Do you blame everyone who comes at you with fact-backed arguments about echo chambers and groupthink? What do you think about vaccines and global warming?
Care to elaborate why you never even try to counter any arguments made by these so-called "self-validators", apart from shouting oppression?
Since you bring it up, "global warming" hasn't actually happened for the past 20 years. But don't let facts get in your way. Just keep telling yourself that those 20,000 "experts" who signed that document, who mostly aren't even scientists, who gain financially from "carbon trading" and sales of "eco-friendly" lightbulbs, and who fly around in tax-payer funded private jets to remote paradise islands for their "conferences," are peddling their mumbo-jumbo solely for the good of humanity.
Looks like the unpaid professional commentator NS2 groupthinkers support the real-world groupthinkers! Why am I not surprized that you would be a psuedo-science alarmist. I'm out.
the hive skill works in most cases but not always, for example the game kept balancing the same teams but each time aliens would get completely slaughted within 5 mins even though both teams had 1600 hive skill. this happened 3x in a row with teams being relatively the same each time.
Everytime i hear about hive "skill" i get a head ache . Maybe because i'm stupid. Maybe because i'm bad. Probably because i am confused... But why is it called a skill when it is not measuring anything than your ability to be on the winning team? As been said before, numbers should have been hidden, but they are not!
Agree 100% with this. Numbers should be hidden. Sadly those in control of the "skill" system like to brandish their own high numbers to prove... something.
Do you blame everyone who comes at you with fact-backed arguments about echo chambers and groupthink? What do you think about vaccines and global warming?
Care to elaborate why you never even try to counter any arguments made by these so-called "self-validators", apart from shouting oppression?
Since you bring it up, "global warming" hasn't actually happened for the past 20 years. But don't let facts get in your way. Just keep telling yourself that those 20,000 "experts" who signed that document, who mostly aren't even scientists, who gain financially from "carbon trading" and sales of "eco-friendly" lightbulbs, and who fly around in tax-payer funded private jets to remote paradise islands for their "conferences," are peddling their mumbo-jumbo solely for the good of humanity.
Looks like the unpaid professional commentator NS2 groupthinkers support the real-world groupthinkers! Why am I not surprized that you would be a psuedo-science alarmist. I'm out.
the hive skill works in most cases but not always, for example the game kept balancing the same teams but each time aliens would get completely slaughted within 5 mins even though both teams had 1600 hive skill. this happened 3x in a row with teams being relatively the same each time.
This shouldn't happen unless hive is timing out or something. (Or unless a different algorithm was being used that didn't use hive? There are so many things out there that do something different that it's hard to tell what is being used on any particular shuffle.)
After the first win, the scores should change. Everyone on the winning team goes up, and everyone on the losing team goes down. This makes the original teams no longer balanced according to hive, so the next shuffle should mix them up substantially. I can see this maybe happening though if there's one god-player on the server whose skill is substantially underestimated and much better than everyone else. It might take a while to push his still up enough that teams are correctly stacked against him.
The true skill system adjusts skill individually for each player while hive adjust the whole team by the same amount. TrueSkill adjusts based on both skill and certainty while hive adjusts based only on skill.
If you scroll to the bottom you can see how these two players hive skill have changed over time compared to their TrueSkill MMR. Hive has these massive spikes going both up and down. Hive is occilating as @Therius described. In comparison the TrueSkill MMR does not have these spikes. Another thing I notice is that the trends are the same, even though hive oscillates. I may just be looking at only two players, but still, it is interesting that hive and TrueSkill MMR both rise at similar times or remain steady around the same skill level.
moultanoCreator of ns_shiva.Join Date: 2002-12-14Member: 10806Members, NS1 Playtester, Contributor, Constellation, NS2 Playtester, Squad Five Blue, Reinforced - Shadow, WC 2013 - Gold, NS2 Community Developer, Pistachionauts
TrueSkill is unfortunately patented, so we probably can't use it in a commercial game, but the basic idea is easy to implement in an approximate form. The gist is that the size updates to each player's skill are inversely related to the confidence we have in that skill, and the confidence in the skill value tends to go up over time.
the hive skill works in most cases but not always, for example the game kept balancing the same teams but each time aliens would get completely slaughted within 5 mins even though both teams had 1600 hive skill. this happened 3x in a row with teams being relatively the same each time.
