I'm curious to know what everyone thinks about typical RPG level progression. Specifically, opinions about exponential versus linear cost/progression. There are.. eh, four possible extremes, I think: 1. Exponential cost, linear improvement. 2. Exponential cost, exponential improvement. 3. Linear cost, linear improvement. 4. Linear cost, exponential improvement. These are not limited to character levels. They can also represent skill levels. Or even progression that wouldn't be considered levels at all. Which do you think is better for a game? Does it depend on linear versus open-ended exporation/gameplay? What about game length (short, long, never ending)? Anything else? My opinion: Exponential cost: Most games seem to go with this. The biggest problem I have with it is that the beginning of most games is already pretty exciting. Everything is new to the player, so it doesn't make a lot of sense to pile on a lot of progression there. Once the game moves along, the leveling aspect becomes so slow that it might as well be shut off. Exponential improvement: There's not much I can say about it. It's just pretty lame when a level 21 NPC/player can squish a level 19 NPC/player like a bug, regardless of how long it took them to get from 19 to 21. Linear cost: I like this one. The player's progress never slows down. That means very slow leveling near the start, but still leveling at that same rate while nearing a maximized character. The most negative aspect is that you're stuck with a beginning character for a longer time. Linear improvement: Pretty much the opposite of my complaints for exponential. A level 19 NPC/player can still get the upper hand on a level 21 NPC/player. That means AI enemies don't become useless/pathetic as quickly, and that the player can compete ahead of what the game expects with playing skills or strategy. What do you think?

Quote:What do you think?

I think it is relative. If the monsters are just going to improve along with the player, then how fast the player improves is irrelevant, except as a reward (and only a perceived one at that as there is no actual improvement due to the fact that the monsters also improve at the same rate).

So it comes down to how often you want the player to be rewarded and how big you want those rewards to appear to be.

As for frequency, a constant rate of reward eventually weakens in its perception over time. So a linear rate of reward (linear level progression) will mean that as the game progresses, the size of the reward perceived by the player will be reduced. So this means that you have to give bigger rewards later in the game or the player will stop seeing the level up as a reward.

So to maintain the rewards effectiveness in the game, you will have to give bigger and bigger rewards for each level up, or have an exponential improvement if you are using a linear level up system.

If, on the other hand you are using a non linear reward rate (exponential cost), then as these are not regular reward periods, you don't suffer the effects of a diminishing returns (due to the psychology of humans). This means that you can get away with a linear progression (but can still use an exponential progression if you want).

I think the big problem with "linear improvement" is what is called "diminishing returns." The amount that you increase may be the same when leveling up from 1 to 2 as it is from 99 to 100, *but* it makes almost no difference at levels 99. If your damage equals 5 * level, then you double your damage by getting to level 2. 5 damage to 10 damage is a big difference (10 hits to kill someone becomes 5 hits). 495 to 500 damage is hardly a change at all (if it took 10 hits to kill someone with 495 damage, it will at best take 9 hits to kill them with 500 damage). To double again, you must then reach level 4, then level 8, then level 16, then 32. Looking familiar now? Yeah, that's just like exponential cost and exponential improvement, but the levels are bigger.

Quote:Original post by Ezbez
I think the big problem with "linear improvement" is what is called "diminishing returns." The amount that you increase may be the same when leveling up from 1 to 2 as it is from 99 to 100, *but* it makes almost no difference at levels 99. If your damage equals 5 * level, then you double your damage by getting to level 2. 5 damage to 10 damage is a big difference (10 hits to kill someone becomes 5 hits). 495 to 500 damage is hardly a change at all (if it took 10 hits to kill someone with 495 damage, it will at best take 9 hits to kill them with 500 damage). To double again, you must then reach level 4, then level 8, then level 16, then 32. Looking familiar now? Yeah, that's just like exponential cost and exponential improvement, but the levels are bigger.

But that's an unreasonable amount of progression (which plenty of games employ). You're creating too much of a gap. 5 damage to 500 damage? There's no conceivable way for a person to range from 5 damage to 500 damage with skill alone. Try something like 100 damage to 200 damage, or 50% damage to 150% damage. That will be much less obvious.

Such as..
sword_scaler = sword_level / 25
damage_scaler = 0.5 * 1-sword_scaler + 1.5 * sword_scaler

If a certain sword caused 100 damage, it would be:
Level 1: 54 damage
Level 2: 58 damage
Level 8: 82 damage
Level 9: 86 damage
Level 24: 146 damage
Level 25: 150 damage

Climbing from 54 to 58 is not much better or worse than 146 to 150.

This type of progression is less noticable per level up, but that also means enemies and areas don't become outdated as quickly (instantly).

EDIT:

Actually, I think the point is that with linear progression, you need to start the player out at a humanly realistic scale. Not like your average out-of-the-box player character, who shoots a ranged weapon and misses by a 90 degree angle.

I think it's not just as simple as which of these methods you use.
The way they fit into the game is what really makes or breaks the system.

You could design a game that works around all of these.
So it's really hard to say that one is definitively better.
(Although I agree with others that you want to avoid diminishing returns problem.)

