Hiveskill convergence rate | Hive 2 better?

2»

Comments

  • NordicNordic Long term camping in Kodiak Join Date: 2012-05-13 Member: 151995Members, NS2 Playtester, NS2 Map Tester, Reinforced - Supporter, Reinforced - Silver, Reinforced - Shadow
    moultano wrote: »
    I'm thinking the most fair method is to compute the average marine skill and average alien skill for the server, then try to match each team against those averages. That prevents the weird behavior when everyone is better at one team than the other.

    Making hive learn separate marine and alien skill values won't be that hard. If you can think of a way to make shuffle work in an acceptable manner with separate marine and alien skill values I think it would be worth the effort.
  • amoralamoral Join Date: 2013-01-03 Member: 177250Members
    greenhorse wrote: »
    MoFo1 wrote: »
    My marine score should be 200-500 at the highest.. While my alien score should be closer to 1500-2000..

    Mofo, how would you shuffle a group of 20 players where each player has this kind of skill break down?

    I think what you could do is randomize top 30 percent then balance around them. So top 5 or 6 players get a random team assigned, then the rest of the server gets team balanced by alien/marine hive score to balance the teams.
  • amoralamoral Join Date: 2013-01-03 Member: 177250Members
    Someone do the math, you've got 4 players 2500, 1500, alien marine or marine alien, one with 2k both and the rest between 800 to 1500, within two points. I think you can still get balanced teams by forcing the top 5 or 6 to teams strictly randomly, and using the rest to fill in the gap so to speak.

    And the top guys will be truly random, and the rest will be randomized by that forcing
  • NordicNordic Long term camping in Kodiak Join Date: 2012-05-13 Member: 151995Members, NS2 Playtester, NS2 Map Tester, Reinforced - Supporter, Reinforced - Silver, Reinforced - Shadow
    edited October 2016
    amoral wrote: »
    greenhorse wrote: »
    MoFo1 wrote: »
    My marine score should be 200-500 at the highest.. While my alien score should be closer to 1500-2000..

    Mofo, how would you shuffle a group of 20 players where each player has this kind of skill break down?

    I think what you could do is randomize top 30 percent then balance around them. So top 5 or 6 players get a random team assigned, then the rest of the server gets team balanced by alien/marine hive score to balance the teams.

    What about the bottom who would also be forced to play the team they are better at? If anything I am more concerned about them than the top.

    That is because the bottom is likely going to be forced to play the team they are better at, compared to the top which would be forced to play the team they are worse at.

    I would rather give the bottom the chance to play the team they are worse at to learn and get better, but they won't get that with separate marine and alien skill values.

    Beyond that I still don't think any one player should have more choice than another player. I would not want the top or bottom 30% randomized and the rest to fall in line. No player group should be forced when others are not. It should be all or nothing.
  • MoFo1MoFo1 United States Join Date: 2014-07-25 Member: 197612Members
    greenhorse wrote: »
    MoFo1 wrote: »
    My marine score should be 200-500 at the highest.. While my alien score should be closer to 1500-2000..

    Mofo, how would you shuffle a group of 20 players where each player has this kind of skill break down?

    You mean a group of 20 people that are all 200-500 skill Marines and 1500+ skill Aliens?

    That would be impossible as no matter who you put on Marine there wouldn't be anyone to counter the average Alien players... And sometimes it is just impossible to balance teams... For instance if you have one 3500+ score 4.0+ kdr player in a game full of 1000-1500 score players.

    Now say you have 10 people who are 200-500 skill Marines/1500-2000 skill Aliens, and 10 people who are 200-500 skill Aliens/1500-2000 skill Marines...

    Ideally it would shuffle to put everyone on the team they are better at... So you'd have two teams of 1500-2000 players.

    That's where choosing your team preference comes in. Say 10 of those 20 players have no preference, while 6 prefer Alien and 4 prefer Marine... Then shuffle could (theoretically) place everyone who has a preference on their chosen team, and use the 10 who have no preference to balance it out. Also most people will tend to like the side they're better at, so it would likely be a higher skilled game.

    Certainly seems a lot better than one lump score for two completely different teams that play nothing even remotely alike, then having them all "randomly" shuffled in hopes of producing a balanced game...


    @Nordic That's a minor concern really...

    For starters if it turned out to be a problem the shuffle algorithm could give more weight to lower skill players team preference and balance around that... Besides they're likely to want to play the side they're better at anyway... The main ones that would be forced to play on the team they don't want and are worse at are the high skill god-like players who ruin game balance and empty servers... And truthfully they should be playing the side they're worse at anyway, both to help server balance AND to practice for their comp/gather games...

    And honestly unless they're ***holes they shouldn't mind being forced onto the team they're worse at if everyone on the server is way worse than they are.. They'd always get to play their preferred team in gathers and servers with higher skill players. Hell something like this could even encourage higher skill players friend each other and congregate in one server, instead of having 1-3 of them ruining balance across 6+ servers in the browser.

