantage that is already hard to overcome.
Shuffle does not do that.
If you were given these 16 players how would you balance the teams? Try to balance these 16 players.
Here is how shuffle would actually balance those teams. Do you notice how the top two players by hive skill are not on the same team?
In comparison, here was the 1-2-2-2 split. Do you not remember the spawn killing without killing an IP that made? I remember games where aliens would just kill you as you spawned without killing the IP just to get their skill value higher. Pepperidge farm remembers.
A lot of people don't understand the value of lifeforms - even if they're bad - and think they go no lifeform or gorge - and they can win.
If aliens don't use their ress they probably never will get to the point of having 70 ress because the "snowballing" prevents this.
Once the marines teched up and aliens still have only skulks and gorges it's so easy to kill the skulks because their is no fade or lerk to draw their fire away.
The same works with marines:
- if aliens have no lifeforms - there is no need to actually buy a lot of shotguns - while when there are tons of fades or ultra-aggressive lerks around... shotguns can push them away / make them more defensive in their playstyle
- Marines who don't use their ress could've in general pushed even more - but if they wait, they are likely to loose more and more mapcontrol.... they wait in the faint chance of getting exos (which they do because aliens don't use their ress yay)...
@skav2 why you're still so low in hivescore has mostly reasons like lack of mapawareness, turtleness like slowly reacting / rotating into fights and stuff... I've seen more than enough 3k hiveskill players with 10-15% accuracy just because they know where they're the most useful :-)
To make your sample of 16 players more accurate make 2 more of them under 1k... then lump all 4 players under 1k with the 3k, 2k, and 1800 player and move everyone else to "balance" it, because that's what shuffle does constantly.
And people still farm kills to this day... I watched a server die last night because of it.
This part breaks pubs for me so hard. According to my observatory profile My score per minute is 11.82 which according to your skill ranks would put me in a 3000-3250 ELO. Yet I am still at about 4k ELO. This is probably due to me playing 6v6 competitively, which is an environment where the skills I have helps me to win more than on pubs. The 2 reasons I can think of why the score per minute is so low for my 4k ELO is either because in 6v6 you have less engagements and therefor don't earn points as much or that my ability to win engagements is much lower compared to other 4k players. (probably both applies to some extent)
Anyway I noticed everytime I join pubs the hive shuffle can't judge me correctly because I jumped into a different environment and the result is that 1 side is getting stomped pretty hard. Unfortunately I can't think of a way to use the stats given to the hive to calculate his winning ability in different environments.
1000 times this. I know that server operators can set minimum and maximum level of allowed skill in the server. I thought this could be used to somewhat acieve that and I saw server that do this. Might be a good idea to get up some more servers that have max 2000 skill or min 1500 and adjust the play now button to put you on servers that have restrictions like that, which you can play on. I think that might be a start but then again I have no idea if the amount of average players online at the same time would allow for a split like that
EDIT: @skav2 sounds for me like you need to play some NSL gathers bro
Because that is the fault of the highskill players amirite? Brb, fetching my victimhoodcloak. The amount of times I get shit for "stacking" is ridiculous, when I am as powerless as everyone else on the server to move after a shuffle. I, or other competent players, don't have magical powers that eligible us to move after a shuffle, or switch to another team when it's full.
People farm kills in a PvP game, where kills is one of the available tools to buy yourself more mapcontrol. Gimme a break. If you see me in the hive killing eggs with a shotgun it's obvious to me, I have a purpose. If I am in there alone, I am biding time til more teammates get there and making sure that skulks can't spawn, playing it safe to secure a win.
EDIT: Or pushing a lane to trick bad skulks to rush me down, proceeding to die over and over again(see mezzanine lane, server hive start). Because, their shit descisions is my fault. Assuming responsibility over your own gameplay is just too mean.
Here's a protip. You want less "stacked" games? Stop winning with gorgetunnels. The only issue here is a ton of bad aliens with inflated elos lacking any sort of skill whatsoever playing skulk. If you have 3,5k elo but skulk like a 1k elo, you're going to have a terrible time when hive assumes you can perform a role you can't. Really isn't more to it, you dug your own hole. Deal with it.
That's not farming kills.. What I'm referring to is when the entire alien team is dead with 4-5+ marines in their main and nobody is shooting the hive!. They're all just camping eggs.
And let's be real, it isn't uncommon to hear "don't end it, let's keep the game going" from some loser who thinks it's "fun" (who is usually in the 1-2k range)
https://observatory.morrolan.ch/player/3683
I probably should since most of my games are spent telling the team how to play. Been wanting to play a serious game where people know what to do.
As for the shuffle system, it would probably be most fair if the top 4 players were placed on opposite teams then everyone else placed randomly to even it out. Or at least top 2 players.
Most of my game I hear this "dont go to |map area| because *x* is there. And X gon give it to ya". We literally only track where the best players go lol. Then if they decide to push him I see this
As @Nordic said before the average player skill is low but to stick to the theme of this thread we just need to get people playing quickly and as fair to both teams as possible.
the ideas so far have been
- Fix shuffle to be more fair
- Gorge tunnel changes due to low cost of use and being super effective
- Queue system like CSGO
^ "Probably wont happen"
- Players Git Good
^ "Pshh yea right."
