1200 is too low.. I myself have reached 1600 before and I'm nowhere near the "pro" level..
@MoFo1 No, it really is not. I have stated the average hive scores as close as I could by memory. Maybe you did not see the graph I showed above.
Let me tell you in direct numbers then by looking at the file.
If I account for all players in hive, the average skill level is 883 and the average KDR is .87.
This is actually fairly close to MrChokes skill level and averages.
The problem with those numbers is that most players in hive do not play right now. So here are the numbers of players with 100 or more hours recorded by hive. Using 100 or more hours makes the data more like what we actually see in any given server. Then the the average skill is 1297 and the average KDR is 1.29.
Average is actually not that good a measure for this sort of thing, so how about I tell you the medians. A median is the number separating the higher half of a data sampled from the lower half. In laymans terms this mean the median is the exact middle value.
The median hive skill for everyone in hive is 998. The median for players with 100 or more hours is 1206.
The median for players with 100 or more hours is 1206. This is why I said 1200 hive skill because it is pretty close to the exact middle between the more skilled players and the less skilled players. So you are wrong when you say that 1200 is too low.
If you you are going to split the player base in two, you might as well do it right at the middle. The middle is ~1200 hive skill.
Just because it is an interesting graph that shows my point I will post it again but this time without spoilers.
Now if he limited it to under like 1500-1800 skill that would go a long way towards truly making the server " noob friendly" as it would get rid of most of the people i see in there ruining games...
1500 is too high, as explained above. 1800 is WAY to high. 1500 and above would be a good starting point if you wanted to make a high skilled only server.
The fourth quartile of players, the top 25%, starts at 1560.
The problem with doing this is we can't realistically segment the players. The player counts are too small to do this. Updates like bootcamp are trying to increase the playerbase by making player retention better.
Of course there are still players with low hive skill that have kdr over 2.0.. But that's a problem with being rated purely by win/loss...
There are players with low hive skills and over 2 KDR recorded in hive. These players do not have much data recorded yet and their skill value will be increasing rapidly.
I have explained to you and others numerous times that KDR is incredibly related to Win Loss. By measuring win loss you indirectly measure KDR.
I have began playing with linear regression to try and predict values based on other values. You don't need to know what this is or what it means.
In this graph I tried to predict Win/Loss by using Kill/Death, Score/Minute, Assists/Minute. I found out that Kill/Death is the strongest predictor of Win/Loss by a lot. If you don't believe me just look at this. Ỏ̷͖͈̞̩͎̻̫̫̜͉̠̫͕̭̭̫̫̹̗̹͈̼̠̖͍͚̥͈̮̼͕̠̤̯̻̥̬̗̼̳̤̳̬̪̹͚̞̼̠͕̼̠̦͚̫͔̯̹͉͉̘͎͕̼̣̝͙̱̟̹̩̟̳̦̭͉̮̖̭̣̣̞̙̗̜̺̭̻̥͚͙̝̦̲̱͉͖͉̰̦͎̫̣̼͎͍̠̮͓̹̹͉̤̰̗̙͕͇͔̱͕̭͈̳̗̭͔̘̖̺̮̜̠͖̘͓̳͕̟̠̱̫̤͓͔̘̰̲͙͍͇̙͎̣̼̗̖͙̯͉̠̟͈͍͕̪͓̝̩̦̖̹̼̠̘̮͚̟͉̺̜͍͓̯̳̱̻͕̣̳͉̻̭̭̱͍̪̩̭̺͕̺̼̥̪͖̦̟͎̻̰
All I read now when ppl complain about their hive skill value or the elo system is
"I think I'm better than what this number is telling me, something must be wrong. I want it to be higher. What kind of bullshit system can I propose to make me look better?"
You could restrict it to just the middle 50% skill between 998 to 1560 skill. That would cut out lower skilled and the higher skilled players. This would result in better team shuffles.
Otherwise I don't understand what you mean by headroom?
There are players with low hive skills and over 2 KDR recorded in hive. These players do not have much data recorded yet and their skill value will be increasing rapidly.
Except the guy I saw the other day with <1000 hive score and a kdr over 2.5 and he had hundreds of hours on the hive. That should be plenty of "data"
Also if you want to make a server for average skilled casual players, the worst thing to do would be to split it right down the middle of that skill range. It's the high skill players who are going to stomp the crap out of average pubbers and rookies and make them quit the game... By your own words you say a good level for a high skill server would be 1500 and up... So not sure why you disagreed with that part.
I also don't have much faith in going purely by the numbers when I use sh_teamstats and it shows the "average" skill of both teams as being even, yet one team has three 2000+ players and the highest skill player on the other team is 1200... That is in no way even remotely balanced.
There are players with low hive skills and over 2 KDR recorded in hive. These players do not have much data recorded yet and their skill value will be increasing rapidly.