    At any rate shuffle will never really work well as long as we have one lump skill score for two vastly different gameplay styles.
  • amoralamoral Join Date: 2013-01-03 Member: 177250Members
    Nordic wrote: »
    amoral wrote: »
    greenhorse wrote: »
    MoFo1 wrote: »
    My marine score should be 200-500 at the highest.. While my alien score should be closer to 1500-2000..

    Mofo, how would you shuffle a group of 20 players where each player has this kind of skill break down?

    I think what you could do is randomize top 30 percent then balance around them. So top 5 or 6 players get a random team assigned, then the rest of the server gets team balanced by alien/marine hive score to balance the teams.

    What about the bottom who would also be forced to play the team they are better at? If anything I am more concerned about them than the top.

    That is because the bottom is likely going to be forced to play the team they are better at, compared to the top which would be forced to play the team they are worse at.

    I would rather give the bottom the chance to play the team they are worse at to learn and get better, but they won't get that with separate marine and alien skill values.

    Beyond that I still don't think any one player should have more choice than another player. I would not want the top or bottom 30% randomized and the rest to fall in line. No player group should be forced when others are not. It should be all or nothing.

    depends on how the "randomize the top" plays out. you're not trying to maximize the hive totals, you're just trying to get as close to parity with the right amount of players.

    so sometimes you pick the shittier of the two, because you need to just throw a body onto the higher team to bring it down.
  • CmdrKeenCmdrKeen Join Date: 2013-05-21 Member: 185321Members, Reinforced - Shadow
    edited October 2016
    If we can't find a good solution for m/a seperated skill for now, how about seperate commander skill ranks? I'd love me some legenda.. I mean rookie comm badge.

    I am pretty sure my comm elo would be much different than my field elo (and also my marine comm elo much different than my khamm elo).

    Edit: I realized with comm skill ranks a shuffle would only be accurate once it is determined who are the comms (which also happens to start the game). But maybe the potential skill avg imbalance would be so small that we could just ignore it?; since only two players (the comms) are "misshuffled"
  • NordicNordic Long term camping in Kodiak Join Date: 2012-05-13 Member: 151995Members, NS2 Playtester, NS2 Map Tester, Reinforced - Supporter, Reinforced - Silver, Reinforced - Shadow
    CmdrKeen wrote: »
    If we can't find a good solution for m/a seperated skill for now, how about seperate commander skill ranks? I'd love me some legenda.. I mean rookie comm badge.

    I am pretty sure my comm elo would be much different than my field elo (and also my marine comm elo much different than my khamm elo).

    Edit: I realized with comm skill ranks a shuffle would only be accurate once it is determined who are the comms (which also happens to start the game). But maybe the potential skill avg imbalance would be so small that we could just ignore it?; since only two players (the comms) are "misshuffled"

    As I understand it, that is a more difficult problem than M/A seperate skill values. In order to record and or use a commander skill value there would need to be gameplay changes. As the game is now there is no way we can know who is going to command in order to balance around. That sounds like an easy fix. Get commanders then balance the game after that. As the game is now, the commander can change multiple times in a single game on both teams. In order to not only use, but to record commander skill the game would need to lock commanders into the chair the entire game. It may even be necessary to not allow commanders to change during a round. I am not sure I am prepared for that kind of game play change, let alone the community.

    TLDR: Commander skill values are not easily recorded or functional to use.
  • SantaClawsSantaClaws Denmark Join Date: 2012-07-31 Member: 154491Members, Reinforced - Shadow
    You don't need gameplay changes, you certainly don't need to lock commanders in, you just need to record time spent in chair. How technically difficult that is, you tell me.

    IF you have time spent as commanders recorded in each game, you could reward players commander rank proportional to the time that they spent in the chair in relation to the round duration - and that same reward will of course be what would otherwise go to his player rank.

    So if a player only jumps in for a quick second to mist himself or what ever, he'll gain a microscopically insignificant bonus in his comm rank. If two players, for what ever reason, spend equal amount of time in the chair, they gain or lose the same bonus in comm rank.

    Conversely, commanders that spend time in the field, should gain player rank proportional to the time they spent on the field, in relation to the round duration.
  • moultanomoultano Creator of ns_shiva. Join Date: 2002-12-14 Member: 10806Members, NS1 Playtester, Contributor, Constellation, NS2 Playtester, Squad Five Blue, Reinforced - Shadow, WC 2013 - Gold, NS2 Community Developer, Pistachionauts
    SantaClaws wrote: »
    You don't need gameplay changes, you certainly don't need to lock commanders in, you just need to record time spent in chair. How technically difficult that is, you tell me.

    IF you have time spent as commanders recorded in each game, you could reward players commander rank proportional to the time that they spent in the chair in relation to the round duration - and that same reward will of course be what would otherwise go to his player rank.