- Ranked matches based on skill.
^ "This one will be hard due to low player base count"
@Mephilles Try not to put too much stock in those averages. Not everyone follows the trend. Last I looked (a long time ago) my spm was higher but my kdr was lower than my skill bracket at the time. An average is an average which means it doens't describe everybody. Also, the upper end of the skill ranks there get questionable stats because the sample size drops off to 1% per skill bracket.
The problem with skill restricted servers is that they are really hard to seed.
This is why I keep saying that good, realistic behaving AI bots would be an insanely huge improvement to NS2. Especially if they could be controlled by the players in the server somehow. (basically to activate/deactivate them and increase/decrease the difficulty and/or number of bots)
I remember an old server (Tactical Freedom?) that used to have a great seeding mode where all the players got lumped to Marine against a swarm of skulk bots. It was hands down the easiest server to seed that I have ever seen... and that includes other games not just NS2.
If I were in charge at UWE I'd hire an AI programmer to get some really good bot AI.. I think it's an investment that would have a very high chance of increasing the # of people playing. It just solves so many of the problems caused by the game having a small playerbase.
When will ppl stop trying to reason with stubborn, unthinking people who outright ignore facts, common sense, and wouldnt recognize a logical thought in front of them if it bit them on the nose and yelled at them for actual years on end.
Good effort, although clearly wasted as ever it has been.
Player numbers are just not high enough and haven't been for like 4 years to sustain some of the other good ideas in this thread.
Think at a certain point, you must accept the fact that you simply cannot have your cake and eat it too. If devs were able to immediately put a new system in place, it would not also magically beckon thousands of players back to the game to really address the issue of skill gap and games being unfun because of it.
The game is old in terms of modern fps games. The fact that it still gets support is astounding. Personally, i can comfortably say i dont play ns2 anymore. It became a chore and almost never fun for all the reasons echoed in this thread and the countless other threads that have existed in this vein year after year after year. Furthermore, it becomes maddening to see the same ppl (or maybe just a select few) spew the same tired bullshit year after year, deaf entirely to reason.
Whether it's incessant adult children, tryhard edgelords immersed in juvenile meme culture, eternal victims shifting blame, or the righteous community warrriors who wouldnt dare speak a word against the game and are there to defend its honor at any given turn or murmur (e.g. having players kicked for expressing dislike of a feature or map - yes there are ppl this petty, I've seen it happen), there is always a source of grating friction which is not enjoyable. This is the sticky, syrupy, aromatic remenant of the reduced wine that is ns2. There arent a whole lot of level-headed players who are terribly vocal. The ones bitching loudly are usually on the extreme end of any spectrum, as with ANY SUBJECT.
You guys are a special herd, and that is both awesome and terrible.
The shuffle algorithm does it the best way it can. If you tried to balance the skills on paper by hand, you wouldn't be able to do it any better.
A possible solution would be to kick/ban the outliers from the server (v.high skill and the v.low skill), which isn't fun to said victims.
@Nordic Great post. I've seen the ELO distribution chart before, and I want to re-express my disbelief that even after rookie-status the massive low-skill outlier is still there. Do you have a ELO distribution chart of rookie-only players? Can we blame this outlier to the concurrent player dropout problem? If the outliers are legit, there is no way we can allow rookies to "graduate" until they break out of that outlier.
A future patch should consider firmly placing all lowest-skilled players in the "rookie-only" category until they rise above it. It may go a long way for improving the average pub experience for non-rookies.
It could do a better job of balancing out so each team has an equal number of the v.high skill and the v.low skill players... Instead of putting both v.high and v.low skill players all on one team and calling it "balanced" because the numbers look even.
I know I'm wasting my time as I lack the mathematical skills to give an example of what I'm talking about... it's just so frustrating to see people go "oh these teams are balanced" because the averages are close, when in reality one team has all the highest skilled players and is almost guaranteed an easy win.
It makes games either frustrating (on the losing team) or boring (on the winning team) which sucks when all you want is to play a balanced match that goes back and forth.
I understand what you mean in mathematical terms. "The difference in standard deviation between the two teams," in technical terms. If I understand the shuffle algorithm correctly, it minimizes this variable as much as it can, so no, it really can't do a better job. Nordic's example dataset is a good one here, if you look at the "difference in standard deviation of the two teams" of the shuffle algorithm, it is approximately 625 minus 621: equals 4, which is a good number. Nordic showed an example right after that one with a different shuffle method, and the "difference in standard deviation" is much higher, which is bad.
Whenever you experience a truly bad shuffle, it really is the best the algorithm can do, and there is no better way to do it. If you don't believe this assertation at all, you can prove me wrong by screen capping a terrible post-shuffle result and posting it here in the forums. Ideally you'd have to mathematically prove it yourself, but I would be content with just a screencap and an example of a shuffle that would have been better.
