D&D 5E Point Buy vs Rolling for Stats

[Maxperson]

You have dismissed the impact as unimportant:

The impact of a +1 in non-prime stats is minor at best. In prime stats, the PC isn't going to have a 12, so it's not a +1.

Seems to have done both. The higher stat version had twice the chance of survival in a difficult situation. That's a difference, even a quantified one, and hardly minor

How does a comparison that won't happen in a real game showing anything of relevance? In a real game it will be a dwarven fighter and a halfing wizard, or a tiefling rogue and an elven sorcerer, and so on.

What it doesn't do is prove that there's anything unfair about it.

That's certainly true.

Are not relevant in an analysis of the effects of differences in stats. Fortunately, they're easy to factor out, by just holding them the same.

Those things will have an impact on the success or failure of the PC. A PC doesn't fight with stats alone, so using stats alone doesn't really tell the proper story.

 

[Oofta]

No.

Feel free. It will have nothing to do with real game situations, though.

In a white room only. It doesn't show diddly for comparisons between PCs of different races, classes and abilities, like will happen in a real game.

You keep harping on that. In other words there is no evidence, no proof no simulation that could show you that there is a difference. You've made up your mind and that's it.

 

[Oofta]

What it doesn't do is prove that there's anything unfair about it.

Fairness is in the eye of the beholder. It's an opinion, one we may differ on. I don't have a problem with that, what I have a problem with is the refusal to accept proof that the variation between characters can be significant.

 

[Maxperson]

You keep harping on that. In other words there is no evidence, no proof no simulation that could show you that there is a difference. You've made up your mind and that's it.

Classic. You can't defeat my argument, so you attack me personally. There's no way for you to simulate all the race/class/ability/feat combinations that affect combat. For instance, if the high stat PC is a tier 3 class and the low stat character is tier 1 class, that could mostly or even completely erase the stat differences, especially depending on the build.

 

[Lanefan]

what I have a problem with is the refusal to accept proof that the variation between characters can be significant.

That a high-low variance in stats will make a difference between two characters (Harry and Larry, say) who are otherwise the same (race, class, skills, feats, h.p.-per-level, etc.) and who are doing what they do in isolation cannot be disputed. Hell, I heard somewhere that a bored programmer came up with the numbers in one afternoon to prove it.

BUT: how much if any of that difference stands up over the long term when these two characters are operating in a larger party - and this can happen in a few* ways - can realistically never be proven. There's so many variables it'd be like trying to model the weather. The problem is, it's this we're trying to discuss; as a party of only two characters is (IME at least) extremely rare: what are each of Harry Histat and Larry Lowstat's long-term survival chances and how much difference is there between these results.

* - such as:

- there are two parties that are exactly the same and who do exactly the same things. Harry is in one, Larry in the other
- Harry and Larry are working together in the same single party with a reasonable mix and spread of other level-appropriate characters (the 'classic' example of high-low stat characters in the same party)
- Harry and Larry are in two different parties - let's say both are level-appropriate - each of which is also randomized as to number of other characters along with which races, classes, and stat-strata [EDIT - and levels] are represented
- Harry and Larry are themselves the entire party.

Lanefan

 

[Lanefan]

There's no way for you to simulate all the race/class/ability/feat combinations that affect combat.

True. See my post just previous to this one.

For instance, if the high stat PC is a tier 3 class and the low stat character is tier 1 class, that could mostly or even completely erase the stat differences, especially depending on the build.

Oofta's analysis did set the two characters in question at the same level. He was strictly looking at the difference raw stats made, while eliminating as many other variables as he could, and - given that - his results are pretty clear.

That said, 5e does forgive some character level variance within a given party far better than 3e or 4e did, so yes: level becomes yet another variable.

Lan-"100,000 Dwarves, one at a time, adds up to one really well-fed hell hound"-efan

 

[Maxperson]

Oofta's analysis did set the two characters in question at the same level. He was strictly looking at the difference raw stats made, while eliminating as many other variables as he could, and - given that - his results are pretty clear.