This shouldn't happen unless hive is timing out or something. (Or unless a different algorithm was being used that didn't use hive? There are so many things out there that do something different that it's hard to tell what is being used on any particular shuffle.)
After the first win, the scores should change. Everyone on the winning team goes up, and everyone on the losing team goes down. This makes the original teams no longer balanced according to hive, so the next shuffle should mix them up substantially. I can see this maybe happening though if there's one god-player on the server whose skill is substantially underestimated and much better than everyone else. It might take a while to push his still up enough that teams are correctly stacked against him.
The main problem is probably that it is a 1600 mean rather a 1600 median. A lot of games really are competitions between the top two-three players of each team and the rest are just fodder. Without teamwork, it is going to be hard to dislodge a good marine in a good position especially when they go in one by one without lifeforms to get slaughtered and waste eggs. A slight change in skill/players isn't going to reverse this because marines have an inherent advantage over skulks and fodder is still going to be fodder on different teams.
the hive skill works in most cases but not always, for example the game kept balancing the same teams but each time aliens would get completely slaughted within 5 mins even though both teams had 1600 hive skill. this happened 3x in a row with teams being relatively the same each time.
This shouldn't happen unless hive is timing out or something. (Or unless a different algorithm was being used that didn't use hive? There are so many things out there that do something different that it's hard to tell what is being used on any particular shuffle.)
After the first win, the scores should change. Everyone on the winning team goes up, and everyone on the losing team goes down. This makes the original teams no longer balanced according to hive, so the next shuffle should mix them up substantially. I can see this maybe happening though if there's one god-player on the server whose skill is substantially underestimated and much better than everyone else. It might take a while to push his still up enough that teams are correctly stacked against him.
The main problem is probably that it is a 1600 mean rather a 1600 median. A lot of games really are competitions between the top two-three players of each team and the rest are just fodder. Without teamwork, it is going to be hard to dislodge a good marine in a good position especially when they go in one by one without lifeforms to get slaughtered and waste eggs. A slight change in skill/players isn't going to reverse this because marines have an inherent advantage over skulks and fodder is still going to be fodder on different teams.
If that's true, then the distribution of skills will adjust to reflect that. The top tier players will achieve skill values so high that the skills of lower players are never sufficient to balance out the teams on their own. You might be able to build this assumption into the system by doing a more "max-like" combination of player skills, like a power mean https://en.wikipedia.org/wiki/Generalized_mean, but I would be hesitant to build this assumption into the skill system directly, since at least in principle, the current system could effectively learn that over time.
the hive skill works in most cases but not always, for example the game kept balancing the same teams but each time aliens would get completely slaughted within 5 mins even though both teams had 1600 hive skill. this happened 3x in a row with teams being relatively the same each time.
This shouldn't happen unless hive is timing out or something. (Or unless a different algorithm was being used that didn't use hive? There are so many things out there that do something different that it's hard to tell what is being used on any particular shuffle.)
After the first win, the scores should change. Everyone on the winning team goes up, and everyone on the losing team goes down. This makes the original teams no longer balanced according to hive, so the next shuffle should mix them up substantially. I can see this maybe happening though if there's one god-player on the server whose skill is substantially underestimated and much better than everyone else. It might take a while to push his still up enough that teams are correctly stacked against him.
The main problem is probably that it is a 1600 mean rather a 1600 median. A lot of games really are competitions between the top two-three players of each team and the rest are just fodder. Without teamwork, it is going to be hard to dislodge a good marine in a good position especially when they go in one by one without lifeforms to get slaughtered and waste eggs. A slight change in skill/players isn't going to reverse this because marines have an inherent advantage over skulks and fodder is still going to be fodder on different teams.
If that's true, then the distribution of skills will adjust to reflect that. The top tier players will achieve skill values so high that the skills of lower players are never sufficient to balance out the teams on their own. You might be able to build this assumption into the system by doing a more "max-like" combination of player skills, like a power mean https://en.wikipedia.org/wiki/Generalized_mean, but I would be hesitant to build this assumption into the skill system directly, since at least in principle, the current system could effectively learn that over time.
