Celus Ivara
DUST University
Ivy League
268
   |
Posted - 2015.01.13 20:49:00 -
[1] - Quote
CCP Rattati wrote:Attributes:
New Player Skill Points/BattleSecond = A (integer)
Veteran Player Skill Points/BattleSecond = B (integer)
New Player Skill Points/War Point = C (integer)
New Player Skill Points/War Point = D (integer)
Minor point of order, I think you mean:
Quote:New Player Skill Points/War Point = C (integer)
Veteran Player Skill Points/War Point = D (integer)
For numbers, I'd say a moderately well played, average length Skirmish should award about 7000SP (before boosters).
A very well played Skirm should hit about 10,000SP.
Let's say an "average length Skirmish" is 15min (I actually don't know how long it is, some metrics would be nice here).
So at 900 seconds, a moderately well performing player should be getting about 7.7Ļā SP per second.
I'm going to do a design hack here, and presume the 80/20 Rule is a design goal for us.
So let's randomly say a new player gets 80% of their SP from match time and 20% from WP, and the vet is the inverse of that.
For a battle rewarding 7000SP this breaks into:
NP: 5600SP from match-time.
Vet: 1400SP from match-time.
So if those are the numbers we want to see from a 900 second battle:
for "New Player Skill Points/BattleSecond" we'd want a value of 6.2 ,
and for "Veteran Player Skill Points/BattleSecond" we'd want a value of about 1.6 .
(A note here, these variables are presently set to integers. Can we change that to a float? When dealing with numbers these small, granularity would be nice.)
Now for WP to SP, we're going to be forced to do some wild guesses at numbers.
First, what is the cut-off between "New Player" and "Vet"?
Let's randomly say 12 million SP.
So, the group of players under 12mil SP, let's guess they get in the range of 0WP to 1500WP in a Skirmish.
For the group of players over 12mil SP, they might get in the range of 400WP to 3000WP in a Skirmish.
Let's grab the mid-point of those two numbers and go with:
"New Players" (heavy quotes) pull 750WP a match.
"Vets" (very heavy quotes) pull 1700WP a match.
So if we go with the goal of an average 7000SP a match, and the 80/20 Rule:
"New Players" would get 1400SP from WP.
"Vets" would get 5600SP from WP.
Using the (highly arbitrary) mid-point values above:
"New Player Skill Points/War Point" would be 1.87 .
"Veteran Player Skill Points/War Point" would be 3.29 .
So let's double-check these values and look at the SP ranges for the Skirm WP performance ranges presumed above.
New Players:
0WP in Skirmish = 5600SP + (0WP * 1.87) = 5600 SP
750WP in Skirmish = 5600SP + (750WP * 1.87) = 7003 SP
1500WP in Skirmish = 5600SP + (1500WP * 1.87) = 8405 SP
Vets:
400WP in Skirmish = 1400SP + (400WP * 3.29) = 2716 SP
1700WP in Skirmish = 1400SP + (1700WP * 3.29) = 6993 SP
3000WP in Skirmish = 1400SP + (3000WP * 3.29) = 11,270 SP
And some of those numbers sound insane, right?!
A 750WP player moving from the New Player tier to the Vet tier will instantly go from a respectable 7000SP to a demoralizing 3880SP.
Players who joined in Beta, but have only been playing occasionally will be instantly thrown into the Vet pool and getting similarly depressing rewards.
I think the main problem here is that two tiers is far too stark to meaningfully represent the vast gradient of skill levels in the playerbase. 
Like others here have proposed, I'd like to see this pushed to a more multi-tiered system.
1 - Recruit - Brand new to Dust, most still in Academy, still figuring out how the game works.
2 - - Understand the basics of the game, working on learning and building into roles.
3 - Mid Tier - Specialized fully into a role and has several half-finished side roles, learning map strategies.
4 - - Multiple roles that can be switched into, has a pretty good understanding of strategies.
5 - God Tier - Highend PC ringers, masterful understanding of most every role and strategy, only ever lose to other God Tier.
Alternatively, we could do away with the idea of stark tiers all together, and have a sliding scale where the more SP you have, the more you are pushed to the WP-focused reward end of the algorithm.
