D&D 5E Point Buy vs Rolling for Stats

[Lanefan]

Yeah, that's a great idea. No variance there at all. O.O Yikes. Low ball of 72 high ball of 103! Nope, there's going to be no noticeable difference in these PC's whatsoever.

Yet even the '72' can cobble together one 18...though the rest of the stats would be a bit of a disaster unless those 6s were spread around a bit instead of clumped together.

That 103 is a real outlier. There's 8 6s and 5 5s, and only 2 2s and 2 1s (dead average is 4 of each number) - over half the dice are either 5 or 6. Crazy! That said, the 96 is deceptively good in that all its discards are 1s - there's 2 18s and 2 15s hiding in there...or 18-17-16-15 if you like.

And, oddly enough, while both the 72 and the 74 can put together an 18 neither of the '76' sets can - in fact the highest stat the second 76 can produce is a 16...which tells me that while getting an 18 seems common in this system it's by no means guaranteed.

Lanefan

 

[FrogReaver]

I've been off the Internet for a week, and I've just read the 20 pages of this thread that I missed.

I can't be bothered to go through all 20 pages to find the post in question, but one post that made a point that motivated this reply was on the subject of "people who roll are only doing so to get Mary-Sue". I won't bother to refute this idea as it should be obvious it refutes itself, but the specific point was made about the CR system being all messed up if high stat PCs are used, and the array '16, 16, 14, 12, 12, 12' was derided for being too high for the system to handle without breaking.

Let's take your scenario idea and have two unarmoured barbarian humans instead of two fighter dwarves that have already made so much money from adventuring that they can both afford full plate armour at 4th level(!)....would you think that this would be at least as 'fair' a test as your own?

Let's say that the 'so unplayable it's broken' array mentioned above was given to one twin, ending up with: Str 16 Dex 14 Con 16 Int 12 Wis 12 Cha 12.

Now compare this with the other twin, made with point-buy: Str 16 Dex 14 Con 16 Int 8 Wis 10 Cha 8. Is there any difference at all when they do what they are intended to do: solo a hell hound, using your program?

And yet, the first is deemed as so good it's breaking the system. Point-buy will not let you buy it, but it allows for more concepts.

It looks like you have racial bonuses to one character and not the other

 

[TwoSix]

Yet even the '72' can cobble together one 18...though the rest of the stats would be a bit of a disaster unless those 6s were spread around a bit instead of clumped together.

That 103 is a real outlier. There's 8 6s and 5 5s, and only 2 2s and 2 1s (dead average is 4 of each number) - over half the dice are either 5 or 6. Crazy! That said, the 96 is deceptively good in that all its discards are 1s - there's 2 18s and 2 15s hiding in there...or 18-17-16-15 if you like.

And, oddly enough, while both the 72 and the 74 can put together an 18 neither of the '76' sets can - in fact the highest stat the second 76 can produce is a 16...which tells me that while getting an 18 seems common in this system it's by no means guaranteed.

Lanefan

103 is a decent size outlier, only about a 1.3% chance, according to anydice. About half the chance of rolling an 18 in a straight 3d6 6 times array.

 

[Oofta]

I've been off the Internet for a week, and I've just read the 20 pages of this thread that I missed.

I can't be bothered to go through all 20 pages to find the post in question, but one post that made a point that motivated this reply was on the subject of "people who roll are only doing so to get Mary-Sue". I won't bother to refute this idea as it should be obvious it refutes itself, but the specific point was made about the CR system being all messed up if high stat PCs are used, and the array '16, 16, 14, 12, 12, 12' was derided for being too high for the system to handle without breaking.

Let's take your scenario idea and have two unarmoured barbarian humans instead of two fighter dwarves that have already made so much money from adventuring that they can both afford full plate armour at 4th level(!)....would you think that this would be at least as 'fair' a test as your own?

Let's say that the 'so unplayable it's broken' array mentioned above was given to one twin, ending up with: Str 16 Dex 14 Con 16 Int 12 Wis 12 Cha 12.

Now compare this with the other twin, made with point-buy: Str 16 Dex 14 Con 16 Int 8 Wis 10 Cha 8. Is there any difference at all when they do what they are intended to do: solo a hell hound, using your program?