I don't think its a flaw in the hive skill measuring system. Its just that sometimes, with the current pool of players on the server, you are not going to get a good game because skill is imbalanced on the server itself. No matter how you split the teams, the difference will only be either laughably bad (most uneven teams) or tortuously drawn out (most even teams).
The skill system would be a lot more respected if we had the player base and matchmaking system to group players of similar skill together rather than the current unfortunate situation of having to try and balance players of all skills in a game.
The skill system would be a lot more respected if we had the player base and matchmaking system to group players of similar skill together rather than the current unfortunate situation of having to try and balance players of all skills in a game.
Except that matchmaking removes your ability to choose what server you join.. Which makes the game not worth playing in the first place.
The skill system would be a lot more respected if we had the player base and matchmaking system to group players of similar skill together rather than the current unfortunate situation of having to try and balance players of all skills in a game.
NS2 can not support large scale matchmaking. The playerbase is too small and the skill ceiling is too high. Forced matchmaking would make for a poor experience. The player base is many times too small to find near skilled games.
There is a long explanation of why that is true in this spoiler.
I used to play a game called Nosgoth. I played it for about 300 hours according to steam. Nosgoth is a similar game to ns2 in a lot of ways. It is a third person FPS vs a third person mellee fighter. Combat is similar to ns2 in many ways, perticularly in how you engage the enemy. Nosgoth had a fairly high skill ceiling, but I think NS2's skill ceiling is far higher.
At the time I was playing Nosgoth had ~3000 players according to steam charts, compared to NS2's ~400. That is more than 7 times the playerbase. According to their skill system, I was in the 55th percentile. I was literally middle skilled. So in theory, the majority of the players should be near me in skill. One would think matchmaking would work well for me.
Nosgoths matchmaking was incredibly poor and is what ultimatly made me quit that game. I, a middle skilled player, would have to wait 15 to 30 minutes for matchmaking to put me into a lobby. I would then have to wait another 15 to 30 minutes for it to be filled with all 8 players. You would think that all that waiting would allow the matchmaking to find players of similar skill. It did not. Games often ended in 30/10 which was a huge stomp.
The problem Nosgoth had was that its playerbase was too small and its skill ceiling too high. This is similar to ns2 except NS2 has a higher skill ceiling and a much smaller playerbase.
NS2's skill ceiling is too high. The average skill of players with more than 50 hours in hive is 1246. Any player with over 2000 hive skill is literally off the charts. In the following spoiler there is a graph and a table that show the distribution of skill in ns2. The conclusion is that NS2 has a ridiculously high skill ceiling.
In this table, I am showing veteran players. For the purposes of this table I have defined veterans as players with greater than 50 hours recorded in hive. These values are from players that played between January 1st 2016 to February 12, 2016.
This histogram shows the distribution of skill for players that played between January 1st 2016 to February 12, 2016.
Standard deviation is a measure of the variation in a distribution or set of data. Basically it is a measure of how far apart the values are. In terms of skill for NS2, standard deviation is a measure of how near skilled teams are. In my experience have seen the standard deviation of skill on a team range from 400 to 1000. Both are ridiculously high values for a standard deviation, but that is the nature of having a small player base.
For an example, what if matchmaking was done for only the middle 50% skill bracket (835 to 1582 hive skill)? In excel I created a simulation of 100 servers. On these servers I put 18 players with a random skill value between 835 and 1582. The resulting teams had an average standard deviation of ~200. That is not great. That is how near skilled we can reasonable expect teams to be. That is what we have to work with.
Continuing that example, the bottom 50% (less than 1124 hive skill) would be less well off. I ran the same simulation with the new parameter of between 0 and 1124 skill value. The resulting average standard deviation of ~320.
The same experiment with the top 50% resulted in an average standard deviation of ~1090. The top 50% would not be able to find anything resembling near skill games. The problem is that the highest levels of skill go incredibly high. Anything above 2000 hive skill is off the charts.
These simulations were simplistic but they highlight the difficulty matchmaking will have. Only the middle 50% would be able to find something resembling near skilled games. The highest level of skills would not be able to find near skilled games what so ever.
This is the nature of the skill ceiling we have. The playebase is just many too small to have enough players to fill out the far ends of the skill bracket. Forced matchmaking would make for a poor experience. The playerbase is too small and the skill ceiling is too high. NS2 can not support large scale matchmaking.
