The unified experience theory

Remove all ads!

The final pieces of the great experience award theory puzzle seem to have fallen to place, so let me present how this actually works. For those interested, this unified theory not only applies to IWD2 but can freely be applied to ANY game using 3rd edition rules.

Could I get a TL;DR please?
Apart from blatant cheating and/or exploiting engine loopholes, the 3rd ed experience system is just the thing you'd want: It works and it also greatly diminishes the "value" of powergaming, while making sure that a few unwise decisions along the way won't permanently set you back to any significant degree. So don't sweat it.

Ok, I can handle the maths. Hit me.
It all starts with the actual game layout. By playing through the game with a standard zero ECL, full size party from the word go is one way to measure it, but whatever works as you can calculate the outcome based on your input parameters. No matter how you slice it, all of the possible permutations of how you want to game the system, finally boil down to what you'll actually meet in terms of enemies and/or quest exp available in the game. The fact that IWD2 allows for level-squatting (ie. making it possible to have your actual and earned levels differ) is a coding mishap, yes, but even that can be calculated for if you're interested in such exploits.

Rule #1: Level that differs from the expected value.
Whenever you're at an (average) level that differs from the expected value discussed above, your exp gain from kills (but not quests, at least in IWD2) will differ as well. Breaking the exp awards table to its terms, you have:
exp award = A x A/B x C, where
A = CR of the monster in question
A/B = CR of the monster in relation to (ie. divided by) your absolute level
C = Alternating 4/3 or 3/2 to a power of the difference between your absolute level and the monster CR. For large differences, this is easily approximated by using square root of two as the base, as 4/3 x 3/2 = 2.

Hence, the relation of exp awards gained at levels past (or behind) the expected level is no simple function. However, as you're most often supposed to be battling monsters that have CR values at least somewhat close to your expected level (+/- 7 CR as per the 3rd Ed. Dungeon Master's Guide), a rule of thumb that IWD2 gratuitously breaks in several places, it's usually enough to only consider the "power of square root of two" term. IWD2 also alters the actual CR value of some bosses based on your average level, but compared to the hordes of standard mobs, this effect is negligible at best.

In practice, this means that even a slight overtramp in the actual level (in comparison to the exptected) is quickly beaten into submission, while staying lower will reap great rewards, even more so during the early levels where the A/B term still has a lot of say on things. Also, for the poor souls using level-squatting, this is THE reason why it's so effective, and only gets better the longer you wait with cashing it in.

Rule #2: ECL penalties.
Short version: Your expected average level is only slightly affected by your ECL penalties...
Long version: ...but it also depends on the absolute expected level range. Early on, the percentual difference in exp required to reach the next level is quite significant, but it slowly decreases as your absolute level increases, reaching mere single-digit percentages past level 22. Hence, even the steepest of ECL penalties will ultimately reach a point where you're only a partial level behind the corresponding zero ECL party. But you WILL always stay behind.

ECL Anomaly: However, as long as your average level is calculated using the floor(x) function, ie. ignoring remainders, the actual ECL vs expected level, will treat anything with a non-zero remainder one level LESS than expected. For example, a party with five ECL zero characters and one with 1 or greater ECL will actually behave as a party having ECL of NEGATIVE ONE would!

Rule #3: Smaller than maximum party size
This has the opposite effect compared to ECL penalties in that your exp awards per character will actually increase in direct relation to party size vs maximum allower party size, both for kills and quests. Even if you consider the fact that the very first levels have steep percentual increases in terms of exp required for the next level, a smaller party reach a slightly higher level than a corresponding full sized really, really fast, despite the effects from rule #1.

However, even if you're scoring sextuple exp per character (IWD2: One char vs 6 char maximum), you only need to reach a mere five levels above the expected before the "power of square root of two" multiplier slows you to a crawl corresponding to a full sized party for your kills. Thus, you'll only ever want to stay at higher than expected level when a vast majority of your exp comes from quests, or else!

Rule #4: Adding characters as you go
The final puzzle piece that I've been missing so far. I've theorized about it, confirmed it in actual gameplay, but still missed WHY it was happening. It has very little to do with "doing certain important parts as lower level party" as I previously thought, but rather how the accrued exp total and the party's average level relate to each other on a mathematical level.