I would say, however, that my favorite is exponentially harder levels with a linear improvement scale. Additionally, in my personal opinion, you don't really need a level cap if you fit the system into your game correctly.

That's a very interesting discussion for me. I've been going through a lot of balancing in my RPG as of late. I've explored a lot of formulas for characters and enemy growth. I'm going to explain what kinds of formulas I chose and why. If you have any comments, be my guess, balancing is fun but hard. Tweaking numbers on an Excel sheet is one thing, but playing the thing in practice is different, because there is way too many variables to check, such as the number of foes per groups, the availability of shops (in case the player needs more potions) the enemy's AI, the specs of the characters, etc.

For level growth, I'm using a mix of an exponential and a quadratic function. (At low level it is quadratic because exponential scales too slowly and I don't want to player to score 10 levels in 5 minutes, after IIRC level 15 it becomes exponential) so I guess you could call it an "exponential cost" according to the TC's criterias.

However, I set the enemies' XP so that if you progress normally through my RPG, it feels actually linear, not exponential. Allow me to explain: what I do when setting enemies' XP is that for level X, I expect the player to kill Y enemies of level X, and the function for Y (average number of monsters I expect to be slain to go from level X to level X + 1) is linear. For example, to get from level 15 to level 16, it takes 4,372 XP and I expect the player to slay in average 25.72 level 15 foes, so the average XP for a level 15 enemy is 4,372 / 25.72 ~= 170.00. (±35%) So, thanks to the exponential XP function, my system puts a soft ceiling to levels by making it incredibly time-consuming to power-level past a certain point, but it also makes a low level character catch up really quickly. In my RPG, a character joins the party at level 2 in a level 8~11 area. After only two fights, I think she was already level 6. The high increase of an exponential function makes unusually low level characters correct very quickly, at any level. A linear cost growth would only work like this at lower levels, at higher levels, it would take a lot of time for a character 10 levels below average to catch up. I'm not saying this is bad, some developers might want this; it is simply not my case.


Concerning the growth in power, I'm still in the process of balancing, but I sure don't want linear damage progression. The diminishing returns here are too pronounced. Imagine a character does 20 damage on level 1 and gains an additional +20 damage per level. The damage increase between level 1 and level 2 would be 100%, then 50%, then 25%, etc... At low level, even-leveled enemies would become obsolete after one level. Then you tone down the scaling to 20 +5 per level so low-level isn't as crazy... Good now? Oops! Now the problem is at higher levels where player barely notice the effect of leveling.

For my part, I opted for a 3 variables which all increase linearly, and those are all multiplied together so that looks like an exponential improvements, though it's more a cubic one. Cubic and quadratic functions still have diminishing returns, but it's much less noticeable. Besides, I find a function with a *small* diminishing returns beneficial. On level 3, fighting a level 8 should be very dangerous, but on level 80, fighting a level 85 should not be a certain doom. Without diminishing returns it would be just as dangerous here.


EDIT: Split the wall of text in paragraphs. ;)

Quote:Original post by Bearhugger
Concerning the growth in power, I'm still in the process of balancing, but I sure don't want linear damage progression. The diminishing returns here are too pronounced. Imagine a character does 20 damage on level 1 and gains an additional +20 damage per level. The damage increase between level 1 and level 2 would be 100%, then 50%, then 25%, etc...

The problem is that you're adding 20 damage per level and starting at 20. If 20 represents a single learning step, your beginning character is worse than a complete moron. He's totally braindead. Something on the level of a newborn baby trying to wage war. Consider it. A 1 year old probably could do double the damage. And a 2 year old would probably triple it.

That's just not realistic. Humans can pick up weapons or dive into actions that they've never seen before and do a decent job with them. Don't start at the bottom.

Quote:Original post by Kest
The problem is that you're adding 20 damage per level and starting at 20. If 20 represents a single learning step, your beginning character is worse than a complete moron. He's totally braindead. Something on the level of a newborn baby trying to wage war. Consider it. A 1 year old probably could do double the damage. And a 2 year old would probably triple it.

I have used these numbers for the sole purpose of magnifying the flaws of linear scaling. A real game would be, I dunno, 2 x (Lv - 1) + 20? The benefits of leveling in a linear system is never uniform. This is just a formula, this might be what a game designer wants, but in my project, I think that linear scaling would be flawed.

Quote:Original post by Bearhugger
I have used these numbers for the sole purpose of magnifying the flaws of linear scaling. A real game would be, I dunno, 2 x (Lv - 1) + 20? The benefits of leveling in a linear system is never uniform. This is just a formula, this might be what a game designer wants, but in my project, I think that linear scaling would be flawed.

So your damage would be 22, 24, 26, 28, 30, 32, etc? What's wrong with that?

It should only be flawed in a system where enemies and challenges climb at an unrealistic rate. That might be the case when you go from figthing rats to fighting demons, but not with any reasonable progression.

I guess it's a personal opinion. I think enemies should become slightly less of a challenge, but not obsolete. They should still influence the game world, rather than being left behind. One benefit of it is that the player isn't limited to a few types of enemies at a time. Nearly all enemies in the game world could be fought with a reasoble chance to lose or win, at any one time.