Given the fact that you could only reasonably find nearish skilled games in the middle 50% of skill for ns2, and how poor a similar game called Nosgoths matchmaking was with over 7 times the playerbase, it is unlikely NS2 will ever have good matchmaking. The skill ceiling is too high and the player base is too small.
NS2 today has been improved far beyond what it was at launch. Even if NS2 was released in the state it is in today, I doubt player counts would be all that much higher. Sure, they would be higher but I doubt they would be high enough to support large scale matchmaking. NS2 is just too niche of a game. It is, and always would of been like this.
That does not mean matchmaking is not possible. It just can't be done in the traditional sense of matchmaking like other games have used. Any attempt at matchmaking in NS2 is going to have to be very creative.
Except that matchmaking removes your ability to choose what server you join.. Which makes the game not worth playing in the first place.
Matchmaking would not necessarily remove your ability to choose a server. Don't narrow your concept of matchmaking like that. You should keep an open mind. There are many ways matchmaking could be implemented, many of which would keep the server browser.
One middle ground idea I have heard is a mix between matchmaking and a gather. People join into a queue of players waiting for matchmaking. While waiting they can play ns2 via the server browser. The matchmaking lets people join the queue until it finds 12 players with hive skill within a standard deviation of 200. Standard deviation in terms of hive skill is a measure of how near skilled teams are.
The matchmaking system would hang onto these players, say there were like 24 in the queue. If 12 of them fit within a standard deviation of 200. When that happens the players would get a notification that they are going to be pulled for matchmaking in 1 minute.
The standard deviation of 200 is variable. A smaller standard deviation is better as teams will be more near skill, but a higher standard deviation will mean more matches made. The 1 minute notification is also variable. Either value could be adjusted to better fit the needs of the matchmaking system.
There is also the problem of having players join servers, and be pulled from them mid game. Maybe make it so servers can prevent players from joining if they are waiting for matchmaking. Just an idea.
Even if this is not an acceptable version of matchmaking, there are other options.
<1000 = Baddies
1000 - 2000 = Normal folk
2000 - 2500 = Really good normal folk
2500 - 3000 = Suspiciously good folk
>3000 = Hackers
Can we write this into the main menu please?
But its so far from the truth? Really anybody over 2000 hive skill is off the charts.
What do you mean by "off the charts"? As in "a good player" or "a player with an inflated score"? (I suspect you mean the former, but understood the latter first.)
<1000 = Baddies
1000 - 2000 = Normal folk
2000 - 2500 = Really good normal folk
2500 - 3000 = Suspiciously good folk
>3000 = Hackers
Can we write this into the main menu please?
But its so far from the truth? Really anybody over 2000 hive skill is off the charts.
Eh, I don't agree with that scaling but 2000 hive skill is probably just slightly above average for a veteran. I suspect if you change your filter from 50 hours to 500 hours the skill level will rise significantly. If 500 has too few players, maybe 300 or something.
Regarding matchmaking, I just don't see it. Hive is useful as a rough gauge of a player's skill but it can't make teams. Balancing players of all skill levels into teams is a delicate act that is easily upset. This is even before accounting for gorges and commanders who remain a problem even with a magic assessing system that gives you perfect hive score value. I don't see a solution if we can't (and we can't) group players of similar skill together.
Off the charts is just me saying that 2000 is ridiculously high skilled, at least compared to the majority of players. There are very few people with over 2000 hive skill.
In this table, I am showing veteran players. For the purposes of this table I have defined veterans as players with greater than 50 hours recorded in hive. These values are from players that played between January 1st 2016 to February 12, 2016.
This histogram shows the distribution of skill for players that played between January 1st 2016 to February 12, 2016.
In the graph, I stopped it at 2000 because it made the graph look better. But the top 10% of skill starts at 2110 by the information in the table. The top 5% starts at 2472. Then the top players are at about 4000 hive skill.
My point is that ultimately, that farewelltowarms categorization of skill is so far from the truth it is silly.
Eh, I don't agree with that scaling but 2000 hive skill is probably just slightly above average for a veteran. I suspect if you change your filter from 50 hours to 500 hours the skill level will rise significantly. If 500 has too few players, maybe 300 or something.