And yet, the first is deemed as so good it's breaking the system. Point-buy will not let you buy it, but it allows for more concepts.

So you have a problem with the fact that I gave them both armor? I'm pitting 4th level characters against a hell hound in a cage match. I thought I'd at least give them a fighting chance. But their AC is identical so it should have no impact on the scenario (assuming they both use heavy armor). Maybe someday I'll allow AC adjustment in my program.

I'm measuring relative power. In relationship to each other the guy with the higher stats is much better off.

That's not a judgement, it's just giving some numbers around how much better off they are. BTW I've also run the numbers with some other options (give me a week or two and a roadtrip to build in a little more flexibility) but the difference is between 20-30% more effective in combat.

I'm not sure where the Mary-Sue came from. It was some tangent. Similar to the tangent of whether or not 4th level PCs should have plate. Suffice to say that it came from the discussion that I don't want to play characters that have above average stats (like the 11 being the lowest number). Well, and something about claiming that you can't build a character to a concept with point buy because if your concept is "I'm the strongest, toughest, fastest, smartest, wisest, prettiest person to ever walk the planet" you can't build it.

As far as someone with 16 16 14 12 12 12 vs 16 14 14 10 8 8, I've never said either one breaks the system or is unplayable. I've just stated that with standard 4d6 drop lowest there will be on average a 2 point difference for every stat. Based on my scenario that equates to a 20-30% difference in combat effectiveness.

I think this ties goes back into the "CR is broken" idea. If people are using systems that provide consistently higher than average points, those characters are going to be significantly more effective at combat on average.

To summarize: I wrote a program to see how much variance you could expect with what some people consider minor differences. While there could be a near infinite variation of generation methods and ways of measuring effectiveness, that does not negate the result I came up with. I chose one scenario that was common but also simple to model. It showed a much more significant power difference than I had expected. That's it.

 

[Tony Vargas]

I can't be bothered to go through all 20 pages to find the post in question, but one post that made a point that motivated this reply was on the subject of "people who roll are only doing so to get Mary-Sue". I won't bother to refute this idea as it should be obvious it refutes itself

Like refuting the idea that people gamble because they want to win money. Obviously, the odds are stacked against them, so they must actually like losing money, right?

but the specific point was made about the CR system being all messed up if high stat PCs are used, and the array '16, 16, 14, 12, 12, 12' was derided for being too high for the system to handle without breaking.

That array doesn't seem that scary. Where's the 2nd 18?
Where's the post-racial 20?

But, I have found that the odd high-stat (that 20 after bonuses, for instance) makes a difference at 1st level: it can actually help.

But, that's only true if your concept is "better than everyone else around". Not really a concept I'm all that interested in the game supporting. Sure, those two barbarians might fight that hell hound the same. Of course, the first barbarian will notice the hell hound more often, and will have a better chance of knowing that it is a hell hound and not some other creature. So, other than just being "better" than our point buy barbarian, what concept do you have in mind, [MENTION=6799649]Arial Black[/MENTION]?

The point isn't better, just different. A concept that required more stats than point buy could be achieved by raising point buy for everyone - at which point, another concept, calling for fewer points (even for exactly the former point-buy value!), would be un-achievable via point buy, but you might get it with a random roll.

I think this ties goes back into the "CR is broken" idea. If people are using systems that provide consistently higher than average points, those characters are going to be significantly more effective at combat on average.

Y'know, it might be that CR is just broken, and it might be that significant variance in PC abilities is broken. But it could also be both.

 

[Lanefan]

I think this ties goes back into the "CR is broken" idea. If people are using systems that provide consistently higher than average points, those characters are going to be significantly more effective at combat on average.

Which really has nothing to do with CR if CR is viewed as only comparable to itself.

To explain: when comparing a high-stat party to a low-stat party you're comparing nothing but party vs. party based on one variable. No worries there.

But when using CR (in any edition that has it, not just 5e), isn't it all just an attempt at comparing challenge vs. challenge in a similar way, only taking more variables into account? When a DM sees that an encounter is CR 5 she can - one hopes - safely assume there's a bit less to it than the CR 6 encounter on the next page, but it's got more teeth than the CR 3 encounter on page 5. It compares challenge vs. challenge quite well, in a broad-brush sort of way.

