Back to SeanKReynolds.com home

A lot of people don't like the Level Adjustment (LA) system for powerful races (such as drow) and think we should use an XP-penalty system, where each powerful race takes a penalty to XP awards to keep them a little bit behind the humans, dwarves, and so on. I helped come up with the current LA system so I'm partial to in and I know it inside and out, and I think it's a reasonable way of balancing the races. It has its flaws, but it works much better than an XP-penalty system, which is seriously flawed to the point of rewarding you for playing a powerful race. This article takes a look at the data behind the XP-penalty idea and shows why it doesn't accomplish what its proponents want it to do.

For this data, we're assuming a robot DM is awarding 100 XP per character play session, modifying each individual XP award by the XP penalty of that character. So each session, the robot DM awards the no-penalty character (let's say, a human) a full 100 XP, the 10%-penalty character 90 XP, the 20%-penalty character 80 XP, and the 30%-penalty character 70 XP.

This example ignores that as characters go up in level, they tend to gain more XP per session because they're facing tougher opponents (I could build a table for that sort of situation, but we'd end up with the same overall result and the math wouldn't be as obvious to the reader).

This table examines the XP totals and character level of each character after each session's XP is awarded (the first line is the zero-value, before any sessions are played).

(If you like, you can skip on past this table to the discussion below, the table is really here just to prove I did the proper calculations to reach my conclusion. You don't need to look at the data in this big table to get what I'm talking about, I explain it all below.)