Players with over 50 hours left me a sample size of 4090. Players with over 300 hours leaves me a sample size of 1325. Players with over 500 leaves me a sample size of 670. As a veteran myself, I only have 450 hours in hive compared to my 2200 in steam.
For players with over 300 hours recorded in hive, the average skill is 1743.
The distribution is:
Off the charts is just me saying that 2000 is ridiculously high skilled, at least compared to the majority of players. There are very few people with over 2000 hive skill.
In this table, I am showing veteran players. For the purposes of this table I have defined veterans as players with greater than 50 hours recorded in hive. These values are from players that played between January 1st 2016 to February 12, 2016.
This histogram shows the distribution of skill for players that played between January 1st 2016 to February 12, 2016.
In the graph, I stopped it at 2000 because it made the graph look better. But the top 10% of skill starts at 2110 by the information in the table. The top 5% starts at 2472. Then the top players are at about 4000 hive skill.
My point is that ultimately, that farewelltowarms categorization of skill is so far from the truth it is silly.
I think you might get a more representative histogram if you weight each player by playtime. This will give you a better picture of "if I enter a game today, what skills am I likely to see." Players with high skill are very few, but they also play much more, so you are more likely to encounter them when you play.
Off the charts is just me saying that 2000 is ridiculously high skilled, at least compared to the majority of players. There are very few people with over 2000 hive skill.
In this table, I am showing veteran players. For the purposes of this table I have defined veterans as players with greater than 50 hours recorded in hive. These values are from players that played between January 1st 2016 to February 12, 2016.
This histogram shows the distribution of skill for players that played between January 1st 2016 to February 12, 2016.
In the graph, I stopped it at 2000 because it made the graph look better. But the top 10% of skill starts at 2110 by the information in the table. The top 5% starts at 2472. Then the top players are at about 4000 hive skill.
My point is that ultimately, that farewelltowarms categorization of skill is so far from the truth it is silly.
It looks a lot more reasonable if you change the terms used.
Baddies -> deadweight
Normal folk -> not a total liability
Really good -> asset
Susp good -> carry
Hacker -> star player
Edit - Thanks for putting in all the work. @Nordic
Baddies -> deadweight
Normal folk -> not a total liability
Really good -> asset
Susp good -> carry
Hacker -> star player
That could be true, but it depends on the sever though.
For further reference the following skills seem to have the following average skill level on them.
Wooza's ~1200
TacticalGamer ~1400
MCG Lounge ~ ~1250
DungeonGaming ~1100
DungeonGaming Rookie only ~550
ThirstyOnos #18 ~1400
ThirstyOnos #22 ~1250
I got those values by approximating the average skill development from each servers wonitor page.
There is actually a lot of information to be looking through in this one. I am a little confused though. The values I get from that website differ from the values on wonitor for the same server.
I think you might get a more representative histogram if you weight each player by playtime. This will give you a better picture of "if I enter a game today, what skills am I likely to see." Players with high skill are very few, but they also play much more, so you are more likely to encounter them when you play.
The closest thing I have to that is this.
Some here might know of this proposal to fix the problem with some rookies starting at 0 skill and others at 1000.
It scales skill values as a function of playtime. This sort of weighs players by playtime. The result was this, but that is for all players.
You missed the key word: on average. If your individual skill raises your team's probability of winning from 50% to 55%, you will win more games in the long run even if you lose many games due to your team and not your lack of skill. Play hundreds of games and it will be correct, on average. Do not get bogged down by anecdotes and individual examples. I've lost count how many times I've explained this on these forums. No doubt people will still come arguing against this concept since yesterday they had an absolute terrible game where they went 50-1 but lost the skill system is broken please fix.
____
I wonder what life is like if you reject everything that goes against your intuition as being nonsense without doing any research. Is it easier to live in ignorance?
____
Do you blame everyone who comes at you with fact-backed arguments about echo chambers and groupthink? What do you think about vaccines and global warming?
Care to elaborate why you never even try to counter any arguments made by these so-called "self-validators", apart from shouting oppression?
This is the first time I have posted about this. I don't care how many times you've explained it. Take your shitty attitude and go away. I am repeatedly getting pissed off by know-it-alls, correct as they may be, having 0 inter-personal skills. Bitch and whine all the time.
