2000 vs 2000 skill games
xtal
aka X-rayCat Join Date: 2009-06-28 Member: 67961Members, Constellation, Reinforced - Supporter
aka X-rayCat Join Date: 2009-06-28 Member: 67961Members, Constellation, Reinforced - Supporter
Comments
Also: I think it's always been the case that alien wins happen in lower skill settings, gradiating to marine wins as skill increases (certainly in ENSL). It's just that marines who shoot 30-40% consistently are pretty dang impossible to counter, which honestly isn't a big problem IMO. It's skill.
So all that to say: What manner of suggestion do you have to fix this?
Reduce Rifle damage by 50% and Shotgun damage by 70% for all servers with skill level 2000+. Should do the trick for aliens.
He ain't wrong...
It's not only the Aim %. It is the number of marines too. More field units means more LOS/exit covered for one room. Also more marine coming to rescue that RT, and more often (fast respawn).
If any player with less than 50 hours recorded in hive is accused of being a smurf by three unique players then he will be forced to use a steam controller the same way as mentioned above. To do this, ns2 will be using an advanced regex code that will parse register if the player's name and the word smurf are used in the same chat message.
Valve will be paying UWE to add this feature. There is no need to complain. This will fund development.
ENSL: Sponsored by Valve Electronics.
Thats true.
Only on wooza this doesnt matter.
Looks like they need more than 21 marines there to win more than 50% of the games.
Maybe 60 slot can help them, 5 marines in each room are not enough.
That is one of the things that would probably be solved with 2 hive skills, one for marines and one for aliens.
With the present skill system maybe you need players with average 2500 skill in aliens to beat a 2000 skill marine team (just saying random numbers to explain my point). With 2 sets of skills you solve this. A team of 2000 alien skill should be balanced to a 2000 marine skill team.
The other point is that maybe some people will get to 4000 skill as marine, but nobody will get there as an alien (because of the reasons that you exposed before)
One time there was a 100 slot server.
Ask Hugh, he is the 'PR Guy' or in colloquial terms, guy with no better answer than ' you are correct '
Perhaps. However, 95% winrate does not seem representative of how favored marines actually are in high level play. I don't know where he got the 95% figure from, but I'm assuming it's correct for the sake of the argument.
What I would guess is going on here. Is that a large majority of 2000+ elo players, have a positive bias for the marine side. They may play that faction more often and as a result be better at this faction.
I think the solution is to separate alien and marine elo.
Worth mentioning is also, that aliens simply require a lot more team work and communication than marines do. So you wouldn't really expect a pub team to win against a marine team at this level. Non the less, I think a separation of the elo between factions would help things for the above mentioned reason.
The thing is they absolutely :
-Aim better (or kill more)
-Know where to go to cover the key locations
-React faster when a key location is cleared by alien
So the frontline shifts from the middle of the map to the alien naturals or worse, at hive doors. Basically in this conditions, the aliens have no word to say in early game. It's usually a come back or death in box. GG is called before it happens.
In an 6v6 envierment its easier to play marines effectivly than aliens, the reason is being that aliens need to coordinate their engagement much more than the marines do.
The point is, whether it is true for you or not, some people, for what ever reason, favor one faction over the other.
The factions are deliberately asymmetrical, so it only makes sense that your hive ranking is asymmetrical as well. Even if you play both factions the exact same amount, does not mean you'd be equally skilled in both factions. Either way, I think separation of the elo would help a great deal.
PS: I sincerely hope for an accurate system. We'll see.
Besides - People seem to know alot more about how to play together as marines than playing together as aliens.
pop corn... check
Beer... check
Cookies... check
Peanuts anyone?
The marines will shoot better, the skulks will be sneakier.
In practice tho, marines might be easier to learn for rookies.
Your problem seems to me more like a mixture of high skill and server size.
See, endgame marines can very quickly kill a hive if they coordinate a strike
which becomes increasingly common as skill increase and increasingly potent as player slots increase.
But on the other hand aliens can rush a cc faster earlygame as well as have many onos endgame so its hard to tell what to balance
better hitreg OR better aiming players both work in favor of the marines, this is a known factor that has haunted NS2/NS balancing for ever
You gain points for winning; you gain more points for winning when the opposing team has a higher skill average (which is to say that you gain more for winning when the opposing team is predicted to win). You lose points for losing; you lose more points for losing when the opposing team has a lower skill (which is to say that you lose more for losing when the opposing team is predicted to lose). Underdogs receive a bigger reward for succeeding, and favorites suffer a bigger penalty for losing.