XP Total	Human	Human's level	10% penalty race's XP total	10% penalty race's level	is the 10% penalty race behind a level?	20% penalty race's XP total	20% penalty race's level	is the 20% penalty race behind a level?	30% penalty race's XP total	30% penalty race's level	is the 30% penalty race behind a level?
0	0	1	0	1		0	1		0	1
100	100	1	90	1		80	1		70	1
200	200	1	180	1		160	1		140	1
300	300	1	270	1		240	1		210	1
400	400	1	360	1		320	1		280	1
500	500	1	450	1		400	1		350	1
600	600	1	540	1		480	1		420	1
700	700	1	630	1		560	1		490	1
800	800	1	720	1		640	1		560	1
900	900	1	810	1		720	1		630	1
1000	1000	2	900	1	yes	800	1	yes	700	1	yes
1100	1100	2	990	1	yes	880	1	yes	770	1	yes
1200	1200	2	1080	2		960	1	yes	840	1	yes
1300	1300	2	1170	2		1040	2		910	1	yes
1400	1400	2	1260	2		1120	2		980	1	yes
1500	1500	2	1350	2		1200	2		1050	2
1600	1600	2	1440	2		1280	2		1120	2
1700	1700	2	1530	2		1360	2		1190	2
1800	1800	2	1620	2		1440	2		1260	2
1900	1900	2	1710	2		1520	2		1330	2
2000	2000	2	1800	2		1600	2		1400	2
2100	2100	2	1890	2		1680	2		1470	2
2200	2200	2	1980	2		1760	2		1540	2
2300	2300	2	2070	2		1840	2		1610	2
2400	2400	2	2160	2		1920	2		1680	2
2500	2500	2	2250	2		2000	2		1750	2
2600	2600	2	2340	2		2080	2		1820	2
2700	2700	2	2430	2		2160	2		1890	2
2800	2800	2	2520	2		2240	2		1960	2
2900	2900	2	2610	2		2320	2		2030	2
3000	3000	3	2700	2	yes	2400	2	yes	2100	2	yes
3100	3100	3	2790	2	yes	2480	2	yes	2170	2	yes
3200	3200	3	2880	2	yes	2560	2	yes	2240	2	yes
3300	3300	3	2970	2	yes	2640	2	yes	2310	2	yes
3400	3400	3	3060	3		2720	2	yes	2380	2	yes
3500	3500	3	3150	3		2800	2	yes	2450	2	yes
3600	3600	3	3240	3		2880	2	yes	2520	2	yes
3700	3700	3	3330	3		2960	2	yes	2590	2	yes
3800	3800	3	3420	3		3040	3		2660	2	yes
3900	3900	3	3510	3		3120	3		2730	2	yes
4000	4000	3	3600	3		3200	3		2800	2	yes
4100	4100	3	3690	3		3280	3		2870	2	yes
4200	4200	3	3780	3		3360	3		2940	2	yes
4300	4300	3	3870	3		3440	3		3010	3
4400	4400	3	3960	3		3520	3		3080	3
4500	4500	3	4050	3		3600	3		3150	3
4600	4600	3	4140	3		3680	3		3220	3
4700	4700	3	4230	3		3760	3		3290	3
4800	4800	3	4320	3		3840	3		3360	3
4900	4900	3	4410	3		3920	3		3430	3
5000	5000	3	4500	3		4000	3		3500	3
5100	5100	3	4590	3		4080	3		3570	3
5200	5200	3	4680	3		4160	3		3640	3
5300	5300	3	4770	3		4240	3		3710	3
5400	5400	3	4860	3		4320	3		3780	3
5500	5500	3	4950	3		4400	3		3850	3
5600	5600	3	5040	3		4480	3		3920	3
5700	5700	3	5130	3		4560	3		3990	3
5800	5800	3	5220	3		4640	3		4060	3
5900	5900	3	5310	3		4720	3		4130	3
6000	6000	4	5400	3	yes	4800	3	yes	4200	3	yes
6100	6100	4	5490	3	yes	4880	3	yes	4270	3	yes
6200	6200	4	5580	3	yes	4960	3	yes	4340	3	yes
6300	6300	4	5670	3	yes	5040	3	yes	4410	3	yes
6400	6400	4	5760	3	yes	5120	3	yes	4480	3	yes
6500	6500	4	5850	3	yes	5200	3	yes	4550	3	yes
6600	6600	4	5940	3	yes	5280	3	yes	4620	3	yes
6700	6700	4	6030	4		5360	3	yes	4690	3	yes
6800	6800	4	6120	4		5440	3	yes	4760	3	yes
6900	6900	4	6210	4		5520	3	yes	4830	3	yes
7000	7000	4	6300	4		5600	3	yes	4900	3	yes
7100	7100	4	6390	4		5680	3	yes	4970	3	yes
7200	7200	4	6480	4		5760	3	yes	5040	3	yes
7300	7300	4	6570	4		5840	3	yes	5110	3	yes
7400	7400	4	6660	4		5920	3	yes	5180	3	yes
7500	7500	4	6750	4		6000	4		5250	3	yes
7600	7600	4	6840	4		6080	4		5320	3	yes
7700	7700	4	6930	4		6160	4		5390	3	yes
7800	7800	4	7020	4		6240	4		5460	3	yes
7900	7900	4	7110	4		6320	4		5530	3	yes
8000	8000	4	7200	4		6400	4		5600	3	yes
8100	8100	4	7290	4		6480	4		5670	3	yes
8200	8200	4	7380	4		6560	4		5740	3	yes
8300	8300	4	7470	4		6640	4		5810	3	yes
8400	8400	4	7560	4		6720	4		5880	3	yes
8500	8500	4	7650	4		6800	4		5950	3	yes
8600	8600	4	7740	4		6880	4		6020	4
8700	8700	4	7830	4		6960	4		6090	4
8800	8800	4	7920	4		7040	4		6160	4
8900	8900	4	8010	4		7120	4		6230	4
9000	9000	4	8100	4		7200	4		6300	4
9100	9100	4	8190	4		7280	4		6370	4
9200	9200	4	8280	4		7360	4		6440	4
9300	9300	4	8370	4		7440	4		6510	4
9400	9400	4	8460	4		7520	4		6580	4
9500	9500	4	8550	4		7600	4		6650	4
9600	9600	4	8640	4		7680	4		6720	4
9700	9700	4	8730	4		7760	4		6790	4
9800	9800	4	8820	4		7840	4		6860	4
9900	9900	4	8910	4		7920	4		6930	4
10000	10000	5	9000	4	yes	8000	4	yes	7000	4	yes

Now that we have 101 data points for all four example characters, we can analyze how they compare to each other.

Looking at each session, we can figure out how many times each character is one level behind the human character, and (knowing how many game sessions were played) can determine a percentage of time that the character spent one level behind the human character.