Thank you for explaining to me how statistics work. The numbers do in fact even out by including all the parts of the game in the equation and the resulting difference you make on the outcome is your skill. I do not think being held responsible for other player's actions is a representation of player skill. This has nothing to do with the math.
The irony of a post like that and calling me the one with a "shitty attitude" is palpable. The last two posts you quoted were not directed at you, but at someone who has had multiple people explain very simple ideas to, but still continues with denial and trash-talking.
You don't like being held responsible for other player's actions? Good, because, in the long run, your skill is not affected by other people's actions. Glad we agree and you understand the statistics behind it.
You don't like being held responsible for other player's actions? Good, because, in the long run, your skill is not affected by other people's actions.
I think you mean in the looooooooooooooooooooooooooooooooooooooooooooooooooooooooooooonnnnngggggg run.
As in, to the point of almost not being useful at all for the average player.
Comments
Since you bring it up, "global warming" hasn't actually happened for the past 20 years. But don't let facts get in your way. Just keep telling yourself that those 20,000 "experts" who signed that document, who mostly aren't even scientists, who gain financially from "carbon trading" and sales of "eco-friendly" lightbulbs, and who fly around in tax-payer funded private jets to remote paradise islands for their "conferences," are peddling their mumbo-jumbo solely for the good of humanity.
Looks like the unpaid professional commentator NS2 groupthinkers support the real-world groupthinkers! Why am I not surprized that you would be a psuedo-science alarmist. I'm out.
What about seal clubbing?
Whats everyones stance on seal clubbing?
Dude, what?
I think I need a mod...
I like seal clubbing, if you know what I mean.
Totes know whatchu mean.
This shouldn't happen unless hive is timing out or something. (Or unless a different algorithm was being used that didn't use hive? There are so many things out there that do something different that it's hard to tell what is being used on any particular shuffle.)
After the first win, the scores should change. Everyone on the winning team goes up, and everyone on the losing team goes down. This makes the original teams no longer balanced according to hive, so the next shuffle should mix them up substantially. I can see this maybe happening though if there's one god-player on the server whose skill is substantially underestimated and much better than everyone else. It might take a while to push his still up enough that teams are correctly stacked against him.
The true skill system adjusts skill individually for each player while hive adjust the whole team by the same amount. TrueSkill adjusts based on both skill and certainty while hive adjusts based only on skill.
TacticalFreedom has been running an experiment where they have been collecting data for TrueSkill. http://www.tacticalfreedom.com/mmr
I picked these two profiles because one has the highest TrueSkill MMR and the other has the most playtime recorded in the system.
http://www.tacticalfreedom.com/stats/51377413
http://www.tacticalfreedom.com/stats/135140684
If you scroll to the bottom you can see how these two players hive skill have changed over time compared to their TrueSkill MMR. Hive has these massive spikes going both up and down. Hive is occilating as @Therius described. In comparison the TrueSkill MMR does not have these spikes. Another thing I notice is that the trends are the same, even though hive oscillates. I may just be looking at only two players, but still, it is interesting that hive and TrueSkill MMR both rise at similar times or remain steady around the same skill level.
Just food for thought.
The main problem is probably that it is a 1600 mean rather a 1600 median. A lot of games really are competitions between the top two-three players of each team and the rest are just fodder. Without teamwork, it is going to be hard to dislodge a good marine in a good position especially when they go in one by one without lifeforms to get slaughtered and waste eggs. A slight change in skill/players isn't going to reverse this because marines have an inherent advantage over skulks and fodder is still going to be fodder on different teams.
If that's true, then the distribution of skills will adjust to reflect that. The top tier players will achieve skill values so high that the skills of lower players are never sufficient to balance out the teams on their own. You might be able to build this assumption into the system by doing a more "max-like" combination of player skills, like a power mean https://en.wikipedia.org/wiki/Generalized_mean, but I would be hesitant to build this assumption into the skill system directly, since at least in principle, the current system could effectively learn that over time.
I don't think its a flaw in the hive skill measuring system. Its just that sometimes, with the current pool of players on the server, you are not going to get a good game because skill is imbalanced on the server itself. No matter how you split the teams, the difference will only be either laughably bad (most uneven teams) or tortuously drawn out (most even teams).
The skill system would be a lot more respected if we had the player base and matchmaking system to group players of similar skill together rather than the current unfortunate situation of having to try and balance players of all skills in a game.