Except the guy I saw the other day with <1000 hive score and a kdr over 2.5 and he had hundreds of hours on the hive. That should be plenty of "data"
Send me the link of a person like that. Please. I wonder what circumstances lead to him having something like that.
Also if you want to make a server for average skilled casual players, the worst thing to do would be to split it right down the middle of that skill range. It's the high skill players who are going to stomp the crap out of average pubbers and rookies and make them quit the game... By your own words you say a good level for a high skill server would be 1500 and up... So not sure why you disagreed with that part.
If you want to make a server for average skilled servers you would not set a max cap at 1500 or 1800 as you suggested. You would also need a minimum cap.
As I just suggested to ironhorse if you made the range from 998-1560 that would restrict it to the middle 50% of skill.
I suggested to Zero that he limit his Noob friendly server to below 1200 because that is the bottom 50% of players. I was not making a suggestion for the average player, but the lowest skilled.
I also don't have much faith in going purely by the numbers when I use sh_teamstats and it shows the "average" skill of both teams as being even, yet one team has three 2000+ players and the highest skill player on the other team is 1200... That is in no way even remotely balanced.
[/quote]
You may not have faith in going purely by numbers because of the problems with how shuffle balanced a team. What you don't understand is that is explained by numbers also.
Shuffle is a really smart system but it only works with what it has. It won't always result in balances teams and I have explained to you and others in other threads. I will do it again.
To speak specifically about one team has three 2000+ players and the highest skill player on the other team is 1200. That is a problem will skill variance, basically how far apart the skills are on the team. It is very easily explained.
It is not the only problem with shuffle though. The reason shuffle does not always result in balances is mostly, but not entirely, a result from these factors.
1. NS2 is an asymmetric game. Aliens and marines need separate hive skill values.
2. The skill system is a bit too relativistic. Someone who has a high skill value on a no rookies pub server would have a higher skill value on a random pub server where the average skill is lesser.
3. New rookies start with a hive skill of 0, and returning players which are often rookies start from a hive skill of 1000.
4. Role selection. If a player is a really good fade, and has gained a high skill value, if he plays as commander his skill value is no longer reflective on his ability to contribute to a win.
5. Skill variance. When you have a server with players ranging from 500 up to 3000 skill there is a lot of skill variance. It is incredibly hard to balance with high skill variance. This skill variance also further enhances the problems caused by role selection.
6. Team selection. Both FET and shuffle try to keep you on the team you joined. It tries to balance with your preference in mind. This restricts the balance functions ability make the most balances teams possible.
7. Smurfs. One smurf can be so misrepresented he can really mess up the skill function. Not only does a smurf hurt balance, it also messes with everybody else skill value temporarily messing balance up even further. No, this is not a way that you can lose unfairly. The odds of you being on the team with the smurf, or against the team with the smurf are equal.
8. People switch after FET or shuffle was made.
Since you like to use sh_teamstats, do this next time. Look at the standard deviation. The standard deviation is the measure of the skill variance.
It looks like this.
I just got that from a server in ns2. I know the averages are not equal and the teams are clearly unbalanced but that is not the point. Look how big those standard deviations are. 440 and 856. That is huge. In my experience of using sh_teamstats I don't see it go below 400 very often.
The average skill on that server is 1050 which is below average even though there are some very skilled players there. The standard deviation is 771 which shows how hard it is to balance for that set of players.
What I am saying is that shuffle can only balances games so well with what it is given. To make balanced games you need near skilled players. That example server does not. Most servers do not. The lower the standard deviation the more near skilled the players are.
So in and ideal shuffle you would have the average skill be near equal, the standards deviations be near equal, and you want as low of standard deviations as possible. This would result in balanced teams.
The problem is that those ideal scenarios do not exist in ns2. We do not have enough players to segregate skill levels to create that.
I did a test today where I made 100 servers with 18 players each. All 1800 of these fake players had a random skill between 998 and 1560. Meaning that all of these fake players are in the middle 50% of skill.
I then found the standard deviation for all 100 servers. The average standard deviation was 161. The maximum standard deviation was 212.
To summarize, if you limited a server to be between 998 and 1560 skill and shuffled the teams with no team switching, you would have a much much higher chance of having balanced games.
You could restrict it to just the middle 50% skill between 998 to 1560 skill. That would cut out lower skilled and the higher skilled players. This would result in better team shuffles.
Otherwise I don't understand what you mean by headroom?
As this sounds rather interesting i was so free to host a server following this concept. It's avaible as Ghoul's Box #Casuals Only
@ZEROibis, if you are going to have your server be called (NOOB FRIENDLY) you should maybe consider using shine to limit it to players with less than 1200 hive skill. This thread shows that there are people looking for that kind of server.
1200 is too low.. I myself have reached 1600 before and I'm nowhere near the "pro" level..