This makes sense, but it's inadequate in the ways that matter most to players. The scheme's problem is caused by heavily abstracting the game's dynamics in the quest for programmatic simplicity. The resulting abstractions often fail as truly predictive models given the vast complexity present in each round. People want to experience "even" games on the micro scale: they want every game to be a robust, interesting experience in which all roles are covered, in which victories and losses don't feel wholly predetermined, and in which they feel rewarded for making clearly beneficial contributions. And they want all this without overly democratizing the game via an over-reliance on contrived turnaround mechanics or overly rigid limitations. They want individual skill to be central to the equation when it needs to be, but not so central that dominance is the game's central theme in regular practice.
To handle a nuanced system in which everything is measured and weighted independently would require a bigger active community, more programming resources, and challenging judgement calls in which programmers would be forced to objectively record and compare a litany of variables (e.g., per lifeform skill levels, gorge counts, average building time, accuracy - the list goes on) in effort to establish "even" teams. This is unappealing because it's a project of labyrinthine complexity and because the results would be of limited use until heavily tweaked.
But the existing system is unappealing for other reasons. And I think an assessment of the existing system will reveal that what's going on is more complex than the OP suggests.
I've written elsewhere that the game is fairly skill-compressed. By this I mean that there aren't that many routinely populated servers if we limit our selection to non-rookie servers and servers that at least approximate the game's intended balance (e.g., the servers that aren't 20 v 20, but which are somewhere between 6 v 6 and 10 v 10). And we must remember that rookie status doesn't last very long. So this leaves us with a handful of servers that ping well during prime hours in which a majority of the game's active players compete. It's not uncommon to see huge skill variance on these servers. Rookies, average players, high average players and occasional pros share space in the same arena. So let's say that we're looking at the following hypothetical 8 v 8:
Team One: Avg. Skill - 1712.5
P1 - 2200
P2 - 2100
P3 - 2000
P4 - 1800
P5 - 1600
P6 - 1500
P7 - 1500
P8 - 1000
Team Two: Avg. Skill - 1712.5
P1 - 3800
P2 - 2050
P3 - 2050
P4 - 1450
P5 - 1250
P6 - 1200
P7 - 1150
P8 - 750
1. Both teams have identical skill averages. But due to a huge difference in intra-team variance, Team One will win a majority of matches for the simple reason that they're better positioned to win a majority of engagements due to having relatively better players entering a majority of engagements.
2. This is then compounded by player role selection. Let's say that P1 from Team Two (3800 skill) wants to command. Or to gorge. Or to play on the team they don't often play on. Or to do anything other than reenact the full bore dominance that probably led to the heightened skill measurement in the first place. In any of these circumstances, Team Two is at an even greater disadvantage (because everyone else - already underpowered relative to Team One - is forced into the nearly impossible task of compensating). This is a vicious circle that entrenches and reinforces the stomp, because when someone is pressured to dominate, they're more likely to do so. I've actually encountered a number of people who have created secondary accounts not for smurfing, but for comming. They do this because their primary accounts have such high hive skills that they're practically forced to carry the team every time they play a shuffle, and it is very challenging to carry a team of underskilled players via comming alone. In other words: some people are actually buying additional copies of NS2 on new Steam accounts not to troll, but to avoid gimping their team whenever they want to command.
3. Smurfing. This is becoming more of a problem. I was previously skeptical about its prevalence, but I've been seeing it more often (I've played the game long enough to know when someone is actually smurfing as determined by their positioning and communication patterns versus a situation where a genuinely new player can simply aim). Let us be clear: a single dominant smurf completely obliterates the existing shuffling mechanism. Granted, it would be challenging to balance a game in the presence of hardcore smurfs even under the most intricate balance scheme, but a scheme which measured and rewarded a greater variety of in-game behaviors and roles other than winning alone would also reduce the incentive to smurf in the first place, because it would change how people looked at their contributions.