If this is the case.. then maybe someone can explain why sometimes you can shuffle and end up with an average that is way off, then people reshuffle and the averages are much closer... (and no, nobody joined or left after the first shuffle)
Does repeated shuffles prioritize the average over the deviation or something?
I've been wondering lately what it would be like to just remove it all... No joining teams, and no ability to shuffle. (just hear me out)
Everyone is put in the ready room at the start.. The doors we currently use could be retextured or whatever to say Khammander/Commander respectively... The first person to walk in each door is the comm for that team, once both sides have a comm teams are autoshuffled and the game starts. (And a reset would put everyone back in the RR)
This would theoretically solve
- repeated shuffles before game start (so.... many.... shuffles!!!)
- High skill players not wanting to command because it will handicap their team (and subsequently having your team handicapped because your best player is comm)
- Games that start when only 1/4th (or less) of the server has joined a team
- Less waiting overall (waiting for commanders, waiting for shuffle, etc)
It would also basically remove the need to have a separate commander score right?
The only downside I see is less of an ability to choose what team you want.. Though with how often I join Alien and get forced over to Marine I doubt it would change much in that regard.
If this is truly the case, it could be giving too much preference to players' team preferences, causing bad shuffles. At the very least a concrete example would help discover whether it's either an error or a preference for low stddev variance. If there is an error, there is no harm in showing a concrete fail-state that can be improved on.
You don't need math skills. Google docs will do everything for you. I made a sheet years ago just for this. Put whatever numbers you want in.
I have actually been taking screenshots of shuffle results for awhile now. I will screenshot if I think about it when shuffle does something weird, or if there is a certain range of skill playing that I want to save the shuffle results of. I have several dozen screenshots of shuffle results. >85% of the shuffle results are weird because people left after the shuffle started. This is not a representative sample at all, so don't take it for that. I have never actually looked at these until now.
Here is a dump of a handful of teams that people didn't leave after shuffle.
[[[Edit: This math was wrong. So very wrong.
Guess what I found? I found one team where the two highest skilled players were on the same team. Look at that. Link to shuffle results. When I first looked at it, the team averages had a difference of about 10, but standard deviation had a difference of nearly 500. My first thought is that the server operator might have set it up so that it had a very loose tolerance for standard deviation. The REALLY weird part is that when I put those numbers into google sheets, the average alien skill calculated to 2653.6 but shuffle is displaying the average alien skill as 2586.9. Shuffle calculated the average incorrectly, which seems to have allowed the two top players to be put on the same team. These teams actually have a difference of 77 hive skill, which gives the alien team about a 60% chance of winning.
I can not begin to explain why shuffle did its math wrong.]]]
There are also two examples in the image dump of a successful shuffle making teams that are not balanced by average team skill at all. As far as I know, this usually happens when the tolerance for standard deviation is too tight, but I can't really explain that either.
First of all, do you have any idea how many times I have tried to explain exactly what squishpoke just said to you Mofo for you to just not believe me? Gah! I am glad you are finally understanding that point though.
Yes. Shuffle prioritizes skill average over the standard deviation. Standard deviation is handled by a tolerance. I believe average skill and standard deviation both have tolerances that are server configurable. Shuffle tries to find teams that are within those tolerances. By default average skill has a much tighter tolerance than STD. Essentially, shuffle tries to find teams where the average skills are as close to equal as possible within a the set STD tolerance. @SupaFred sets his tolerance a little looser because he thinks it gives players a better experience because it allows them to stay on their preferred team more often.
The average that is calculated is correct. Sum the values in a calculator and divide by 9, you'll get 2586.9 rounded to 1 decimal place.
The problem here is likely the very large standard deviation tolerance, which would allow the teams to quickly deteriorate in the name of improving how close the averages are. If you want more balanced teams, the standard deviation tolerance should be low (e.g. the default value of 40 or potentially less if you're feeling optimistic).
Even with the configuration setup, it is still possible for shuffle to get the teams wrong from time to time, simply due to the constraints in play. Not only does it try to respect team preferences and various tolerance values, but it also is limited by execution time and resources.
Currently it uses a hill climbing method to try to find an optimal solution, which is quite cheap in execution and often provides good results. However, this approach can often end up stuck at a local minimum instead of reaching the global minimum, and it will not attempt any further iterations to try to prove that it has found the global minimum. This can mean the final solution may not be the absolute best, and in extreme cases, may in fact be quite far off.
If I had the time and motivation I could implement a smarter algorithm that would either restart the process from a different starting point after finding a solution (though this would throw out part of the team preference bias), allow more solutions to be considered by relaxing the rules (tabu search), or change it to a probabilistic approach such as simulated annealing. However, all of these would increase the time needed to compute the teams and may often result in little improvement, especially when people do not trust the underlying values anyway.
4351 + 3989 + 3610 + 2579 + 2447 + 2253 + 1775 + 1179 + 1099 = 23282
23282 / 9 = 2586.88888888888888888888888...
Not sure how you're getting something different, unless one of our calculators/computers is lying to us.