Now if he limited it to under like 1500-1800 skill that would go a long way towards truly making the server " noob friendly" as it would get rid of most of the people i see in there ruining games...
Of course there are still players with low hive skill that have kdr over 2.0.. But that's a problem with being rated purely by win/loss...
Even I have higher skill than that, I would not be able to join my own server...
Additionally, this starts to split up and kill off the community. Back before they finished killing tribes with things like this I often could never find servers b/c the only ones with players were out of my skill range b/c I was 1 level to high.
Having god-like pros stomping every single round seems to me like it would also kill off the community...
Frankly I'm just tired of seeing games completely ruined by 1-3 high skill players... Yesterday there were six populated servers in NA and they all had 2-3 high skill pros. That's more than enough for them to play against each other instead of ruining game after game after game (after game after game...) across six servers...
I guess 1200 (or even 1000) would still be better than nothing. I'd just be forced to intentionally play like crap on the high skill servers so I can get under 1000..
Anything to play without high skill pros skewing the balance of every single game...
@ghoul. Any chance that server is in NA?? My ping to Euro servers is too high...
You win and lose rounds like everyone else.
So where are all the nonstop losing stomp rounds you descibed?
Your hive history shows that you won 50% of the last rounds.
And 13 of the games where BEFORE februar 2nd (where the rookie only servers where released)
You have lost 7 of these 13 rounds. How is this possible?
So your "nonstop" stomp rounds are the 7 rounds after the "rookie only servers" release?
You won 4 of them btw.
In MrChoke's defence, he was talking about him getting stomped, not his team. I mean, imagine you're a guy in a server where both teams have avg skill of 2000, but you've got 200. It wouldn't mean your team would necessarily be stomped, but you as an individual might be. Nordic states the average skill is about 980, so if we take Choke to be between 700 and 800 and that he has no access to the rookie only servers (where the avg skill is, I presume, well below 980), it is actually quite reasonable to expect that he as an individual player is being stomped because outside of such servers the average should be a fair bit above 980.
SantaClaws suggests a reasonable alternative though (which I think might be something I have said in a previous post too?), where you isolate servers on the basis of skill and not hours played. i.e. skill score not level.
Even I have higher skill than that, I would not be able to join my own server...
Additionally, this starts to split up and kill off the community. Back before they finished killing tribes with things like this I often could never find servers b/c the only ones with players were out of my skill range b/c I was 1 level to high.
And bootcamp is not killing off the community??? I assure it is. I would prefer bootcamp get pulled. But I guess that is not going to happen. At leas the skill-based cap idea for servers will mitigate the mess bootcamp has created.
Frankly I'm just tired of seeing games completely ruined by 1-3 high skill players... Yesterday there were six populated servers in NA and they all had 2-3 high skill pros. That's more than enough for them to play against each other instead of ruining game after game after game (after game after game...) across six servers...
Do you think it doesn't happen to me? I have a hive skill of 1800 and I get stomped on by higher skilled players also. I also experience the pain of playing with players of much lower skill level than I am.
This is not an easy problem to fix because NS2's playercounts are just that low. NS2 has roughly a peak 24 player count of 471 right now. That is not enough for skill segregation. A similar asymmetric game called Nosgoth had a peak 24 hour player count of over 3000 and it was still not enough for skill segregation.
I guess 1200 (or even 1000) would still be better than nothing. I'd just be forced to intentionally play like crap on the high skill servers so I can get under 1000..
Splitting it down the middle at 1200 is one way to do it. If you were going to do a rookie noob friendly server I would probably say any player less than 1000 or 1200 hive skill would be appropriate.
While this could be attributed to a number of sales, the increase in new player retention is clearly from the changes we're making to the new player experience.
I guess another part is, that the official developer studio of this game seems to actually fully care about its game again.
CDT has been pretty great from what i see (especially want to mention ns2+ and the loading time improvements), but everything is on a different scope once there is money behind it again.
At least my reason to look into ns2 again after 2 years of inactivity is just because of that.
(edit: but i'm not really a new player, ill keep the post up anyway, since i think its still worth to be mentioned ...)
PS: the worst thing that happened during my inactivity, ns2:combat. I miss combat.
PPS:i am kinda disappointed that in those 2 years, mechanics and gameplay-wise barely anything really changed, the biggest thing imo seems to be the clunky implemented switch from shadowstep to metabolize having it on the shift key and as a main weapon slot... consistent inconsistent, but since we don't have decent keybinding options still, its okay that way.
While this could be attributed to a number of sales, the increase in new player retention is clearly from the changes we're making to the new player experience.
I guess another part is, that the official developer studio of this game seems to actually fully care about its game again.
CDT has been pretty great from what i see (especially want to mention ns2+ and the loading time improvements), but everything is on a different scope once there is money behind it again.