    So if a player only jumps in for a quick second to mist himself or what ever, he'll gain a microscopically insignificant bonus in his comm rank. If two players, for what ever reason, spend equal amount of time in the chair, they gain or lose the same bonus in comm rank.

    Conversely, commanders that spend time in the field, should gain player rank proportional to the time they spent on the field, in relation to the round duration.

    You can track it, but if you aren't going to use it to balance teams in some way, what use is it?
  • NordicNordic Long term camping in Kodiak Join Date: 2012-05-13 Member: 151995Members, NS2 Playtester, NS2 Map Tester, Reinforced - Supporter, Reinforced - Silver, Reinforced - Shadow
    SantaClaws wrote: »
    You don't need gameplay changes, you certainly don't need to lock commanders in, you just need to record time spent in chair. How technically difficult that is, you tell me.

    IF you have time spent as commanders recorded in each game, you could reward players commander rank proportional to the time that they spent in the chair in relation to the round duration - and that same reward will of course be what would otherwise go to his player rank.

    So if a player only jumps in for a quick second to mist himself or what ever, he'll gain a microscopically insignificant bonus in his comm rank. If two players, for what ever reason, spend equal amount of time in the chair, they gain or lose the same bonus in comm rank.

    Conversely, commanders that spend time in the field, should gain player rank proportional to the time they spent on the field, in relation to the round duration.

    Sorry I was not clear. I did not say it could not be recorded. Gameplay changes are needed to effectively use a commander hive skill value to balance teams.
  • SantaClawsSantaClaws Denmark Join Date: 2012-07-31 Member: 154491Members, Reinforced - Shadow
    moultano wrote: »
    SantaClaws wrote: »
    You don't need gameplay changes, you certainly don't need to lock commanders in, you just need to record time spent in chair. How technically difficult that is, you tell me.

    IF you have time spent as commanders recorded in each game, you could reward players commander rank proportional to the time that they spent in the chair in relation to the round duration - and that same reward will of course be what would otherwise go to his player rank.

    So if a player only jumps in for a quick second to mist himself or what ever, he'll gain a microscopically insignificant bonus in his comm rank. If two players, for what ever reason, spend equal amount of time in the chair, they gain or lose the same bonus in comm rank.

    Conversely, commanders that spend time in the field, should gain player rank proportional to the time they spent on the field, in relation to the round duration.

    You can track it, but if you aren't going to use it to balance teams in some way, what use is it?
    I'd use it to balance the teams

    So I guess what you are referring to, is that players could abuse this, by swapping commanders at the start to boost the avg. elo. In my opinion that's not a big deal, if people really blatantly abuses this, in worse case admins can deal with it or you can vote kick.

    While I won't totally dismiss that concern (perhaps you have others), I think this solution does solve a much more pressing issue, than what it might cause. Namely the fact that veteran players (your lovers of the god damn game) with high elo, cannot command without screwing over their entire team as it is right now.
  • CmdrKeenCmdrKeen Join Date: 2013-05-21 Member: 185321Members, Reinforced - Shadow
    edited November 2016
    moultano wrote: »
    You can track it, but if you aren't going to use it to balance teams in some way, what use is it?

    You will (as it looks right now) not be able to use both commander skill values for balancing, but I believe you will be able to use one value regularly. From my experience, there is often the case that one team has a comm while the other team does not, and a shuffle is invoked.

    On some servers (Thirsty onos comes to mind), shuffles are done regularily shortly after a round has started.

  • xtalxtal aka X-rayCat Join Date: 2009-06-28 Member: 67961Members, Constellation, Reinforced - Supporter
    Someone wrote hive 2.0 will be on on b310, so is new hive working already or we need to wait more ?
  • antouantou France Join Date: 2016-07-24 Member: 220615Members
    xtal wrote: »
    Someone wrote hive 2.0 will be on on b310, so is new hive working already or we need to wait more ?

    It has been added yes, but there is no frontend for the moment, so you can't see the data.
  • barniebarnie Join Date: 2016-07-26 Member: 220695Members
    How is the current state of shine integration?
    Does shuffle or plugins that show average team skill use the old or the new data?
    Or is "force even" the only thing that is currently using Hive v2?
  • NordicNordic Long term camping in Kodiak Join Date: 2012-05-13 Member: 151995Members, NS2 Playtester, NS2 Map Tester, Reinforced - Supporter, Reinforced - Silver, Reinforced - Shadow
    edited November 2016
    barnie wrote: »
    How is the current state of shine integration?
    Does shuffle or plugins that show average team skill use the old or the new data?
    Or is "force even" the only thing that is currently using Hive v2?

    @GhoulofGSG9 has stated that Shine fully uses Hive 2.0.
    I am positive about NS2+ which is maintained by @remi
Sign In or Register to comment.