In my previous article, I described the impressive way in which the level progression in fourth edition Dungeons & Dragons has been balanced such that character abilities retain more or less the same balance relative to each other as the characters gain levels. In general, attack, defense, damage, and skills all go up at the right rate to remain generally in proportion as the characters gain levels, and prevent high-level characters from becoming progressively more polarized. However, there is one major discrepancy in this system - the fact that character statistics do not go up at the same rate. This creates a number of discrepancies as the characters gain levels.
1. Each character build has certain primary and secondary statistics needed for the class. Many of the builds, and the suggestions for those bills, imply that you would want to concentrate on 3 statistics. However, this makes no sense as far as level progression is concerned because you can only increase 2 of your statistics at the maximum rate. Of course, it is possible to divide your statistic increases between 3 statistics. But the problem with this is that you will fall farther and farther behind a character who divides his statistic increases between only 2 statistics. Primary statistics are used to calculate your attack bonus, so it is absolute essential that they be increased at a maximum rate to comply with the mathematics of the game. Otherwise, you will become less accurate than the other characters as you gain levels. This is why the idea of a character class with 2 primary statistics and a secondary statistic, like a paladin, just doesn't work. It is relatively feasible (if not optimal) to make such a character at low levels, but as the character increases in levels, one of those statistics has to be left behind or the character has to become less and less powerful compared to the other characters. Some secondary statistics are not quite as critical, affecting only things like damage or healing or other things that are not as sensitive to single points. In this case, it would be more feasible to have a character with one primary statistic that splits his points between 2 secondary statistics. However, this isn't very tempting and still results in a character who, at high levels, won't be quite as appealing as his 2 statistic counterparts.
Once you realize this is the case, you can banish the idea of 3-statistic characters from your mind and concentrate on 2-statistic characters, which work properly. However, it remains peculiar that the game tempts you to build multi-statistic characters because they appear to work okay low-level, but you have to rethink your design once you discover they don't quite translate properly at higher levels.
2. The most significant level progression imbalance is with the skill system. Fourth edition gets rid of skill points in favor of fixed bonuses for training in the skill and universal bonuses to all skills according to the level of the character. This design promises to prevent huge discrepancies in skills between the characters at high levels, and in general provide the benefits that I described in my previous blog article. However, the fact that the statistics increase at an uneven rate disrupts the symmetry. At low level, the skill system allows just about everyone to participate in skill checks whether they are skilled or unskilled. And a character who takes skill training in a skill governed by a statistic is not especially good in, still has a pretty good chance of having the best skill roll in the party if no one else took that skill. But at high levels, each character will have 2 statistics that are much higher than the others, and the skills controlled by the statistics will start to get much bigger bonuses. Now the characters who are trained in skills that match their statistics far outshine every other character. Not only does this mean that the balance of the game changes at high levels, it has effect that which class/build you have taken has more effect on your skills and which skills you are trained in. This is an effect I always disliked in games, where the streetwise fighter is forced to be less streetwise than the otherworldly fey warlock, regardless of the background stories and options chosen in character generation.
3. There are some other small effects which diverge in ways similar to the first 2 points. For instance, characters with Dex or Con bonuses gain more of a relative initiative / healing surge bonus compared to the rest of the party as they gain levels. So an infernal warlock may have fewer healing surges than a paladin at low level, but more at high level. This isn't really fatal, but it doesn't quite fit the symmetry of the uniform level progression idea. Also, there are some situations where characters make attacks which don't use primary statistics, such as longbow attacks from melee characters. At low levels these may be at least somewhat effective, but at high levels that will become less and less effective compared to the main attack to the point of uselessness.
So the next question is, what would you do if you decided you wanted to fix this imbalance? Well, a very straightforward way to do so would be to have every statistic increase every 4 levels instead of just 2 statistics increasing every 4 levels. This would have a significant advantage of fixing the problems listed above. I should also note that it would make all sorts of unusual multi-class combinations possible and fun. However, it would also carry some potential disadvantages:
1. As mentioned above, primary and secondary statistics often have effects where the same amount of bonus is just as good at high-level is a low level. I think a good word to capture this is that some bonuses are "logarithmic". A 2-point bonus to attack or defense is just as good for a high-level character with a +30 attack as it is for a local character with a +4 attack. Other benefits are non-logarithmic. A +2 damage bonus for a high-level character who does about 22 points of damage is much less effective than the same bonus for a low-level character who does about 8 points of damage.
If every statistic increased at the same rate, then for non-logarithmic abilities controlled by secondary statistics, the actual value of your statistic would mean less and less as you gain level. That is, at low level, a barbarian with a relatively high constitution might gain 3 temporary hit points when he defeated a foe, while barbarian who had basically ignored constitution might only gain 1 temporary hit point, a big difference. But at high-level, it might be the difference between 8 temporary hit points and 6 temporary hit points, not a big deal. So in this case, having all of the statistics increase at the same rate would cause some abilities to scale less well with level. I think this would be the major disadvantage of allowing every statistic to increase.
Although this definitely causes some problems with the scaling, some of the other effects it has on the game might be good or bad. It would mean that at high-level, classes with 2 secondary statistics would now find it very practical to take powers relying on their lesser secondary statistic. My general impression is that such powers are normally much less tempting to take then the flavor of the game would indicate (in particular, they seem to imply that you should take some of these powers when they describe the class builds, but I find it quite difficult to ever want to take a power which is based on a secondary statistic that I am not improving, unless the power is really broken to begin with). So feeling more free to take these powers might be a good thing. Or maybe not, I can't really say.
2. With constitution increasing at a faster rate, high-level parties would have noticeably more healing surges than low-level parties. Since they have noticeably more healing powers, I don't know whether this is a good or bad thing.
3. Many feats have prerequisites that prevent certain class builds from getting those feats. With every statistic increasing, pretty much every build would eventually be able to get most feats. If those statistic minimums were carefully designed and considered as part of the class balance (it isn't clear whether this is true), this might mean the class balance is disrupted somewhat. For instance, normally only fighters who invest in high Dex scores can take scale armor specialization, but now even Con-based axe fighters could do so.
4. I would expect that a major actual reason that not all statistics are made to increase with level is a combination of inertia/tradition (only one statistic increased with level in third edition, and none before that), and the idea that it is fun getting to choose where your statistic points to go with levels. It is true that there is a small amount of discretion in the way you allocate your statistic bonuses, it is sometimes possible to make little trade-offs, not increasing a secondary statistic in order to raise constitution or a statistic necessary to qualify for a feat. But in general, the game balance so strongly demands that you increase the 2 statistics associated with your class build, that I haven't found getting to choose where your statistic bonuses go to be a particularly interesting process in practice.