	10% penalty	20% penalty	30% penalty
% Behind Levels 1-5*	86.1	73.3	55.4

* Technically this header means "For all of first level and up until the breakpoint where the human character reaches a new level," so it covers XP values from 0 to 10,000, which is level 1 to the very beginning of level 5 for the human character (at 10,000, all of the XP-penalty characters are less than 5th level).

So that seems reasonable, right? The 10%-penalty character is one level behind about 86% of the time for the time it takes the human character to reach level 5, the 20%-penalty character is one level behind about 73% of the time, and the 30%-penalty character is one level behind about 55% of the time. So the mildly-penalized character (10% penalty) is almost always a level behind the human character, and is only the same level less than ten percent of the time.

But that doesn't give a good view of the situation -- the single line of data in that very short table actually hides some important information. If you take a look at the data for the lower levels, you find that the characters spend much more time at the same character level.

	10% penalty	20% penalty	30% penalty
% Behind Levels 1-5*	86.1	73.3	55.4
% Behind Levels 1-4*	88.5	80.3	68.9
% Behind Levels 1-3*	90.3	87.1	80.6
% Behind Level 1-2*	90.9	90.9	90.9

Note in particular that last line, which represents the first ten gaming sessions (enough to get the human character up to 1,000 XP and the start of 2nd-level): the only time the XP-penalty characters are a level behind the human character is right when the human character reaches 2nd-level. That means that for the first nine game sessions, the four characters are exactly the same character level ... which means that for those first nine game sessions, the XP-penalty characters are more powerful than the human character because they're the same character level and they get all of the racial goodies that justify their XP penalty in the first place!

The XP penalty isn't penalizing them at low levels where their racial powers should be making the most difference. It's only after 100 game sessions that the XP penalty starts to cause them to lag behind on a regular basis (always just when the human character has gained a level), and they spend the rest of their time at the same character level as the human character, with all of their racial goodies. So, looking at the % Behind Levels 1-5 data line again, realize that the numbers mean that the "penalized" character is simply better than the human character (same character level, plus racial goodies) 86% of the time for a 10% penalty race, 73% of the time for a 20% penalty race, and 55% of the time for a 30% penalty race ... and when they're not simply better (the balance of those one hundred percentage points) the "penalized" characters are exactly the same power level as the human character (because the point of the penalty is to lower their power to that of the human character, so when the penalty "works" they must be the same power level, yes?).

It doesn't matter what penalty you choose; you could give a race a 50% XP penalty (say, for a half-dragon) and they'd still be the same character level (plus racial goodies) as the human 20% of the time for the time it takes the human to reach level 5. Even a 90% XP penalty (say, a half-celestial half-dragon) would still be the same character level (plus racial goodies) as the human almost 10% of the time for the time it takes the human to reach level 5.

XP penalties make little to no difference at low character levels when the differences between races should be the strongest.

The LA system was build so you could have a human, a dwarf, a drow, and an aasimar all in the same party at a comparable power level (particularly at low character levels where differences in ECL have a much greater effect) so that it's fair to all the players (nobody gets to play an über-character while someone else has to play a weenie character). The XP-penalty system throws that out the window because it encourages you to play a more powerful race because at worst your penalty makes you just as powerful as the human and at best you're stronger than the human.

From the "Behind Levels" table you can see that the more game sessions you look at (and thus the higher number of number of character levels the human attaints) the more often the XP-penalty characters are going to fall behind the human character. In other words, at higher levels, the XP-penalty characters are significantly weaker than their human counterparts, even though you'd think that a human Clr20 and an aasimar Clr20 would be almost indistinguishable in terms of power. (Compare to the XP-penalty system, where when the human reaches level 20, the 10% penalty race has just reached level 19, the 20% penalty race is almost at level 18, and the 30% penalty race is almost at 17th ... much like the LA+1, +2, and +3 that everyone hates so much, except the XP penalty system rewards you at low level and penalizes you at high level ... oddly like 1E/2E AD&D.)

And that's why LA, for all its flaws, is a better system than an XP penalty--because LA actually does what it's supposed to (penalize powerful races at low levels rather than reward them).