Different scores for rookies?
dePARA
Join Date: 2011-04-29 Member: 96321Members, Squad Five Blue
Join Date: 2011-04-29 Member: 96321Members, Squad Five Blue
Whats the point of having rookies with a different starthive score at there 1st play?
This is a huge problem at the moment.
Some rookies start with 1000 (dont know why) others with 0.
So what happen when hive-based shuffle kicks in?
You can have a team FULL of rookies against a team with three 2000skill and some 0 point rookies.
But the teamskill seems "balanced" around 1000.
These games end in a slaughter.
Saw this countless times these days and these games are frustrating like hell.
Im sure for both, rookies and vets.
From 10 rounds shuffled with hive-skill, 9 are like above.
This is a huge problem at the moment.
Some rookies start with 1000 (dont know why) others with 0.
So what happen when hive-based shuffle kicks in?
You can have a team FULL of rookies against a team with three 2000skill and some 0 point rookies.
But the teamskill seems "balanced" around 1000.
These games end in a slaughter.
Saw this countless times these days and these games are frustrating like hell.
Im sure for both, rookies and vets.
From 10 rounds shuffled with hive-skill, 9 are like above.
Comments
The split in skill is quite evident in the skill distribution.
If you are interested in other statistics from hive, check out this thread.
Its just another example like many many before how NS2 failed to attract the game to the new players cause mistakes or forgotten stuff that has to be changed before a sale or free weekend.
Right now this happen:
"Shuffle, every1 press f1 to shuffle".
"Ah, hive score looks good, lets start"
Slaughter
Repeat the above multiple times and then:
- unstall, shitgame. Only boring one sided rounds.
This is why I was saying the shuffle needs to split all rookies evenly, not just players at a certain number and then divide the rest of the players based on hive score. You still see way too many games with most of the rookies on one team which leads to crappy games. I've mentioned it before and it wasn't changed.
We should fix this. I think a fair way would be to subtract 800 from all scores, or from all scores greater than 500. I'm not sure when we will be able to make changes to hive again, but when we can I will propose this.
I really didn't suspect that players coming back to the game after a year would be a common thing.
And its prophet the sacred ELO.
What dePara brings up is a valid point and i agree, this must be fixed.
Right now its unplayable for me. I have lost so many games, cant count them.
Its just too frustrating right now.
After coming back im done with the game after 4 days again.
I will give a another try later maybe.
This need to be fixed ASAP.
Every server is forcing or voting for shuffle and nearly every game end in an slaughter right after a sale.
I really cant understand what you are waiting for.
Some dont understand that players with high Accuracy/KDR have a HUGE impact on the game.
They can kill lieforms, controlling the early game vs skulk and harassing harvesters/RTS.
Thats why i always said:
Dont show KDR on the scoreboard but everyone can check his own stats like in TF2 but shuffle the teams based on a mix between kdr/accuracy/elo.
I was doing this back in the days on the HBZ server.
The 4 with the highest elo (carry player) where split to each team. The other players shuffled based on there average KDR of the last 5 rounds.
A assist gave you a half killpoint.
Worked pretty good most of the time.
Obviously having returning players at a skill level of 1000, and new players joining at 0 does make shuffle useless. Besides small, but major, problems like described in this thread the hive systems should be much more accurate at predicting wins. It accounts for everything including K/D.
Give it a few weeks and those scores will converge on each other.
A few good points have been argued; marines and aliens should have a separate rating, the 0-1000 rookie mess should be fixed and something else than simple average should be used for assessing team skill. These are all problems that can be fixed without scrapping the entire system, and would ALL still be present in a KDR-based system.
The hive ranking surely has ways to improve. There are exploits and such. But it IS by far the best system to predict outcomes. I have said my peace on other occasions about this, so I won't go in to detail on that.
The following, is the reason why you will NEVER have it as accurate as people want it, for NS2 specifically - even if it worked perfectly with no exploits.
It's because the skill levels are not separated. This is what happens in NS2:
Herakles goes to play a collection of gathers. He wins 10 games in a row and subsequently sinks the other gatherers ELO as his own inflates. Assume the ELO is correctly being distributed among the gatherers at this point.
Then the gatherers with their deflated ELO go to play against pubbers, and start sinking their ELO's as they re-inflate their own.
Now, with their re-inflated ELO's, they go play another round of gathers with Herakles - and pour more ELO points in Herakles.
