@dePARA I am fairly certain shine already has options to shuffle by kdr or spm. I think it is great if a server community wants to try something like this, like hbz used to. If that makes them happy, then I think that is good.
You should look into what @Tik is doing on the sudamerican server. He is using ns2+ data from wonitor to get per server skill metrics and doing interesting things with it. He has some very interesting graphs showing hive skill variance too.
Modrequest: Make a "Alternative shuffle mod" based on Observatory https://observatory.morrolan.ch/leaderboard
This Site is tracking all the data you need for an real skill based shuffle.
K/D, Accuracy and hive skill are the most important.
If only the hive server was open source, I'd have done this by now.
I know there's at least one mod out there where it sends a list of steam IDs out to a web server, and the server responds with a shuffle configuration while taking into account team-specific skill and K/D.
Modrequest: Make a "Alternative shuffle mod" based on Observatory https://observatory.morrolan.ch/leaderboard
This Site is tracking all the data you need for an real skill based shuffle.
K/D, Accuracy and hive skill are the most important.
If only the hive server was open source, I'd have done this by now.
I know there's at least one mod out there where it sends a list of steam IDs out to a web server, and the server responds with a shuffle configuration while taking into account team-specific skill and K/D.
I don't see why an open source hive server is necessary. There is observatory or even local wonitor data to work with. Shamess of the old tactical freedom server used local data to calculate a players mmr based on Microsofts patented trueskill. Trueskill is open source for non commercial use. There are plenty of options.
Also, in my testing hive was actually more accurate than trueskill at ranking players at least compared to typical skill metrics.
dePARAJoin Date: 2011-04-29Member: 96321Members, Squad Five Blue
The point is: you can use the observatory data to find the 4 carry player , 2 for each team. Then shuffle 50% of the players based on hive stats, the remaining ones by average KDR of the last 5 rounds.
That was the HBZ shuffle. (or something near like that)
The important point is: find the carrys and split them.
Ontop it used the ns2stats data to ban smurfs like mentioned above.
Based on stats, smurfs are easy to identify.
Modrequest: Make a "Alternative shuffle mod" based on Observatory https://observatory.morrolan.ch/leaderboard
This Site is tracking all the data you need for an real skill based shuffle.
K/D, Accuracy and hive skill are the most important.
If only the hive server was open source, I'd have done this by now.
I know there's at least one mod out there where it sends a list of steam IDs out to a web server, and the server responds with a shuffle configuration while taking into account team-specific skill and K/D.
I don't see why an open source hive server is necessary. There is observatory or even local wonitor data to work with. Shamess of the old tactical freedom server used local data to calculate a players mmr based on Microsofts patented trueskill. Trueskill is open source for non commercial use. There are plenty of options.
Also, in my testing hive was actually more accurate than trueskill at ranking players at least compared to typical skill metrics.
It doesn't have to be open source but it would make it much easier. Implementing stuff from scratch is a pain.
Modrequest: Make a "Alternative shuffle mod" based on Observatory https://observatory.morrolan.ch/leaderboard
This Site is tracking all the data you need for an real skill based shuffle.
K/D, Accuracy and hive skill are the most important.
If only the hive server was open source, I'd have done this by now.
I know there's at least one mod out there where it sends a list of steam IDs out to a web server, and the server responds with a shuffle configuration while taking into account team-specific skill and K/D.
I don't see why an open source hive server is necessary. There is observatory or even local wonitor data to work with. Shamess of the old tactical freedom server used local data to calculate a players mmr based on Microsofts patented trueskill. Trueskill is open source for non commercial use. There are plenty of options.
Also, in my testing hive was actually more accurate than trueskill at ranking players at least compared to typical skill metrics.
It doesn't have to be open source but it would make it much easier. Implementing stuff from scratch is a pain.
Is it really from scratch? As I understand it, it can be a lot easier to begin by adjusting someone elses code rather than beginning from scratch. Forgive my ignorance if I am wrong; I am not a programmer. There are multiple other mods in the past that have a custom shuffle,or even a custom mmr system you could look at.
Shuffle is shine and is open source already. Are you saying that you want to make an alternative skill system or a shuffle system?
Shuffle is shine and is open source already. Are you saying that you want to make an alternative skill system or a shuffle system?
Shine is not open source in the sense that I can modify it. It's open to read, but if I made a mod based on any shine code, that would be infringing copyright.
Shine is not open source in the sense that I can modify it. It's open to read, but if I made a mod based on any shine code, that would be infringing copyright.
dePARAJoin Date: 2011-04-29Member: 96321Members, Squad Five Blue
@Nintendows
2 options then: Make a independent mod or a shine plugin
There are different shuffle methods already. Tik mentioned in his mod "You are free to look, modify or reuse the code."
So, you discuss non relevant "problems".
@Nintendows
2 options then: Make a independent mod or a shine plugin
There are different shuffle methods already. Tik mentioned in his mod "You are free to look, modify or reuse the code."
So, you discuss non relevant "problems".
You know, it's weird. Coming from Siege onto Vanilla. The last time I played Vanilla was 2014. Now the Proving Ground server is hosting vanilla competitive. Everything feels like HLStatsX. You know HLStatsX, right? The thing about HLSTatsX is that it was optional. 10 years ago when I placed HLStatsX on my TF2 server, people weren't sure how to react. Back in the day, I've certainly gained a negative attraction against ranked gameplay. People liked the idea of "playing for fun", "not too seriously". Well it seems today in NS2 there's no seperation. Badges are enforced, everything a player does recorded. I really don't know how to react other than my opinion of Siege gameplay being superior .