Sure, but since at the same level race/class/feats do much more for combat than ability scores, eliminating race/class/feats to look at ability scores is just slightly more useful than eliminating those to look at the combat effectiveness of picking your nose. It's interesting as a white room exercise, but does little for looking at what happens in actual game play.

 

[Oofta]

Classic. You can't defeat my argument, so you attack me personally. There's no way for you to simulate all the race/class/ability/feat combinations that affect combat. For instance, if the high stat PC is a tier 3 class and the low stat character is tier 1 class, that could mostly or even completely erase the stat differences, especially depending on the build.

I can't defeat your argument because it is based on opinion, not logic. There's no reason to compare different classes and builds. This is an attempt to measure the impact of a different set of ability scores for a particular build.

If I've decided to play Character/Race/Class X and I get high numbers, that character will be more effective than if I rolled low. It doesn't matter what class X is. I don't see how that statement can be refuted. I've shown how much of a difference it is for one particular scenario. If you say "you should play Y" then my character would have been more effective at Y as well if I had been lucky and gotten higher numbers. How is that statement incorrect?

Just out of curiosity I took 5 minutes to adjust my scenario to one that is, I believe extremely realistic. So 4th level fighters (at full hp) badly wounded hell hound (half hp). Basically a fight where the dice all went wrong and you're looking at a TPK but fighting a bloodied opponent. Result is pretty much the same. Lorne (low man) wins 63% while Max wins 83%. It does jump to 95/99% if the fighters are 5th level, but it makes sense that at a certain point the fighters are going to win the majority of time (and every once in a while in Max's case just have bad luck).

As far as "you never have the same race/class/build", I've done it 3 times. Well, technically the third time it was a set of 3 gnome triplets we ran with our niece but I think it still counts. I've run games at least twice with duplicates. Once with dwarvish fighter twins, another with gnoll barbarian brothers.

Saying that you've come to a foregone conclusion is not a personal attack. It's an observation.

 

[Tony Vargas]

I can't defeat your argument because it is based on opinion, not logic

Like 'fair' being all opinion?

There's no reason to compare different classes and builds. This is an attempt to measure the impact of a different set of ability scores for a particular build.

The mechanic at issue is stat generation. Random generation means that any given character could have higher or lower stats then the exact same character created with point-buy.

It shouldn't require 'proof' that the character with better stats is better than it would have been with worse stats.

As far as "you never have the same race/class/build", I've done it 3 times.

The first time I ran 5e, the party had 2 gnome bard entertainers.
FWIW (nothing, IMHO, just sharing because it was amusing - they both loved vicious mockery, too, and providef some bright moments in an otherwise horrid adventure).

Classic. You can't defeat my argument

You would have to make an argument, rather than just an assertion.

For instance, if the high stat PC is a tier 3 class and the low stat character is tier 1 class, that could mostly or even completely erase the stat differences, especially depending on the build.

True enough. Of course. If the high stat PC were the Tier 1 class, it might only amplify the disparity with a low-stat, Tier 3 character.

It's not completely irrelevant to argue that one source of imbalance is less significant than another, though, the effort involved in avoiding stat imbalances (use array) is much easier than avoiding class imbalances (redesign all the classes).

 

[Oofta]

Like 'fair' being all opinion?

And I freely acknowledge that what is "fair" is an opinion. Google "studies on fairness" sometime.

I shouldn't have to google "+1 is greater than +0" (BTW that's the average difference: +/-1 for every ability mod).

How much of an impact it had on my test scenario kind of surprised me. What was even more surprising was that if I didn't use the breath weapon for the Hell Hound (I didn't want to over-emphasize the Dexterity save) it changed to Low-Guy wins 51% and Max-Guy winning 80%. It appears that winning initiative (ties always go to the PC) is more important than I thought, especially in a close fight. Or maybe because you always take damage even if you do save?

So no, my 20% difference wasn't a best case scenario to prove my point. It was just one sample out of a nearly infinite number of possibilities. However, I think it should be obvious that the guy with higher ability scores is going to fair better every time.