Why do game designers like diminishing returns so damn much?

[MrWaeseL]

This is mostly from my experience with computer role playing games, since the only pen and paper one I play is D&D.

In Vampire: The Masquerade Bloodlines spending your first point in hacking costs 3 xp. The cost of points after that rises until the last point costs 15 xp. Each point adds the same amount to your hacking skill.

In System Shock 2 your first point of strength costs 6 Cyber Modules. Your fifth point costs something like 40. All points in strength add the same amount of inventory space and melee damage.

In Fallout 1 and 2, raising your skills beyond a certain percentage cost double and eventually triple normal skill points.

In D&D, greater weapon focus, improved two-weapon fighting and other feats all do the same as the base feat, but have higher prerequisites.

So in short, diminishing returns abound in RPG's. Why?

[Fwib]

Because that's the way it works 'in real life'? Most things in reality are more difficult to improve from good to better than average to good.

If the game-world did not imitate life, then our knowledge of how the world works would be of no use to us in-game.

[AlphaNerd]

Because it rewards generalits (gish) over specialists. Given that specialization is generally better anyway, it seems a reasonable metric.

It also feels more realistic.

Joshua

[Username17]

Realism doesn't enter into it, as the meanings of the numerical values that are being raised are as subjective and modifiable as the costs to raise the values.

But Joshua has a point. Essentially the Bard isn't really overppowered, and diminishing returns is a very easy and comprehensible way to ensure that people who want to play an enchanter/singer/vagabond end up with a Bard and not something even worse like a Rogue/Sorcerer.

And that's the deal. Somehow you want to ensure that if a player decides that he wants to be "knife man" that he stays on the same random number generator as the rest of the party. Somehow you want to ensure that if a player decides that they want to dabble in Ice Sculpture that they don't fvck themselves forever off the growth curve.

A level based system handles this by giving people points into column A and column B, therefore forcing players to spend some of their points on knowledge of Romulan wine or whatever. A skill based system has other tools available, but it basically comes down to either charging people increasingly large amounts to raise their knife skill (thereby incentivizing character diversity), putting in caps (thereby forcing character diversification), or simply instituting so many tasks that characters will need in order to complete their mission that monomeric specialization is simply detrimental to your character and the team (thereby encouraging character diversification).

---

That being said, the standard triangular groth path evidenced by games like Vampire and Shadowrun fail on first principles. They are not, mathematically speaking, sharp enough of a curve to actually make non-specialization a good life choice. But they are enough of a pain in the ass that psychologically they seem like a big deal. That is, while it's still in your interests to simply set fire to every point of XP you ever get to be the total Dominator, the amounts of XP involved are eventually large enough that people bail and do other things anyway.

It's an exploit of the human factor, and it does often bring about the desired result of people doing several things with their character. That being said, if you actually wanted an extensible game based on a skill system with diminishing returns, you would have to go to exponential costs. Seriously, any geometric system is always eventually going to push the guy maxxing one skill off the RNG with another character spending the same points on maxxing two skills.

-Username17

[RandomCasualty]

I don't think diminishing returns are necessarily bad. While Frank is right that eventually, they end up going off the RNG, this happens at a much slower rate than straight linear systems. THe fact is that while no system extends to infinity, no game goes to infinity either, so we just need a system to carry the game to a point and then stop. So diminishing returns can help you keep the game sane for a little longer.

In truth however, there really isn't a heck of a lot you can do to truly stop specialization if you have a game that rewards it. One of the main problems with combat actions is that you only get a few actions per combat. Thus, you're better off having a really good sword attack as opposed to a moderate sword attack and a moderate bow attack. The reason being that you cannot attack with both the sword and the bow at the same time. So long as you can always use your sword, there's no reason not to super specialize.

[AlphaNerd]

So long as you can always use your sword, there's no reason not to super specialize.

Yes, economy of actions is one of the reasons that specialization tends to win out.

With an exponentials system, you might have

Swords 0, Bow 0
Swords 4, Bow -8
Swords -8, Bow 4

If you can always choose your attack, then sure, either specialist is probably better (assuming swords and bow are the same goodness). If you have to use both frequently, then the generalist is probably best. If it's somewhere in between, then which is best might be iffy.

I don't see why an exponential system would have to run out of steam ever. If people invest a consistent fraction of their xp in each of the basic skills, then you can end up with a long-term situation where chances to hit, etc. remain static.

As a matter of game design, the exponential factor you use plays a big part in determining the relative strengths of a generalist and a specialist. if you assume a factor of two, then the generalist is always one behind the specialist, which probably makes for a strong generalist. If you use a factor of 1.2 or so, then the generalist is behind about 4 points. Note that this is assuming long term behavior and two-skill generalist and one-skill specialist.