Shuffling by spm might work, but kdr is not accurate at all. You can camp all day in useless spots without doing any dmg to your enemy tech, have a really good kdr, and lose because you cared too much about it.
Shuffling by spm might work, but kdr is not accurate at all. You can camp all day in useless spots without doing any dmg to your enemy tech, have a really good kdr, and lose because you cared too much about it.
Going by SPM wouldn't be any different from KDR. If anything it might even be worse.
I've had games where I accomplish absolutely nothing on marine to contribute to the win, yet my score will be super high.
Wouldn't the best solution be to use all three? Win rate, KDR, and SPM?
Score per minute is a good measure for how productive you are, which is related to your overall skill. Server administrators can set up shuffle to use score per minute already.
Kdr is a good measure for how good at getting kills you are, which is related to your overall skill. Server administrators can set up shuffle to use kdr already
You could do all sorts of fancy math trying to use kdr and spm to balance with, and it would work about as effective as hive skill does. A good example of this done well would be ns2stats elo formula.
Edit: Would anyone like to see a comparison between how well hive skill ranks players compared to kdr, spm, and other metrics? By rank I mean that the highest player would be #1, the second highest would be #2 and so on. It is a simple method, but you can see how well hive compares to metrics like KDR or SPM in ranking players from best to worst.
Edit: Would anyone like to see a comparison between how well hive skill ranks players compared to kdr, spm, and other metrics? By rank I mean that the highest player would be #1, the second highest would be #2 and so on. It is a simple method, but you can see how well hive compares to metrics like KDR or SPM in ranking players from best to worst.
Yes! In particular I would be interested if hive skill level directly correlates with kdr and spm, or if there's different "player profiles", i.e. players who have a high skill level because they shoot good, and players with a high skill level because they hit the objective good.
Comments
You should look into what @Tik is doing on the sudamerican server. He is using ns2+ data from wonitor to get per server skill metrics and doing interesting things with it. He has some very interesting graphs showing hive skill variance too.
If only the hive server was open source, I'd have done this by now.
I know there's at least one mod out there where it sends a list of steam IDs out to a web server, and the server responds with a shuffle configuration while taking into account team-specific skill and K/D.
I don't see why an open source hive server is necessary. There is observatory or even local wonitor data to work with. Shamess of the old tactical freedom server used local data to calculate a players mmr based on Microsofts patented trueskill. Trueskill is open source for non commercial use. There are plenty of options.
Also, in my testing hive was actually more accurate than trueskill at ranking players at least compared to typical skill metrics.
That was the HBZ shuffle. (or something near like that)
The important point is: find the carrys and split them.
Ontop it used the ns2stats data to ban smurfs like mentioned above.
Based on stats, smurfs are easy to identify.
Everything needed is in the game, just use it.
It doesn't have to be open source but it would make it much easier. Implementing stuff from scratch is a pain.
Is it really from scratch? As I understand it, it can be a lot easier to begin by adjusting someone elses code rather than beginning from scratch. Forgive my ignorance if I am wrong; I am not a programmer. There are multiple other mods in the past that have a custom shuffle,or even a custom mmr system you could look at.
Shuffle is shine and is open source already. Are you saying that you want to make an alternative skill system or a shuffle system?
Shine is not open source in the sense that I can modify it. It's open to read, but if I made a mod based on any shine code, that would be infringing copyright.
Thats wrong.
https://github.com/Person8880/Shine/wiki#for-developers
The goal of person8880 (shine author) was and is: easy to extend and maintain
I found the old shine shuffle based on ns2stats with a fallback method.
https://github.com/Person8880/Shine/wiki/Vote-Random/ad94e5dc55929345dd2c5d31b27b3a1325189da8
The HBZ shuffle based on this with some small modifications.
Replace the NS2stats part with Observatory data and you got it.
@Nordic
This is Tiks shuffle system but i have not looked in the code yet:
https://steamcommunity.com/sharedfiles/filedetails/?id=1410401007
That link doesn't explain that I'm wrong. Here's where I once asked to use and modify a shine file and was denied: https://github.com/Person8880/Shine/issues/639
2 options then: Make a independent mod or a shine plugin
There are different shuffle methods already. Tik mentioned in his mod "You are free to look, modify or reuse the code."
So, you discuss non relevant "problems".
lol
Edit: Here's the Proving Ground server for NS2 for reference https://www.gametracker.com/server_info/209.222.16.170:27045/
Edit: Yes I know hiveskill can be turned off for ns2 server. However, who is going to turn it off ? no one.
Going by SPM wouldn't be any different from KDR. If anything it might even be worse.
I've had games where I accomplish absolutely nothing on marine to contribute to the win, yet my score will be super high.
Wouldn't the best solution be to use all three? Win rate, KDR, and SPM?
Kdr is a good measure for how good at getting kills you are, which is related to your overall skill. Server administrators can set up shuffle to use kdr already
You could do all sorts of fancy math trying to use kdr and spm to balance with, and it would work about as effective as hive skill does. A good example of this done well would be ns2stats elo formula.
Edit: Would anyone like to see a comparison between how well hive skill ranks players compared to kdr, spm, and other metrics? By rank I mean that the highest player would be #1, the second highest would be #2 and so on. It is a simple method, but you can see how well hive compares to metrics like KDR or SPM in ranking players from best to worst.
Yes! In particular I would be interested if hive skill level directly correlates with kdr and spm, or if there's different "player profiles", i.e. players who have a high skill level because they shoot good, and players with a high skill level because they hit the objective good.