## A Really Quick Nitpick and a Discussion of Skill versus Talent

January 13, 2010 by Bilsybub

I know I pick on Tobold sometimes, but in this case I agree with the gist of his point. He’s speaks in this post about Gearscore and Blizzard’s own similar system, built into the Dungeon Finder. To get my feelings on Gear Score out of the way: it annoys me. Greatly.

Anyway, my nitpick is with this assertion: “Skill in playing World of Warcraft, like many naturally occuring things, is distributed over a bell curve, a Gaussian distribution.” I’ve a couple complaints with this. First, there’s a bit of a conflation of the concepts of “skill” and “talent”. For our purposes here, talent is in-born potential, unrealized or realized, and skill is existing ability. So, a talented player is someone who has the natural ability to perform very well, regardless of whether they actually do, while a skilled player is actually able to perform well. I know that’s not wholly consistent with word usage, but my meaning ought to be clear: there’s some difference between some who could perform well, due to inborn potential, and someone who does perform well. It’s generally assumed that it is talent which is distributed along a bell curve, not skill. I’m generally skeptical of any such broad generalization, myself, due to the ease with which counterfactuals can be found. For instance, I assert that perfect talent at Tic Tac Toe exists in the majority of people (that is, most people have the potential to learn how to make every game of Tic Tac Toe either a win or a draw). Whether people have learned that or not, the talent would not follow a Gaussian distribution in that case.

But let’s assume talent does indeed fit some form of Gaussian distribution. Even assuming this, skill itself is learned. When speaking of learning, we talk about the “Learning Curve”, which is an asymptotic curve where time invested initially yields rapid gains in facility then tapers off such that additional time invested yields diminishing improvements (note that the slope of the curve is always positive). We can assume (this simplifies this discussion almost to the point of absurdity, but most assumptions do) that people move along the curve at precisely the same rate. It seems reasonable to think that someones inborn skill functions as a sort of multiplier on this curve. Most people get a kind of standard curve, while others get a higher potential ramp up and higher ceiling on their potential capabilities, and still others have a very short jump before they simply can’t get much better, regardless of time/effort invested. This has a very similar effect to simply putting a ceiling on the learning curve, but maintains the ability to continue to marginally improve (the positive slope of the curve).

So take a normal distribution of multipliers, from 0 to 1 with 50 as the mean, and take a random distribution of start times (so people are at random places along their respective curves), You’ll find that a very large portion of people (at a guess it looks like a supermajority, or 2 standard deviations) are actually below the average maximum skill level.

### Like this:

Like Loading...

*Related*

## A Really Quick Nitpick and a Discussion of Skill versus Talent

January 13, 2010 by Bilsybub

I know I pick on Tobold sometimes, but in this case I agree with the gist of his point. He’s speaks in this post about Gearscore and Blizzard’s own similar system, built into the Dungeon Finder. To get my feelings on Gear Score out of the way: it annoys me. Greatly.

Anyway, my nitpick is with this assertion: “Skill in playing World of Warcraft, like many naturally occuring things, is distributed over a bell curve, a Gaussian distribution.” I’ve a couple complaints with this. First, there’s a bit of a conflation of the concepts of “skill” and “talent”. For our purposes here, talent is in-born potential, unrealized or realized, and skill is existing ability. So, a talented player is someone who has the natural ability to perform very well, regardless of whether they actually do, while a skilled player is actually able to perform well. I know that’s not wholly consistent with word usage, but my meaning ought to be clear: there’s some difference between some who could perform well, due to inborn potential, and someone who does perform well. It’s generally assumed that it is talent which is distributed along a bell curve, not skill. I’m generally skeptical of any such broad generalization, myself, due to the ease with which counterfactuals can be found. For instance, I assert that perfect talent at Tic Tac Toe exists in the majority of people (that is, most people have the potential to learn how to make every game of Tic Tac Toe either a win or a draw). Whether people have learned that or not, the talent would not follow a Gaussian distribution in that case.

But let’s assume talent does indeed fit some form of Gaussian distribution. Even assuming this, skill itself is learned. When speaking of learning, we talk about the “Learning Curve”, which is an asymptotic curve where time invested initially yields rapid gains in facility then tapers off such that additional time invested yields diminishing improvements (note that the slope of the curve is always positive). We can assume (this simplifies this discussion almost to the point of absurdity, but most assumptions do) that people move along the curve at precisely the same rate. It seems reasonable to think that someones inborn skill functions as a sort of multiplier on this curve. Most people get a kind of standard curve, while others get a higher potential ramp up and higher ceiling on their potential capabilities, and still others have a very short jump before they simply can’t get much better, regardless of time/effort invested. This has a very similar effect to simply putting a ceiling on the learning curve, but maintains the ability to continue to marginally improve (the positive slope of the curve).

So take a normal distribution of multipliers, from 0 to 1 with 50 as the mean, and take a random distribution of start times (so people are at random places along their respective curves), You’ll find that a very large portion of people (at a guess it looks like a supermajority, or 2 standard deviations) are actually below the average maximum skill level.

## Share this:

## Like this:

RelatedPosted in Commentary | Leave a Comment

Comments RSS