Consider the function for a character's LEVEL, given the amount of EXP (in thousands):
Lvl = [(sqrt(8*exp+1)) +1] / 2 - ECL_PENALTY

Adding characters as you go will introduce a character-specific offset to the exp value, based on the point of time when this specific new character was added, resulting in:
Lvl = [(sqrt (8*[exp-offset]+1)) +1] / 2 - ECL_PENALTY

Taking an average of such values to calculate the party's average level will quickly reduce the effect of the greater exp totals for your first character(s) to insignifigance as you keep on increasing your party's exp total. This has the effect of making any early exp gained for them "free" for all practical purposes.

However, just as I've also proven in previous playthroughs, there's a practical limit in how long you can keep the lower character count before it starts to eat into the profits. Specifically, even a solo character should NEVER consider anything past +5 levels above the expected, unless going on nothing but quest experience, and anything past a duo is sawing its own branch already at +2. Conversely, any longer patch of consecutive exp from killing stuff should be a priority target when your average level is low, ie. right after adding new party member(s).

So, what's the "best" way of doing things in IWD2, then?
Just as I've proven before, you really get double benefits by starting with a smaller party. Just as long as you can get through Prologue with a smaller party, you'll get all the important stuff and reach the point of smooth sailing faster. The only difference is that now I have the required maths backing it up.

 

I could follow almost nothing of that, since I'm an idiot who hasn't practiced math in over half a decade, but I did catch 1 tidbit:
Are you saying that a party with a small ECL penalty will actually get an ECL boost to XP, due to that anomaly?! How far does this continue? My favourite party had 3 drow and 1 DG. What would my ECL have been? +1, +2, 0, -1?!

Kudos on the research, i just wish I was able to appreciate it fully.

 

This is impressive, Sir Rechet. A lot of the empirical stuff is starting to make more sense now.

 

My favorite party has an Aasimar, no one else has any ECL penalty, it does seem to be moving quite fast.

 

Don't forget to mention the impact of having characters with xp penalties from multiclassing...

 

Yes. A party with nothing but zero ECL characters is actually trumped by one that has one with ECL 1+.

For the exact same amount of total exp, you could have a bunch of characters, all at level X, or a (slightly smaller) bunch of characters at level X and one with level X-1. While this DOES seem to favor the all-zero ecl party, the second party is a whole level below the first in terms of party's average level. Thus, the second party will get the exp rewards for being at level X-1, even if a majority of them are actually at level X.

This, combined with rule #1, turns a seemingly "behind by one level" situation into an "ahead by party size minus one levels" one. In due time, not immediately, of course.

For the 3 Drow + 1 DG party, you get a calculated ECL of +2.25. However, since you have a remainder, you minus one, ending up at +1.25 "effective" ECL, ie. +1, the same you would get with an all-Aasimar party!

Rule #2, subsection A: Multiclassing penalties.
MC penalties function much in the same way as ECL penalties do, with one major exception: You end up at lower average level, giving you an exp boost, but the boost is then reduced back by the MC penalty. Considering that the exp req deltas reach single digit percentages past level 22, that's two or more whole levels dragging behind for every 20% of MC penalty past that point. The "single level behind" breakpoint is at absolute level 12.

 

Well, it's actually 2 humans, 3 drow and a DG. Does that make much difference?

 

That's actually a bit better as you get a total of 9 levels of ECL penalty in a party of six, ie. +1.5, which translates into +0.5 effective ECL due to the remainder.

Please note that unless all of the party's characters start simultaneously, the total ECL penalty is only a rough guideline for the expected level average. It's a razor's edge balancing between pushing further into the "free" exp zone and getting penalized for being at too high level due to having too few characters doing the pushing. Generally though, adding a high ECL character later on will let the rest of the party push to slightly higher level, while a low ECL character gets into the usable level range faster. It's all about what levels you wish to reach, not so much in the absolute sense, but in relation to the other party members.

 

I knew my party rocked!
Thanks.