Stats: The Good And The Ugly

(Stroggafier) #1

Persistent stats are an integral part of any game, and electronic games are no exception. However, a key evolution in video games is that permanent stats are implemented in the hopes of supporting both publisher’s and designer’s contentions that the game has a long play life, rather than simply analysing play results, as used in other types of games. Since long play life is a key sales feature of video games, persistent stats become an important marketing tool.

From a marketing perspective the longer a player is engaged, the better for the longevity claim. Simply put, the more hours played, the stronger the claim. However, using hours played as a key ranking stat, would be too obvious. A less obvious measure is needed. Points, instead of hours, can become a thinly veiled but effective alternative. First, points can allow some variation from a simple hours tally. More importantly, it is obvious that the longer a player plays, the better a player becomes and the more points a player acquires. Hence, points can be legitimized. Rank is the name of this legitimacy, or so, that is the marketing contention.

Whether or not the marketing contention itself is legitimate depends entirely on how the stat system is designed and implemented. And it may be in the design, and not the implementation that the problems in ETQW permanent stats rest.

As can be demonstrated, the stats system, as implemented, does not support the premise that the better the player, the higher the rank. Several key reasons why this is so:[ol][li] ETQW contends that objectives are at least, if not more important than points. Yet, in a typical player point tally, points from completing objectives accounts for only about 10% of the tally, yet points determine rank.
[/li][li] Points are not aligned with how difficult it is to obtain them. That is, some points are more difficult to get than others, yet that difficulty is not reflected in the tally.
[/li][li] There exists an ability to manipulate the game dynamics to maximize points at the expense of other, possibly more important, game elements.[/ol]
[/li]To get the discussion started, here are three, of possibly many, solutions that might work:[ol][li] If completing objectives is more important, assign 10x the current number of points to objectives completion[]Infantry kills are much harder than vehicle kills, so assign 5 pts per infantry kill and only 1 per vehicle kill.[]include stats such as: Number of TKs of Team-mates Working On An Objective (to prevent TKing just to personally complete the objective)[/ol]
[/li]This thread can hopefully explore these issues further, identify key characteristics of good stats, discuss how to fix bad aspects, and possibly show how future systems can benefit from these insights.

(light_sh4v0r) #2

The following thread was made just yesterday, same subject:
http://splashdamage.com/forums/showthread.php?t=20277

(Stroggafier) #3

@light_sh4v0r, at first I thought so too. But on closer reading, that thread discusses a different aspect of stats - that is, the mechanics of stat distribution.

This thread is about which stats best reflect the game, how they are accumulated/calculated and how bad stats can misrepresent rank. Once that is figured out, then we can go to @Cankor’s post and debate how best to access, report, distribute them.

(Stroggafier) #4

Scores are usually calculated using a technique of weighting. The current point system in ETQW is a weighted system. One can build a table for gaining or losing points for any action. This type of design can be refined as a game is play tested. I’m suggesting that the stats systems need to be actively designed and refined just like every other aspect of the game.

The attached table is simply my view of how to better weight points.

(Cankor) #5

[QUOTE=Stroggafier;207656]
To get the discussion started, here are three, of possibly many, solutions that might work:[ol][li] If completing objectives is more important, assign 10x the current number of points to objectives completion[]Infantry kills are much harder than vehicle kills, so assign 5 pts per infantry kill and only 1 per vehicle kill.[]include stats such as: Number of TKs of Team-mates Working On An Objective (to prevent TKing just to personally complete the objective)[/ol][/li]This thread can hopefully explore these issues further, identify key characteristics of good stats, discuss how to fix bad aspects, and possibly show how future systems can benefit from these insights.[/QUOTE]

In ETQW the points (XP) awarded for various activities isn’t as arbitrary as you seem to think, and the first two of your three points are already addressed in some fashion or other.

To your first point: You get more XP for completing objectives than any other task you could perform, by a significant margin. You get more XP for repairing the MCP than you do for repairing a deployable or any other vehicle. You get more XP for building a bridge (assuming you build the whole thing yourself) than you do for blowing up a mining laser, generator or hacking a shield generator. But yeah, some things are harder than others and the XP awards are roughly the same, but the XP awarded is still more than enough to be compelling (if it weren’t already compelling enough just to help your team win).