Joshua

[RandomCasualty]

1161204630[/unixtime]]
With an exponentials system, you might have

Swords 0, Bow 0
Swords 4, Bow -8
Swords -8, Bow 4

The problem with something like this is that your two attacks rapidly get to the point where they're not even on the same RNG anymore, which makes characters boring and one dimensional.

The problem isn't penalizing other actions, already people are looking for every reason to never use another weapon other than their build's chosen weapon. The limitations of actions favor that, the magic item system favors that, the tactical combat system favors that and the bonus system favors that. We don't need to kick a person further for doing something that's not in their specialization. We do that too much already.

Really, the change has to be more to the game itself rather than on the advancement system. You need to force people to do stuff they're not specialized in. Sometimes the archer is going to have to engage in close combat with his short swords. And sometimes the barbarian warrior will have to use thrown weapons or a bow instead of his great axe. Right now, the paradigm is "specialize in one style and take abilities to ensure that you can always use that style." That has to change.

Which probably means no 5' step'n'fire tactics to avoid AoOs from firing a bow, and no more flight or super movement to allow meleers to always be able to fight everything.

[Username17]

The problem with something like this is that your two attacks rapidly get to the point where they're not even on the same RNG anymore, which makes characters boring and one dimensional.

To an extent.

An exponential system of advancement prevents a character invested in 2 things from falling farther behind the character invested in one. But it also makes changing your character's focus prohibitively expensive.

And of course it doesn't do anything about Ability Drop. Here's the deal on that:

When modifiers are relatively low (which in many games maps to inexperienced characters), characters can perform default actions with a noticable chance of success. In d20, for example, a 1st level Wizard with a strength of 11 has a pretty decent chance of laying out an enemy goblin with a blow from a stick. +0 on the d20 has a decent chance of hitting the DC of 14, and while the specialist (raging strength of 22 and a greatsword) certainly performs that task better, he still enjoys a chance of failure and the totally defaulting wizard with a stick still enjoys a chance of success.

However, as modifiers become higher, the challenges that the team faces are still geared towards the specialists, and the non-specialists can't do shit for shit. At 10th level, our specialist melee monkey has bonuses in the +25 range and our Wizard has a +5 with his stick. When you face the big monster in the dungeon it's a Steel Predator (AC 30) and the Wizard might as well not even swing that stick.

And the exponential system does not prevent ability dropoff. If you set an exponent such that each +2 costs twice as much as the last +2, then for 8 points you can have three different characters:

1. Bows +4, Swords +4
   
2. Bows +0, Swords +6
   
3. Bows +6, Swords +0

I could see how all of those might be usable characters. And I could see circumstances where a character might want to use swords or bows even if that was the thing they were shitty at (a penalty of -6 is overcomable on a d20). But imagine the sam group on 64 points:

1. Bows +10, Swords +10
   
2. Bows +0, Swords +12
   
3. Bows +12, Swords +0

The generalist is syill a viable life choice, but the specialists essentially no longer have the ability to fight poorly with the other fighting style, they may as well erase that from their character sheet altogether. They can't even contribute in the other field, even though they are no worse than when they started. That's ability drop off right there.

---

Now one thing you could do is to set things on an exponential and then give people points that could only be spent on abilities below a certain threwshold relative to their total ass kicking. This would ensure that people always had some -6 tricks up their sleaves...

-Username17

[AlphaNerd]

Having problems with ability drop -- don't put that last point into your specialty skill. Instead, put like a few points in the second skill:
Bows +8, swords +11

With a steep xp curve, it's really cheap to buy yourself off the ability drop, so there's no real excuse for not doing it. If the curve is too shallow, then you have other problems. Like I said, exactly what xp curve you use has a huge impact on how the game is played. I think I like 2^(1/4) as the exponential factor. 2^(1/2) seems too sharp and favoring the generalist too much.

If you can apply a "level" to the player, perhaps you could do that pretty easily. Level would presumably be a exponential function of experience spent, with the same factor as the single skills.

But I think if you gave players the option of 11->12 in swords or 0->8 in bows, most would choose 0->8 in bows. So long as people spend a constant fraction of their xp on the basic skills, then the curve doesn't change as you gain in power. Sally is alway at +5 against Bob with primary, -2 with secondary, and Bob is always +4 agains Sally with primary and +0 with secondary, or whatever. Trouble brews if people don't invest (somewhat) distributionally -- it makes them wonky, usually to their detriment.


Joshua