Edit - I didn't get the maths part. Sorry.
Except that matchmaking removes your ability to choose what server you join.. Which makes the game not worth playing in the first place.
I'm a life peer of the Seal Cub Clubbing Club.
There is a long explanation of why that is true in this spoiler.
At the time I was playing Nosgoth had ~3000 players according to steam charts, compared to NS2's ~400. That is more than 7 times the playerbase. According to their skill system, I was in the 55th percentile. I was literally middle skilled. So in theory, the majority of the players should be near me in skill. One would think matchmaking would work well for me.
Nosgoths matchmaking was incredibly poor and is what ultimatly made me quit that game. I, a middle skilled player, would have to wait 15 to 30 minutes for matchmaking to put me into a lobby. I would then have to wait another 15 to 30 minutes for it to be filled with all 8 players. You would think that all that waiting would allow the matchmaking to find players of similar skill. It did not. Games often ended in 30/10 which was a huge stomp.
The problem Nosgoth had was that its playerbase was too small and its skill ceiling too high. This is similar to ns2 except NS2 has a higher skill ceiling and a much smaller playerbase.
NS2's skill ceiling is too high. The average skill of players with more than 50 hours in hive is 1246. Any player with over 2000 hive skill is literally off the charts. In the following spoiler there is a graph and a table that show the distribution of skill in ns2. The conclusion is that NS2 has a ridiculously high skill ceiling.
In this table, I am showing veteran players. For the purposes of this table I have defined veterans as players with greater than 50 hours recorded in hive. These values are from players that played between January 1st 2016 to February 12, 2016.
This histogram shows the distribution of skill for players that played between January 1st 2016 to February 12, 2016.
Standard deviation is a measure of the variation in a distribution or set of data. Basically it is a measure of how far apart the values are. In terms of skill for NS2, standard deviation is a measure of how near skilled teams are. In my experience have seen the standard deviation of skill on a team range from 400 to 1000. Both are ridiculously high values for a standard deviation, but that is the nature of having a small player base.
For an example, what if matchmaking was done for only the middle 50% skill bracket (835 to 1582 hive skill)? In excel I created a simulation of 100 servers. On these servers I put 18 players with a random skill value between 835 and 1582. The resulting teams had an average standard deviation of ~200. That is not great. That is how near skilled we can reasonable expect teams to be. That is what we have to work with.
Continuing that example, the bottom 50% (less than 1124 hive skill) would be less well off. I ran the same simulation with the new parameter of between 0 and 1124 skill value. The resulting average standard deviation of ~320.
The same experiment with the top 50% resulted in an average standard deviation of ~1090. The top 50% would not be able to find anything resembling near skill games. The problem is that the highest levels of skill go incredibly high. Anything above 2000 hive skill is off the charts.
These simulations were simplistic but they highlight the difficulty matchmaking will have. Only the middle 50% would be able to find something resembling near skilled games. The highest level of skills would not be able to find near skilled games what so ever.
This is the nature of the skill ceiling we have. The playebase is just many too small to have enough players to fill out the far ends of the skill bracket. Forced matchmaking would make for a poor experience. The playerbase is too small and the skill ceiling is too high. NS2 can not support large scale matchmaking.
NS2 today has been improved far beyond what it was at launch. Even if NS2 was released in the state it is in today, I doubt player counts would be all that much higher. Sure, they would be higher but I doubt they would be high enough to support large scale matchmaking. NS2 is just too niche of a game. It is, and always would of been like this.
That does not mean matchmaking is not possible. It just can't be done in the traditional sense of matchmaking like other games have used. Any attempt at matchmaking in NS2 is going to have to be very creative.
Matchmaking would not necessarily remove your ability to choose a server. Don't narrow your concept of matchmaking like that. You should keep an open mind. There are many ways matchmaking could be implemented, many of which would keep the server browser.
One middle ground idea I have heard is a mix between matchmaking and a gather. People join into a queue of players waiting for matchmaking. While waiting they can play ns2 via the server browser. The matchmaking lets people join the queue until it finds 12 players with hive skill within a standard deviation of 200. Standard deviation in terms of hive skill is a measure of how near skilled teams are.