Infantry kills are harder than vehicle kills? I’m not sure if you accidentally typed that backwards or you think that’s the case. XP for kills is assigned by how much damage you do, and which weapon caused the damage. Light weapons yield more XP/damage than heavy weapons (same for light vehicile weapons vs heavy vehicle weapons), but the amount of damage you can do is also much less.

XP amounts awarded are already generally more per damage point vs infantry than they are vs deployables or vehicles, deployables and vehicles just happen to have a lot more “health” than infantry, so if you destroy them completely you end up with more points.

Your last point is better solved with better anti-griefing methods or just a well admined server. Guy shoots you to do the objective, tell everyone and then vote-kick him. I guess this doesn’t work on consoles (part of the reason why PC gamers hate Infinity Ward now).

The amounts awarded are generally pretty good and you can tell they at least tried to reward positive behavior. For instance a medic gets more XP for reviving a fallen team mate than he gets for killing an enemy soldier, therefore the medic is encouraged to do what medics should do. The only ones I think are screwed up are the awards for destroying a spawnhost (if I remember correctly it’s .125XP, in other words you have to destroy 8 of them to equal killing a 100 point enemy when he’s at full health) and disarming mines (can’t remember what it is, but it’s pretty low).

They also incorporated XP awards into player callable missions as another way to provide incentive to help your team. There could have been more of these types of missions (I wish there was one for deploying smoke for instance), but the ones for destroying deployables and so forth are pretty good.

Stats and balanced XP awards are different (but related) subjects. You want XP awards which provide incentive for positive (team based) behavior, and you want to be able to get roughly equal XP awarded per hour of play regardless of the class you choose (that’s what I mean about “balanced”). You want stats which also provide positive incentive, but this is harder. A K/D ratio which is publicly available for everyone to look at for instance will tend to make players play more conservatively (doing fewer objectives because you die a lot doing them) and focus more on killing the enemy than fulfilling a role which is most beneficial for the team (medic running in to revive the engie who just died diffusing the charge). This is why I think you should be able to make your personal stats private, then you’re not so worried about having a crappy K/D ratio or something like that.

Regardless of how it is now, it’s not going to change for ETQW. Also, seems like the ETQW stats is down for good along with the community forums.

(Stroggafier) #6

yesir, I hear all of that @Cankor and agree with most of it. Solid background info.

First, I was hoping to get some concensus around the idea that bad stat design is the culprit and not merely “stats” as we so often hear.

e.g. Does anyone feel that the current ranking system, as described by points, correctly reflects players’ abilities? If not, shouldn’t it?

I hope people speak-out on that.

Second, Changing the in-game stats calcs, sounds at best like a mod, and at worst its not do-able; I have no illusions in that area. What sounds possible is a separate stat database that feeds off the in-game data, then is revised/adjusted to create a new set. The revision algorithm can be expressed in tablular form. A re-calculation matrix, as it were. We’d have the “official” stats and we’d have the “cumminty approved” stats. :slight_smile:

Next, the sample table I showed was only a preliminary and incomplete picture. There may be several tables that form the adjustments and a further level of granularity in the mapping of action vs affected is needed (like you were getting into in your reply). Please note that I believe a good design is not simply a damage vs hitpoint equation, but must also consider the desirability of that action. Indeed, I believe this is the central error that SD may have made in many of the ETQW stats.

Some examples:
Even though completing an objective curently gets quite a few points (more than other actions) it may still be not enough to incent the right behaviour - maybe should be doubled/tripled;
And yes, agreed, the medic/technician stats need fixing as 0.125pts-per spawnhost is rediculously low;
Also, a kill is not worth the same regardless of how its done, some are worth more to the game flow and some are more difficult to do.

BTW. The logic for granting 5 pts for infantry-on-infantry kills while reducing vehicle points is to reduce the vehicle spam. The numbers may not be exactlly right. Its the concept that I was trying to get across.

Agree that incentive rewards are related yet different, but work in tandem with points.

Finally, agreed, admin support for some of the more questionable actions is a very good mechanism, as I have advocated that myself on the old forums. The stat system might be able to strengthen that support by providing stat evidence.