This process repeats, as the gather scene perpetually hoardes all the ELO points.
Every time you go from the gather environment to the pub and vice versa, you compromise the closed system in which the ELO ratings are accurate. This applies between public servers, and community servers as well.
While you only have pure regulars on Hellrunners server i.e. - the rating will be accurate. But the moment that some rookie joins, or someone from Wooza, or Tactical Gamers, the closed system is compromised.
If you segregate the playerbase, based on skill, then you won't have the top players, the gatherers, picking from the casuals ELO jars. There would still be inaccuracies, and the groups would be compromised here and there. But the process is much longer, as you won't have the highest skilled group picking directly from the lowest skilled groups ELO and hoardering it. The highest skill group can only steal ELO from the group directly below them if you segregate properly.
Obviously, segregation is never going to happen in NS2, sadly. But that is why, you will never have an accurate ranking system to the extent that people expect.
EDIT: The best thing I can think of. Is instead of using HIVE, implement your own private skill ranking, like ns2stats, on your servers. Like HBZ had. EVEN if it uses the exact same algorithm as HIVE - which I highly recommend, as I said, it is the best way to predict outcomes.
I don't think this global system will ever work for NS2 without proper separation though.
So you try to hold new players by simplifying things with endless discussions and on the other hand you say this after a sale.
In a few weeks all the players who bought the game are gone again cause one bullshit round after another.
Im sure its very simple to fix this issue compared to changing gamemechanics.
Just wow.
Thats the way to hold players.
I... I never thought of my score as being a valuable commodity that is being leeched from me by the %1
Class warfare!
I happen to think that these things DO have a massive impact - which is why I generally see the players with these accuracies have high Hive scores.
@dePARA : welcome to my world. Bad game after bad game?? You'll get used to it.
But it's a sign. If more regular players play more on public servers it means, maybe some other left...
If you want "balanced" games, then play with your team. The rest is practice. Dont have a team? Too bad. Never going to have a "balanced" match in this asymetrical multiplayer. A fair match? The rookie servers provide this for the rookies. The rest of us will have to live with it, untill you join or create a team, or queueing as multiple members becomes a thing.
Nope, if you know how to play NS it's not needed. You can meet many equal to "comp" players in terms of skill (the so called) on public servers. Teams are mixed all the time. In fact when some new team try a public rampage, they get spanked, and that's it.
So its Herakles' fault! I KNEW it!
In all seriousness, Santa is correct about the 'closed system' aspect of skill rating.
It's a knock-on effect of community-run servers.
It's supposed to be me ! Damn it ! You quit now!
Yeh, been asking medi to get a mod to calculate skill based only on rounds play at his the no rookies captains server. Would be fantastic
None of us think that the hive skill system is perfect. Never have.
Something like in Battlefield4.
This + and top10 for each would give progression and you could use this data for real rookie servers for example.
If an smurf would running around with his 2nd or 3rd account you could block him on these rookie servers based on these stats.
So if someone constantly have a accuracy around 30% and slaughter all the rookies in every round, he cant be a rookie.
Something like that would help that new players to learn the game in an stomp free environment.
And this data could be used for an better skill system.
On the KDR issue, I agree, we could get a much better estimate of a player's skill much more quickly if we looked at kdr. The problem is incentives. If there's a number, people will try to make that number go up. The nice thing about the current system is that the only incentives are aligned with the game. The problem with KDR is that it is easy to exploit, and exploiting it makes the experience worse for everyone else. (And in fact, when kdr was the primary metric for the hive skill system, people spawn camped until it killed the server every round, which was my motivation for writing this up in the first place. The goal wasn't a "perfect" skill system, or even an accurate one. The goal was an "abuse proof" skill system, one that by its design would not distort the game. )
For the statisticians out there, we are making a bias-variance tradeoff. We're accepting much higher variance in exchange for 0 bias, because bias causes bad behavior.
If a server disables the "Force Even Teams" vote and replaces it with a "Shuffle Based on KDR", how does that round get factored into Hive skill ranking?
Would that skew the results at all considering that the KDR shuffle is based on a much smaller quantity of data and puts the players into a round that is balanced a different way than the system was designed? Or am i just being paranoid?
Personally, I've had many more games that "feel" more balanced with the official "Force Even Teams" vote rather than any custom shuffle/balance vote.