One final note about tiering:
There are situations in which a player is high SP, but low in WP earning ability. Since we don't want to lose the since of reward that is the SP payout, some people here have suggested moving to a more skill focused system to deternine where you should be placed in the tiers.
On the surface this sounds good, but there is a big problem in that the main purpose of this system is to hinder the reward for AFKing. Since a dedicated AFK character will in most cases have a low ++ (mu), then they will always be placed in the lowest tier, receiving the hgiest SP reward for time, and thus we'd miss a massive amount of the AFKers.
We will still catch people AFKing on their main, though, and that is something.
CCP Rattati wrote:War Point Threshold = T (WP)
Veteran Minimum War Point Threshold = M (WP)
I'd suggest making these a factor of time spent in battle.
Good values for T & M will in Skirm will be bad in Ambush.
Also, Scotty putting you in battles that are 50%+ over is already unfun enough. The idea of being put in a battle with not enough time to hit the minimum WP threshold will be highly upsetting to players.
To start to offer some input on values, though, I tend to pull around 2000WP in a Skirmish, and on a few perfect matches I've hit 3000WP. And I know that A-tier Logis can get into the high 3000's when the stars align.
For a Skirmish played beginning to end, a safe value for T would probably be 4000WP.
And in that same situation, a value of about 200 would probably be a safe value for M.
For P if we're working with "safe" values for the other constants, I'm happy going with something aggressive for the diminisher.
I'd put the "War Point Diminisher" at 50%; maybe harsher. |
Celus Ivara
DUST University
Ivy League
276
|
Posted - 2015.01.14 15:23:00 -
[2] - Quote
So something that is happening in this proposal and discussion that I don't think any of us fully realize, is that there are kind of two goals to this initiative.
1: Reward talented players directly for their performance on the battlefield, while keeping a time based reward for unskilled player so they still get payed out while they're learning the game.
2: Greatly hinder AFKing by decreasing/eliminating their match-time SP payout.
A sub-topic of #1 is "How do we determine if a player is a New Player or a Vet?"
There are two ways to make that determination: Player talent based, or SP total based.
For "Player talent based", a number of suggestions have popped up.
Honestly, I think Regis's is the best I've heard.
Regis Blackbird wrote:To minimize the risk of a SP "cliff" when moving from "New Player" to "Veteran", and to accommodate different match lengths, I would propose the following mechanic:
For a player classified as "New Player", BOTH calculations execute EOM and the highest calculated SP is taken as payout. If the "Veteran" payout is the highest more than half the time the last 20 games (example), the player is bumped up to "Veteran" status.
A variant option is to just use the hidden mu stat (the in-game estimation of how good a player is, used by the Team Balancer in matchmaking).
These are very good methods of solving the Tier selection sub-problem of #1.
But the problem is that in the great majority of cases, a AFKing player looks exactly like a normal not good player to the computer!
So if we utilize a Player talent based system, we'll successfully put New Players and Vets into the correct reward schemes, but we'll also be pushing most of the AFKers into the highly match-time rewarding New Player tier.
And if we instead make the tier decision based upon SP totals, we'll be pushing a ton of casual players into the "pull 2k WP or go home broke" reward scheme.
Both situations are very bad, and I'm starting to worry that this conflict is unresolvable!
There may be no way to force AFKers into a play-to-be-paid tier without screwing over new-berries. :\
I will say though that implementing #1 is worthwhile in and of itself, even if it has no effect on AAFKers.
And, any of the variations of this system we've been talking about will greatly hinder the AFKers who are doing it on their main; which by itself would also make this endeavor worth it.
My main point is that we probably can't use tiering to eliminate AFKing on alts without screwing over normal low WP players.
Now on to a slight tangent:
If we want to lessen AFKers, we need to take a multi-faceted approach. Now, we already are doing this, through filtering out players engaging in AFK behavior, and adjusting the match-time WP reward to lessen the incentive to AFK.
But one thing that hasn't been brought up that we should be discussing is that the weekly SP cap was tripled a few months ago by accident!
The change was left in place as an experiment to see if it broke anything. Most of the game was unhurt by it, but one new thing did appear, players could get a massive SP reward for AFKing their characters.
It may be time to reevaluate whether having an SP cap so high it's only reachable through AFKing (or a self-destructive level of no-lifeing) is something we want in our game. :\ |