At least my reason to look into ns2 again after 2 years of inactivity is just because of that.
(edit: but i'm not really a new player, ill keep the post up anyway, since i think its still worth to be mentioned ...)
PS: the worst thing that happened during my inactivity, ns2:combat. I miss combat.
PPS:i am kinda disappointed that in those 2 years, mechanics and gameplay-wise barely anything really changed, the biggest thing imo seems to be the clunky implemented switch from shadowstep to metabolize having it on the shift key and as a main weapon slot... consistent inconsistent, but since we don't have decent keybinding options still, its okay that way.
NS2+ has never been part of the CDT scope, that's always been my own thing, and still is. Unless you mean the merging of numerous NS2+ features and improvements into vanilla, which did happen during the CDT period. Although in the end it's the same to the end-user, considering the volume of servers running the mod.
dePARAJoin Date: 2011-04-29Member: 96321Members, Squad Five Blue
edited February 2016
I was always for skill split servers. I think i told that countless times over the past years but nothing happen.
While i hosted a high level server and tried to keep it that way, we had no real low to mid servers back in the days.
Now its a bit to late with the low playerbase for this kind of "full split"
Just give the rookie only server concept with the upcoming tutorial improvements more time.
Its the only chance we have to hold atleast some of these players.
And i must say:
If the rookie only servers can help to increase the playerbase by 100 or more, well, who cares about the 5 players who dont like the this change?
In the end the change was successfull.
So instead of whining and complaining, how about a break and come back later after some of the rookies left the green state?
Dear UWE,
As you finally decided to split the playerbase please address it more reasonably. My proposition:
1. Only 2 types of servers: for LOW-MID skilled players (only!) and for MID-HIGH skilled players (only!)
2. Divide Hive skills in 3 ranges:
- LOW (25%)
- MED (50%, or even more, so the great majority of players can play on EVERY server
- HIGH(25%)
3. Clear information in "Main Menu": what is my current skill, and similar in server browser.
Sincerely,
Casual players
Players who are completely new to the game finally have a place now where they can learn the game without getting stomped mercilessly, and still people find a way to complain about it.
dePARAJoin Date: 2011-04-29Member: 96321Members, Squad Five Blue
Looks like everyone need the rookies:
- high skill: to warm up against easy to shoot targets
- med skill: to feel a bit like a pro
- low skill: to atleast make some kills
Shame on you UWE, you ruined the game for everyone by removing the rookies.
Meanwhile the rookies can have some fun on there servers or can join "normal" servers to learn NS2 the old hard way.
Players who are completely new to the game finally have a place now where they can learn the game without getting stomped mercilessly, and still people find a way to complain about it.
People should be able to play the game and have fun - if you're not able to improve such that you can you will likely stop playing. Understandable. In other words, people who come out of the rookie stage and don't improve that much over time are going to get repeatedly rolled until they just quit.
i.e. rookies can learn without getting stomped if they play with someone like the OP, so people like the OP should be able to play with them. This invalidates the current system as it could be improved by segregating by skill. That doesn't mean we need 10000 tiers - 2 or 3 would probably be enough. Severs for people with a skill below 1200 and those above that. That might then create the problem of "Hey my skill is 1500 and I can't compete with people who are above 2000". Yeah, you can't, but so many people don't have a skill level that high anyway.
It just seems weird to make people who have 50 hours in the game with a skill score of 400 play on a server where the average skill is 1200. Teams so rarely have an average skill of 2500 that a player with a skill of 1500 would rarely feel like he's useless.
It just seems weird to make people who have 50 hours in the game with a skill score of 400 play on a server where the average skill is 1200. Teams so rarely have an average skill of 2500 that a player with a skill of 1500 would rarely feel like he's useless.
In an ideal world we would segment the players, but that is not what we have. What we have is a peak of 479 players. With that few players we can hardly afford to segment the players with that few of us.
Server operators have had the ability to segment servers for a long time now. Those servers have been mildly popular at best.
One problem with segmenting is that the bottom and top 10% get screwed if we don't have enough players, which we do not.
@BestProfileName
Again, this is not about UWE doing anything. You don't need them doing absolutely anything, you have billions of empty servers around. What you are asking requires:
1 - servers admins in your region agreeing in a simple skill distribution for their servers.
2 - players in your region agreeing to play in the server of their skill, regardless of the community server who likes and that he probably have played on for years
3 - players populating servers instead of coming to the forums to complain.
And even if all this happens (which is unlikely at the moment), many players will come to complain here because:
a) "Now you killed my community, which was the only reason I was still playing NS2"
b) "Now there is so much segregation that servers are always full or empty so I am never able to play"
Edit: if that isn't enough, then players want even more segregation because they look for servers with highly different number of slots...
TL;DR: We do what we can with the playercount that we have. Just don't annoy the players with less than 10 hours. That doesn't seem too much to ask.
