Article Index

Sorry, but some mathematics is coming. And this is not only about dice. Simply because 3d6 as a way of rolling an ordinary person's attributes is a little odd. Initially it seems reasonable enough. Then we realize that a village of just under five hundred would generally contain two people with Strength 18, and another six with 17. If a character has ten attributes, the chance of all of them being below 14 would be around 17%. Player characters who want to stand out from the crowd could use better attributes than that. Inevitably we move somewhat toward powergaming. A knight who would ordinarily fall outside the ten strongest people even in such a small village is not necessarily a dream character.

A player reasonably decides 14 is not necessarily enough. In a village of five hundred, eighty-one people are generally at least that strong. The character would not win many arm-wrestling contests. If a player character wants to stand out in this environment, they must seek values granting substantial advantages. They powergame because they want to be just a little better than an ordinary person. And it is rather funny when someone in a Facebook debate is surprised I do not raise ordinary people to player-character levels. If I did, players would want to stand out from those ordinary people. They could only do so by exceeding that raised level, the party's average strength. Combined with more XP and rapid advancement from stronger opponents, this soon leads to the level-47 necromancer munchkin. 

We began assembling the ÚTK NPC example collection. One of the biggest obstacles was the 125 distributable points. Why 125? Because we asked. Had we not asked, we would have said that for races of this level, 135 was the rules-compliant solution. With the average around 10—9 to 11—ten attributes make an “all average” character cost 100 points. The remaining twenty-five, distributed as 8+8+5+4, produce two 18s, a 15, and a 14. Does that fit the desire that “this is fantasy, so most supporting characters should stand out in something too”? We soon decided the young man's father owned the inn; by the time he becomes innkeeper, age will impose a five-point deduction. Is it surprising that someone wanting to stand out a little from this starts powergaming? Or, when the system constrains them, that they examine their remaining options?

Only Three Attributes at 18...

Distributing 135 points could produce three 18s, a 17, and a 14. For an innkeeper, peasant, or waitress, say. The result? Many players will powergame. Many people see this as a problem. They try to remedy powergaming at events through the system. Its numerous restrictions force people to abandon many character concepts, iconic abilities seen in novels, and much else. Many players lose their motivation to use that system, while the powergamer receives a small challenge: opponents face the same restrictions, and they will find another loophole. D&D fourth edition was locked down so successfully that many forums questioned whether it was still a roleplaying game, yet powergaming continued. Instead, powergaming eventually makes higher attributes widespread; when we choose only among these, the system's resolution shrinks.

What happens if we reverse the situation? If characters do not need outrageous abilities to stand out? If we assume characters choose tasks they can accomplish with considerable confidence, so players gain almost nothing from powergaming, earn approximately zero points for it at events and competitions, and elsewhere the group decides whether it wants to play with the powergamer... If, alongside all this, boundaries are loose and many colorful, interesting characters exist? And if the system's reward mechanism favors good roleplaying and development instead of characters who start out overpowered? I think much of the problem is then solved. 

But this requires good reference points for attributes, and both players and game masters need sound guidance for human-scale attributes. It is easy here to discuss standard deviation and median alongside the mean. Why distinguish median from mean? Because selection may affect them differently. Suppose roughly 23–26% of people had Health 10: the median would remain 10 if, say, 27% were below 10, 24% at 10, and 49% above it. It would remain so in the reverse situation too. Yet different attributes would make someone stand out. Suppose the mean expected attribute at birth were 10... But far more people with attributes below 10 died in war or of disease... We can then suspect that truly low Health will be very rare. The mean will exceed the median. Evolution, however, may raise both together over the long term. 

In a darker world where people with excessive natural gifts are likely to be executed because they could threaten the ruler? The reverse situation could arise. Clarifying this is therefore part of the world description, not a core system. Standard deviation helps determine what proportion of characters have above-average values. What if, instead of the roughly 3 standard deviation seen with 3d6, we seek around 1.5, assuming a normal distribution? Then a village of five hundred would generally contain about six people—5.85—with Strength 14 or higher. Strength 16, however, would suffice for the title of strongest person in a small town of eight thousand. 

chartBut if selection raises the mean by 0.5–1 point while the median stays... Then many NPCs could stand out a little. All right, “everyone stands out” is impossible, since nobody would stand out, but generally everyone would be talented to some extent. Instead of natural selection or Spartan methods, divine will, the world's fantasy nature, and so forth can produce similar effects. If you also accept one or two attributes of 9 for many supporting characters—hardly something striking—and distribute, say, 105 points rather than 125, or at most 110, interesting NPCs are still possible. Using more of the attribute scale immediately lets you emphasize differences between NPCs better. But look a little to the left! This odd little graph appeared in an earlier article. There we discussed attribute checks too. And the importance of having a reasonable chance on many checks even with the declared average attribute of 10. Partly because players will not accept complete hopelessness on many realistically achievable tasks. They will begin longing for far-above-average attributes that grant bonuses, developing their characters to remedy the problem; and if their attributes are “bland and hopeless” too, they will stand out with some ancient-magic race, overpowered class, or school hunted down online. 

Abandoning rolls under only the portion above 10 and using the full attribute increases the system's resolution and solves numerous problems. But what should we roll? Some point to a twenty-sided die, which must be bought. An immediate alternative is rolling two ten-sided dice. The chart shows the chance of success. What if we roll 3d6 under the attribute? Again, the difference is standard deviation. Also, if the combat system and percentage skills did not require ten-sided dice... But they do. This change in standard deviation would make each single-point attribute difference much more pronounced. 

That can be decided too. As can how much influence chance receives versus reliable knowledge. Or why so many players are attached to ten-sided dice. But let us accept that we have a good solution for attribute checks, perhaps the proposed one. All that remains is deciding how common an attribute of 13 or 17 should be in the world. How should these frequencies be understood at all? And surely nobody will be surprised when I say illustrating them is an even greater problem. We will return to that too.