4. If P1 from Team Two leaves, the net effect on Team Two is dramatically greater than if any player from Team One leaves. Although an 8 v 7 should not be considered balanced, the loss of P1 from Team Two would create a scenario in which Team One maintains an average skill of 1712.5, and Team Two's average skill plummets to 1414.29. Further, Team Two becomes even more likely to lose a majority of engagements.
5. Following from #4, it gets worse: if a new player joins the server in the aforementioned scenario, and that player is newer to the game, they'll probably join Team Two out of impulse. After all, it's not like anyone from Team One is able to switch teams in order to help compensate after P1 leaves Team Two (unless they want to risk disconnecting and reconnecting, possibly losing out on their slot). This would drop Team Two's average even further. But if a new player joins the server and that player is well-versed enough to know how the skill system works, there's a reasonable chance that they're unlikely to want to join Team Two if it means that they're going to lose hive skill for a round they're not even "responsible" for - a round which seems like an inevitable loss.
6. This effect is further compounded by server placement and sudden changes in "regular" populations. There's a difference between achieving a high skill on a casual server vs. via gathers, for example. There really aren't that many populated servers that ping well for a majority of players, but it's not uncommon for some people to stick to one or two favored hangouts. Whenever this pattern is disrupted by the presence of one or more high skill players who enter a new space from more competitive environments - at no fault of anyone involved - the shuffle system breaks down even further.
7. In my experience, some servers have quirks which exacerbate the issue. We all know that it usually takes some time for people to join a team with or without a shuffle. On IBIS, for example, high skill players (myself included) can quickly join a preferred team before a round begins. When multiple high skill players do this, it usually triggers a not unreasonable "let's shuffle" reaction from the others in the server. But once the shuffle kicks in, the server attempts to average the teams by making as few changes to existing player allocation as is possible. This means that the server seemingly tries to keep your preference in mind. Did you join aliens before the shuffle? It's going to try to keep you on aliens. So let's say that a group of 3-4 top skill buddies join marines as soon as the map loads, and everyone else initiates a shuffle. Well, the server is going to try to keep those players exactly where they are, and it'll first try to reach an average by allocating everyone in the RR accordingly. This is a problem, but the reason that high skill players do it isn't always malicious: in some cases they simply want some measure of control over their choice of team, since that seems to be a vanishing privilege in the era of "shuffle and forget". In other cases, it's just a deliberate stomp that is, once again, encouraged by the existing system.
8. The predictive value of hive skill is further diminished by the fact that some rookies have a skill of 0, and older rookies - who stopped playing the game and then returned to it - have a hive skill of 1000. Although these two groups of players make identical contributions to a given team, the shuffle system weighs them completely differently. This effect is at its worst when combined with issues #1 and #2. And this occurs frequently.
I don't have an easy solution to offer UWE. But I do think that a greater emphasis needs to be placed on recording and measuring the vast array of positive contributions players can make in each round. Give players feedback and statistical information about these contributions. Break it down as finely as you can. And then, even if some generalized abstraction is necessary in order to arrive at a functioning balance scheme that doesn't resemble a five-dimensional Rube Goldberg device, make sure that those abstractions take more into account than performance relative to a frankly inaccurate winner prediction measurement.
@moultano, what are your thoughts on the above post.
Maybe one good on paper player has a bad day and cant hit s**t (happenes).
Or a team that looks weaker on paper has much better teamplay cause good communication or better decisions.
Even luck is a factor.
So you cant have perfectly balanced rounds, its impossible.
Especial in times with 1000 points and 0 point rookies.
What a system like hive can do is to break a obvious stack (1900 vs 500 skill), the result of the round comes from the human factor.
The solution is to have people of similar skill play together, but we have neither the player base nor system necessary for that. My solution is to lower expectations to avoid disappointment.
None of this is relevant to the OP where marines gradually become more dominant as the skill level increases. I can't imagine a solution that buffs/nerfs a side that would not affect the balance of the other skill range. Maybe the OP could look into playing the pub balance comp mod? Not sure if it will help, but it seems an attempt at balance for the higher skill levels.
Everything you said is true. But none of those problems would be solved by having the skill measure include judgmental factors instead of pure W/L. It would only introduce bias and incentivise bad player behaviour (skill farming), but all of the problems mentioned above would still exist. Teams would still be composed of an amalgamation of different skill values and playing a role one's not used to would still distort the prediction.