Although I don't think servers should be segmented yet, any server operator want help finding a range I can help. I have the the skill distribution data to give you exact ranges.
For example the middle 50% of players sit between 998 and 1560 hive skill.
If you wanted the to only allow players within the 10-40th percentile, I can tell you those values.
Protip for casuals: Play during not so high player count hours. Like for me it's somewhere 15pm- 19pm CET. After dark, all pro comp players crawl out of theirs caves and you see guys like 3000 skill.
Another tip: Find good community server which suits your playstyle. Ppl will know you after while whatever your skill is.
FrozenNew York, NYJoin Date: 2010-07-02Member: 72228Members, Constellation
I sign on to NS2 now with twice as many populated servers to choose from. I have no idea what the fuck anyone is talking about. NA East is better than it was, except no rookies pub was killed by gathers.
I have began playing with linear regression to try and predict values based on other values. You don't need to know what this is or what it means.
In this graph I tried to predict Win/Loss by using Kill/Death, Score/Minute, Assists/Minute. I found out that Kill/Death is the strongest predictor of Win/Loss by a lot. If you don't believe me just look at this.
How come your prediction ("r" value I assume) is so different from the correlation values from the other thread?
How come your prediction ("r" value I assume) is so different from the correlation values from the other thread?
W/L <-> K/D : 0.546
W/L <-> SPM : 0.434
Although we both are working with the same data, I don't know what subset of data moultano was using. Those correlations you have are from the subset of the whole data that has only players with 100 or more hours recorded in hive and a large amount of level 1000 players removed. If I have given different correlations in different threads it is because I keep jumping around to different subsets of data. I am doing this for fun with no real purpose to it, so I may lack a little bit of consistency.
Moultano was not using Win/Loss or Kill/Death either. He was using win rates and kill rates.
Win rate = wins/(wins+losses)
kill rate = kills/(kills+losses)
As I have stated in another thread my hive data is only ~77% of all the players that were recorded in hive in March 2015. There are 59,489 players in this data in total. A lot of these players are junk data which is why I have have been using different subsets of that data.
Here is a correlation table I got out of SPSS today with all the data, no subsets. I was playing with killrates and winrates when I made this instead of W/L and K/D.
Just played a 3 minute game where the oh so great shuffle put a 3300+ player and a 2700+ player against a team of 1600 and below players... Oh and each team only had 5 people...
The "average" skill of both teams was off by less than 100 points also... Gotta love the awesome math producing those numbers...
Just played a 3 minute game where the oh so great shuffle put a 3300+ player and a 2700+ player against a team of 1600 and below players... Oh and each team only had 5 people...
The "average" skill of both teams was off by less than 100 points also... Gotta love the awesome math producing those numbers...
Because numbers can never be wrong.
No one, not even me, says that the numbers are perfect and can never be wrong. I even explained to you multiple times why balancing with such large standard deviations in skill level often produces imbalanced games.
I also asked you to stop looking only at the average skill, but also at the standard deviation showed by sh_teamstats. So for that game where the averages were close, what were the standard deviations of the teams? This is a serious question.
Also, this weekend Rusty of DiamondGamer's will also be running a server that is limited to the middle 50% of the skill range between 998 and 1560 hive skill. This is an NA server. I have requested that I be whitelisted so I can join and observe from time to time. I will not be playing any games though. I hope to see you there.
Comments
Let me tell you in direct numbers then by looking at the file.
If I account for all players in hive, the average skill level is 883 and the average KDR is .87.
This is actually fairly close to MrChokes skill level and averages.
The problem with those numbers is that most players in hive do not play right now. So here are the numbers of players with 100 or more hours recorded by hive. Using 100 or more hours makes the data more like what we actually see in any given server. Then the the average skill is 1297 and the average KDR is 1.29.
Average is actually not that good a measure for this sort of thing, so how about I tell you the medians. A median is the number separating the higher half of a data sampled from the lower half. In laymans terms this mean the median is the exact middle value.
The median hive skill for everyone in hive is 998. The median for players with 100 or more hours is 1206.
The median for players with 100 or more hours is 1206. This is why I said 1200 hive skill because it is pretty close to the exact middle between the more skilled players and the less skilled players. So you are wrong when you say that 1200 is too low.
If you you are going to split the player base in two, you might as well do it right at the middle. The middle is ~1200 hive skill.
Just because it is an interesting graph that shows my point I will post it again but this time without spoilers.
1500 is too high, as explained above. 1800 is WAY to high. 1500 and above would be a good starting point if you wanted to make a high skilled only server.
The fourth quartile of players, the top 25%, starts at 1560.
The problem with doing this is we can't realistically segment the players. The player counts are too small to do this. Updates like bootcamp are trying to increase the playerbase by making player retention better.
