How smart is a 3 INT in d20?

[Dogbrain]

If you use the 3d6 generation method, 1 in 36 characters will have a 3 in a stat. Nobody uses the 3d6 method, but it's there.

One in thirty-six?
One in thirty-six?

In what reality would this be the case?

You realize, of course, starting out with such a trivially disprovable statement (a 3 score in 3d6 is 1 in 216 and this can be mathematically demonstrated with great rigor), means that nothing else you wrote can be trusted.

 

[Dogbrain]

Also if someone can perform very simple taks with supervision, we need to define simple. Tie a shoe lace? Simple maths? Cut paper and remain unharmed?

Approximately equivalent to a normal human six-year-old. The normal human six-year-olds that I've met have remarkably fluency in their native language.

"Me go to see fancy face man" would be a D&D Int of 1 or lesss.

 

[Dogbrain]

I did that imitation too harshley, perhaps I should have said that they would make obvious gramatical mistakes.

Or spelling errors?

 

[drothgery]

Still, the game assumes that you can still go adventuring with a stat of 3, no matter which one.

If you use the 3d6 generation method, 1 in 36 characters will have a 3 in a stat. Nobody uses the 3d6 method, but it's there.

So, a 3 in a stat is quite low, but would not forbid you to function as an adventurer, so I think we should refrain to assume that a 3 in intelligence will make you a complete 'tard.

A 2 in a stat would make you sub-human, thus a Con 2 person could have a chronical disease, a Wis 2 person could be totally insane, a Int 2 person could have some chronical mental affliction (not being able to differentiate friend from foe, thus unfit for adventuring).

A 3 sucks, but you're still functional. A 3 int character can communicate and recognise a foe.

Except for the bad math (as others have pointed out, the odds of getting a 3 on 3d6 are 1 in 216, not 1 in 36, and combinatronics are funny -- I'm pretty sure the odds of at least one 3 in 6 individual 3d6 events is a lot lower than 1 in 36), you're pretty much correct.

3-18 covers three standard deviations of the human population at age 16 or so (i.e. when you're a real 1st level character, and not suffering any penalities for being a child). There are extremely rare humans who are outside this range to start with, though in D&D worlds this is normally caused by magic (or the super-prodigy who's 4th level by age 16).

 

[Anabstercorian]

Dogbrain, don't live up to your name so. He said those were the odds of having a 3 in a stat, not the intelligence stat. I'm not sure if his numbers are exactly correct, but if 1/216 are the odds of ONE stat being 3, then it seems plausible that 1/36 are the odds of any of six stats being 3.

 

[Kalanyr]

MENSA? That means you used IQ. At least, when I posted, I had the good sense to use "g" instead of merely IQ. Please demonstrate that IQ is an accurate measure of overall intelligence or at least it is a better measure than is "g".

Would you like to show where I said it was a better representation than a g-value ? It'll be pretty hard since I never mentioned a g value. And since I had the distribution for IQ done , but have no g-value formulaes handy I provided what I had, nice to be criticised for it, thanks so much.

 

[Kalanyr]

One in thirty-six?
One in thirty-six?

In what reality would this be the case?

You realize, of course, starting out with such a trivially disprovable statement (a 3 score in 3d6 is 1 in 216 and this can be mathematically demonstrated with great rigor), means that nothing else you wrote can be trusted.

Mm, and again you're wrong, 1 stat in every 216 will be a 3, however everyone has 6 stats. So in a population 36 people you have SURPRISE 216 stats! one of which will be a 3, statistically speaking.

 

[Al]

Strictly speaking, of course, the chances of getting a 3 in any particular stat is not simply equal to (1/216)x6, since the logical extension of this indicates that if you got any 36 people the odds of getting a 3 in some stat is 1 (certain). Yet some permutations of selecting those 36 would doubtless produce combinations whereby not a single 3 is encountered.

Rather, the real probability is the 1-(215/216)^6. Which equates to 2.746% (3dps), or about 1 in 36.4



PS Dogbrain, using IQ as a model for intelligence, 3 Int = 60 IQ = mental age of nine, not six.

 

[Asmor]

Strictly speaking, of course, the chances of getting a 3 in any particular stat is not simply equal to (1/216)x6, since the logical extension of this indicates that if you got any 36 people the odds of getting a 3 in some stat is 1 (certain). Yet some permutations of selecting those 36 would doubtless produce combinations whereby not a single 3 is encountered.

Rather, the real probability is the 1-(215/216)^6. Which equates to 2.746% (3dps), or about 1 in 36.4



PS Dogbrain, using IQ as a model for intelligence, 3 Int = 60 IQ = mental age of nine, not six.

Wait, so how does adding .4 people suddenly make it certain?

If I flip a coin two times, both times I've got a 50% chance of getting either heads of tails. Does that mean I have a 100% chance of getting a tails? No. Does it change the fact that I still have a 50% chance each time? No.

There's no such thing as certain when you're talking about probabilities, however the more people involved the closer it gets to the actual percentage.

If every person on earth rolled 3d6 6 times, the number of 3s rolled would be statistically indistinguishable from one for every 36 people.

 

[Mercule]

Wow, so now we're arguing about statistal probability.

A probability of 1/36 does not mean that you are guaranteed of finding that 1 if you gather any 36 subjects at random. What it means is that, as your sample size grows, you should find it reflects a 1/36 portion of subjects meeting the selection criteria.

In easier terms, if you've got a lot of people, you should be able to put them in groups of 36 people, each group having on person with a stat that is a 3.

The smaller the total sample size is, the less likely the sample numbers will match the statistical probability. When we're talking all the people in a given setting, I think that's a large enough sample size to assume the math works.