The matchmaking system would hang onto these players, say there were like 24 in the queue. If 12 of them fit within a standard deviation of 200. When that happens the players would get a notification that they are going to be pulled for matchmaking in 1 minute.
The standard deviation of 200 is variable. A smaller standard deviation is better as teams will be more near skill, but a higher standard deviation will mean more matches made. The 1 minute notification is also variable. Either value could be adjusted to better fit the needs of the matchmaking system.
There is also the problem of having players join servers, and be pulled from them mid game. Maybe make it so servers can prevent players from joining if they are waiting for matchmaking. Just an idea.
Even if this is not an acceptable version of matchmaking, there are other options.
Can we write this into the main menu please?
What do you mean by "off the charts"? As in "a good player" or "a player with an inflated score"? (I suspect you mean the former, but understood the latter first.)
Eh, I don't agree with that scaling but 2000 hive skill is probably just slightly above average for a veteran. I suspect if you change your filter from 50 hours to 500 hours the skill level will rise significantly. If 500 has too few players, maybe 300 or something.
Regarding matchmaking, I just don't see it. Hive is useful as a rough gauge of a player's skill but it can't make teams. Balancing players of all skill levels into teams is a delicate act that is easily upset. This is even before accounting for gorges and commanders who remain a problem even with a magic assessing system that gives you perfect hive score value. I don't see a solution if we can't (and we can't) group players of similar skill together.
In this table, I am showing veteran players. For the purposes of this table I have defined veterans as players with greater than 50 hours recorded in hive. These values are from players that played between January 1st 2016 to February 12, 2016.
This histogram shows the distribution of skill for players that played between January 1st 2016 to February 12, 2016.
My point is that ultimately, that farewelltowarms categorization of skill is so far from the truth it is silly.
Players with over 50 hours left me a sample size of 4090. Players with over 300 hours leaves me a sample size of 1325. Players with over 500 leaves me a sample size of 670. As a veteran myself, I only have 450 hours in hive compared to my 2200 in steam.
For players with over 300 hours recorded in hive, the average skill is 1743.
The distribution is:
For players with over 500 hours recorded in hive, the average skill is 1965.
The distribution is:
And again, these values are from players that played between January 1st 2016 to February 12, 2016.
So for the small select few players who have over 500 hours recorded, ~2000 is average. That is my point though, there are not very many of them.
Which is kind of my point about matchmaking.
I think you might get a more representative histogram if you weight each player by playtime. This will give you a better picture of "if I enter a game today, what skills am I likely to see." Players with high skill are very few, but they also play much more, so you are more likely to encounter them when you play.
It looks a lot more reasonable if you change the terms used.
Baddies -> deadweight
Normal folk -> not a total liability
Really good -> asset
Susp good -> carry
Hacker -> star player
Edit - Thanks for putting in all the work. @Nordic
For further reference the following skills seem to have the following average skill level on them.
Wooza's ~1200
TacticalGamer ~1400
MCG Lounge ~ ~1250
DungeonGaming ~1100
DungeonGaming Rookie only ~550
ThirstyOnos #18 ~1400
ThirstyOnos #22 ~1250
I got those values by approximating the average skill development from each servers wonitor page.
Edit:
There is actually a better place for this information.
http://ns2servers.net/grafana/dashboard/db/natural-selection-2-server-hivescore-autogen
There is actually a lot of information to be looking through in this one. I am a little confused though. The values I get from that website differ from the values on wonitor for the same server.
EditX2: The closest thing I have to that is this.
Some here might know of this proposal to fix the problem with some rookies starting at 0 skill and others at 1000.
It scales skill values as a function of playtime. This sort of weighs players by playtime. The result was this, but that is for all players.
This is the first time I have posted about this. I don't care how many times you've explained it. Take your shitty attitude and go away. I am repeatedly getting pissed off by know-it-alls, correct as they may be, having 0 inter-personal skills. Bitch and whine all the time.
Thank you for explaining to me how statistics work. The numbers do in fact even out by including all the parts of the game in the equation and the resulting difference you make on the outcome is your skill. I do not think being held responsible for other player's actions is a representation of player skill. This has nothing to do with the math.
You don't like being held responsible for other player's actions? Good, because, in the long run, your skill is not affected by other people's actions. Glad we agree and you understand the statistics behind it.
As in, to the point of almost not being useful at all for the average player.