There are players with low hive skills and over 2 KDR recorded in hive. These players do not have much data recorded yet and their skill value will be increasing rapidly.
I have explained to you and others numerous times that KDR is incredibly related to Win Loss. By measuring win loss you indirectly measure KDR.
I have began playing with linear regression to try and predict values based on other values. You don't need to know what this is or what it means.
In this graph I tried to predict Win/Loss by using Kill/Death, Score/Minute, Assists/Minute. I found out that Kill/Death is the strongest predictor of Win/Loss by a lot. If you don't believe me just look at this. Ỏ̷͖͈̞̩͎̻̫̫̜͉̠̫͕̭̭̫̫̹̗̹͈̼̠̖͍͚̥͈̮̼͕̠̤̯̻̥̬̗̼̳̤̳̬̪̹͚̞̼̠͕̼̠̦͚̫͔̯̹͉͉̘͎͕̼̣̝͙̱̟̹̩̟̳̦̭͉̮̖̭̣̣̞̙̗̜̺̭̻̥͚͙̝̦̲̱͉͖͉̰̦͎̫̣̼͎͍̠̮͓̹̹͉̤̰̗̙͕͇͔̱͕̭͈̳̗̭͔̘̖̺̮̜̠͖̘͓̳͕̟̠̱̫̤͓͔̘̰̲͙͍͇̙͎̣̼̗̖͙̯͉̠̟͈͍͕̪͓̝̩̦̖̹̼̠̘̮͚̟͉̺̜͍͓̯̳̱̻͕̣̳͉̻̭̭̱͍̪̩̭̺͕̺̼̥̪͖̦̟͎̻̰
"I think I'm better than what this number is telling me, something must be wrong. I want it to be higher. What kind of bullshit system can I propose to make me look better?"
So sick of hearing about it in servers and here.
Otherwise I don't understand what you mean by headroom?
Except the guy I saw the other day with <1000 hive score and a kdr over 2.5 and he had hundreds of hours on the hive. That should be plenty of "data"
Also if you want to make a server for average skilled casual players, the worst thing to do would be to split it right down the middle of that skill range. It's the high skill players who are going to stomp the crap out of average pubbers and rookies and make them quit the game... By your own words you say a good level for a high skill server would be 1500 and up... So not sure why you disagreed with that part.
I also don't have much faith in going purely by the numbers when I use sh_teamstats and it shows the "average" skill of both teams as being even, yet one team has three 2000+ players and the highest skill player on the other team is 1200... That is in no way even remotely balanced.
If you want to make a server for average skilled servers you would not set a max cap at 1500 or 1800 as you suggested. You would also need a minimum cap.
As I just suggested to ironhorse if you made the range from 998-1560 that would restrict it to the middle 50% of skill.
I suggested to Zero that he limit his Noob friendly server to below 1200 because that is the bottom 50% of players. I was not making a suggestion for the average player, but the lowest skilled.
[/quote]
You may not have faith in going purely by numbers because of the problems with how shuffle balanced a team. What you don't understand is that is explained by numbers also.
Shuffle is a really smart system but it only works with what it has. It won't always result in balances teams and I have explained to you and others in other threads. I will do it again.
To speak specifically about one team has three 2000+ players and the highest skill player on the other team is 1200. That is a problem will skill variance, basically how far apart the skills are on the team. It is very easily explained.
It is not the only problem with shuffle though. The reason shuffle does not always result in balances is mostly, but not entirely, a result from these factors.
2. The skill system is a bit too relativistic. Someone who has a high skill value on a no rookies pub server would have a higher skill value on a random pub server where the average skill is lesser.
3. New rookies start with a hive skill of 0, and returning players which are often rookies start from a hive skill of 1000.
4. Role selection. If a player is a really good fade, and has gained a high skill value, if he plays as commander his skill value is no longer reflective on his ability to contribute to a win.
5. Skill variance. When you have a server with players ranging from 500 up to 3000 skill there is a lot of skill variance. It is incredibly hard to balance with high skill variance. This skill variance also further enhances the problems caused by role selection.
6. Team selection. Both FET and shuffle try to keep you on the team you joined. It tries to balance with your preference in mind. This restricts the balance functions ability make the most balances teams possible.
7. Smurfs. One smurf can be so misrepresented he can really mess up the skill function. Not only does a smurf hurt balance, it also messes with everybody else skill value temporarily messing balance up even further. No, this is not a way that you can lose unfairly. The odds of you being on the team with the smurf, or against the team with the smurf are equal.
8. People switch after FET or shuffle was made.
Since you like to use sh_teamstats, do this next time. Look at the standard deviation. The standard deviation is the measure of the skill variance.
It looks like this.
I just got that from a server in ns2. I know the averages are not equal and the teams are clearly unbalanced but that is not the point. Look how big those standard deviations are. 440 and 856. That is huge. In my experience of using sh_teamstats I don't see it go below 400 very often.
The average skill on that server is 1050 which is below average even though there are some very skilled players there. The standard deviation is 771 which shows how hard it is to balance for that set of players.
What I am saying is that shuffle can only balances games so well with what it is given. To make balanced games you need near skilled players. That example server does not. Most servers do not. The lower the standard deviation the more near skilled the players are.
So in and ideal shuffle you would have the average skill be near equal, the standards deviations be near equal, and you want as low of standard deviations as possible. This would result in balanced teams.
The problem is that those ideal scenarios do not exist in ns2. We do not have enough players to segregate skill levels to create that.
I did a test today where I made 100 servers with 18 players each. All 1800 of these fake players had a random skill between 998 and 1560. Meaning that all of these fake players are in the middle 50% of skill.
I then found the standard deviation for all 100 servers. The average standard deviation was 161. The maximum standard deviation was 212.
To summarize, if you limited a server to be between 998 and 1560 skill and shuffled the teams with no team switching, you would have a much much higher chance of having balanced games.
As this sounds rather interesting i was so free to host a server following this concept. It's avaible as Ghoul's Box #Casuals Only
Even I have higher skill than that, I would not be able to join my own server...
Additionally, this starts to split up and kill off the community. Back before they finished killing tribes with things like this I often could never find servers b/c the only ones with players were out of my skill range b/c I was 1 level to high.
Frankly I'm just tired of seeing games completely ruined by 1-3 high skill players... Yesterday there were six populated servers in NA and they all had 2-3 high skill pros. That's more than enough for them to play against each other instead of ruining game after game after game (after game after game...) across six servers...
I guess 1200 (or even 1000) would still be better than nothing. I'd just be forced to intentionally play like crap on the high skill servers so I can get under 1000..
Anything to play without high skill pros skewing the balance of every single game...
@ghoul. Any chance that server is in NA?? My ping to Euro servers is too high...
In MrChoke's defence, he was talking about him getting stomped, not his team. I mean, imagine you're a guy in a server where both teams have avg skill of 2000, but you've got 200. It wouldn't mean your team would necessarily be stomped, but you as an individual might be. Nordic states the average skill is about 980, so if we take Choke to be between 700 and 800 and that he has no access to the rookie only servers (where the avg skill is, I presume, well below 980), it is actually quite reasonable to expect that he as an individual player is being stomped because outside of such servers the average should be a fair bit above 980.
SantaClaws suggests a reasonable alternative though (which I think might be something I have said in a previous post too?), where you isolate servers on the basis of skill and not hours played. i.e. skill score not level.
Do you think it doesn't happen to me? I have a hive skill of 1800 and I get stomped on by higher skilled players also. I also experience the pain of playing with players of much lower skill level than I am.
This is not an easy problem to fix because NS2's playercounts are just that low. NS2 has roughly a peak 24 player count of 471 right now. That is not enough for skill segregation. A similar asymmetric game called Nosgoth had a peak 24 hour player count of over 3000 and it was still not enough for skill segregation.
Splitting it down the middle at 1200 is one way to do it. If you were going to do a rookie noob friendly server I would probably say any player less than 1000 or 1200 hive skill would be appropriate.
The fix is the get more people playing ns2. Bootcamp is attempting to do just that.
His server is EU, but I was trying to find an NA server operator willing to do an experiment.
I guess another part is, that the official developer studio of this game seems to actually fully care about its game again.
CDT has been pretty great from what i see (especially want to mention ns2+ and the loading time improvements), but everything is on a different scope once there is money behind it again.
At least my reason to look into ns2 again after 2 years of inactivity is just because of that.
(edit: but i'm not really a new player, ill keep the post up anyway, since i think its still worth to be mentioned ...)
PS: the worst thing that happened during my inactivity, ns2:combat. I miss combat.
PPS:i am kinda disappointed that in those 2 years, mechanics and gameplay-wise barely anything really changed, the biggest thing imo seems to be the clunky implemented switch from shadowstep to metabolize having it on the shift key and as a main weapon slot... consistent inconsistent, but since we don't have decent keybinding options still, its okay that way.
NS2+ has never been part of the CDT scope, that's always been my own thing, and still is. Unless you mean the merging of numerous NS2+ features and improvements into vanilla, which did happen during the CDT period. Although in the end it's the same to the end-user, considering the volume of servers running the mod.
While i hosted a high level server and tried to keep it that way, we had no real low to mid servers back in the days.
Now its a bit to late with the low playerbase for this kind of "full split"
Just give the rookie only server concept with the upcoming tutorial improvements more time.
Its the only chance we have to hold atleast some of these players.
And i must say:
If the rookie only servers can help to increase the playerbase by 100 or more, well, who cares about the 5 players who dont like the this change?
In the end the change was successfull.
So instead of whining and complaining, how about a break and come back later after some of the rookies left the green state?
As you finally decided to split the playerbase please address it more reasonably. My proposition:
1. Only 2 types of servers: for LOW-MID skilled players (only!) and for MID-HIGH skilled players (only!)
2. Divide Hive skills in 3 ranges:
- LOW (25%)
- MED (50%, or even more, so the great majority of players can play on EVERY server
- HIGH(25%)
3. Clear information in "Main Menu": what is my current skill, and similar in server browser.
Sincerely,
Casual players
Guess i should have worded it differently...
While there has been some nice progress during and around the CDT timeframe, especially loading times, or NS+ by Mendasp...
- high skill: to warm up against easy to shoot targets
- med skill: to feel a bit like a pro
- low skill: to atleast make some kills
Shame on you UWE, you ruined the game for everyone by removing the rookies.
Meanwhile the rookies can have some fun on there servers or can join "normal" servers to learn NS2 the old hard way.
People should be able to play the game and have fun - if you're not able to improve such that you can you will likely stop playing. Understandable. In other words, people who come out of the rookie stage and don't improve that much over time are going to get repeatedly rolled until they just quit.
i.e. rookies can learn without getting stomped if they play with someone like the OP, so people like the OP should be able to play with them. This invalidates the current system as it could be improved by segregating by skill. That doesn't mean we need 10000 tiers - 2 or 3 would probably be enough. Severs for people with a skill below 1200 and those above that. That might then create the problem of "Hey my skill is 1500 and I can't compete with people who are above 2000". Yeah, you can't, but so many people don't have a skill level that high anyway.
It just seems weird to make people who have 50 hours in the game with a skill score of 400 play on a server where the average skill is 1200. Teams so rarely have an average skill of 2500 that a player with a skill of 1500 would rarely feel like he's useless.
Server operators have had the ability to segment servers for a long time now. Those servers have been mildly popular at best.
One problem with segmenting is that the bottom and top 10% get screwed if we don't have enough players, which we do not.
Again, this is not about UWE doing anything. You don't need them doing absolutely anything, you have billions of empty servers around. What you are asking requires:
1 - servers admins in your region agreeing in a simple skill distribution for their servers.
2 - players in your region agreeing to play in the server of their skill, regardless of the community server who likes and that he probably have played on for years
3 - players populating servers instead of coming to the forums to complain.
And even if all this happens (which is unlikely at the moment), many players will come to complain here because:
a) "Now you killed my community, which was the only reason I was still playing NS2"
b) "Now there is so much segregation that servers are always full or empty so I am never able to play"
Edit: if that isn't enough, then players want even more segregation because they look for servers with highly different number of slots...
TL;DR: We do what we can with the playercount that we have. Just don't annoy the players with less than 10 hours. That doesn't seem too much to ask.
For example the middle 50% of players sit between 998 and 1560 hive skill.
If you wanted the to only allow players within the 10-40th percentile, I can tell you those values.
Another tip: Find good community server which suits your playstyle. Ppl will know you after while whatever your skill is.
How come your prediction ("r" value I assume) is so different from the correlation values from the other thread?
W/L <-> K/D : 0.546
W/L <-> SPM : 0.434
Although we both are working with the same data, I don't know what subset of data moultano was using. Those correlations you have are from the subset of the whole data that has only players with 100 or more hours recorded in hive and a large amount of level 1000 players removed. If I have given different correlations in different threads it is because I keep jumping around to different subsets of data. I am doing this for fun with no real purpose to it, so I may lack a little bit of consistency.
Moultano was not using Win/Loss or Kill/Death either. He was using win rates and kill rates.
Win rate = wins/(wins+losses)
kill rate = kills/(kills+losses)
As I have stated in another thread my hive data is only ~77% of all the players that were recorded in hive in March 2015. There are 59,489 players in this data in total. A lot of these players are junk data which is why I have have been using different subsets of that data.
Here is a correlation table I got out of SPSS today with all the data, no subsets. I was playing with killrates and winrates when I made this instead of W/L and K/D.
The "average" skill of both teams was off by less than 100 points also... Gotta love the awesome math producing those numbers...
Because numbers can never be wrong.
No one, not even me, says that the numbers are perfect and can never be wrong. I even explained to you multiple times why balancing with such large standard deviations in skill level often produces imbalanced games.
I also asked you to stop looking only at the average skill, but also at the standard deviation showed by sh_teamstats. So for that game where the averages were close, what were the standard deviations of the teams? This is a serious question.
Also, this weekend Rusty of DiamondGamer's will also be running a server that is limited to the middle 50% of the skill range between 998 and 1560 hive skill. This is an NA server. I have requested that I be whitelisted so I can join and observe from time to time. I will not be playing any games though. I hope to see you there.