Where it falls down is when DMs (and the designers, for all that) expect the CR rating to directly mirror what a particular party at a particular table can handle. It can't. No two parties are the same...and even if they were, no two groups of players running them are the same. Only after some trial and error will a DM be able to say with reasonable confidence how the CR numbers will line up with what her particular party can take, based on a whole bunch of variables only one of which is the characters' base stats.

In short: CR isn't broken, it's just being asked to do more than it's really capable of.

Lan-"and never underestimate the resilience of an adventuring party - they're like weeds"-efan

 

[Hussar]

Yet even the '72' can cobble together one 18...though the rest of the stats would be a bit of a disaster unless those 6s were spread around a bit instead of clumped together.

That 103 is a real outlier. There's 8 6s and 5 5s, and only 2 2s and 2 1s (dead average is 4 of each number) - over half the dice are either 5 or 6. Crazy! That said, the 96 is deceptively good in that all its discards are 1s - there's 2 18s and 2 15s hiding in there...or 18-17-16-15 if you like.

And, oddly enough, while both the 72 and the 74 can put together an 18 neither of the '76' sets can - in fact the highest stat the second 76 can produce is a 16...which tells me that while getting an 18 seems common in this system it's by no means guaranteed.

Lanefan

My issue here, really, is a balance thing. It's similar to how balance was achieved in earlier editions. The idea of balance over time. My problem though is that's not balance. It's simply multiple points of imbalance, which is not balance, IMO.

Sure, if we take a hundred rolls, the bell curve is going to smooth out and fair enough. Large enough sample size and all that. But, a group is never that large. The sample size in any given group is only about 4-6, maybe 10 at the most. Which means you never get a bell curve in a given group. You get wildly varying stats between the PC's. So, one guy has that 103 or the 94 and another guy has the 72 or 74.

Which, to me, is just too imbalanced. Sure, that 103 is only a 1.5% chance. Ok, fair enough. But, it WILL happen in some groups. Take 20 tables and the odds are that someone in those 20 tables now has a super powered character and someone will have the 72.

To me, this does not add anything to the table. All it means is that Bob is now a super star because he is just that much better at pretty much everything while Dave is relagated to the side because he just can't really contribute most of the time. Particularly in a fairly well balanced campaign across all three pillars. Bob can fight, explore and talk with the best of them. Better than the best of them in some cases but rarely much worse. While Dave, with his 72 point character maybe contributes well in one pillar and then goes and plays Xbox for the other 2/3rds of the session.

 

[Caliban]

To me, this does not add anything to the table. All it means is that Bob is now a super star because he is just that much better at pretty much everything while Dave is relagated to the side because he just can't really contribute most of the time. Particularly in a fairly well balanced campaign across all three pillars. Bob can fight, explore and talk with the best of them. Better than the best of them in some cases but rarely much worse. While Dave, with his 72 point character maybe contributes well in one pillar and then goes and plays Xbox for the other 2/3rds of the session.

But Dave will take those crappy stats and roleplay the hell out of them. Sure, his character sucks on almost every level imaginable, but that's just an opportunity to show the rest of the group that he's a true roleplayer who can create a memorable character with any set of stats. They'll be telling stories of asthmatic stableboy who tagged along with their heroes for years to come.

I am not Dave. I'll take point buy every time.

 

[Oofta]

In short: CR isn't broken, it's just being asked to do more than it's really capable of.

What do you know. We agree on something!

I agree that CR (and the XP guidelines) are just a starting point. The developers had to use a baseline, which presumably would be a 4 party team of casual players with average stats, no feats and minimal (if any) magic items.

 

[Oofta]

But Dave will take those crappy stats and roleplay the hell out of them. Sure, his character sucks on almost every level imaginable, but that's just an opportunity to show the rest of the group that he's a true roleplayer who can create a memorable character with any set of stats. They'll be telling stories of asthmatic stableboy who tagged along with their heroes for years to come.

I am not Dave. I'll take point buy every time.

<sarcasm>
Wait, are you saying that Dave isn't the awesomest ever because in that one game the die roll lottery winner rolled a 1 and Dave rolled a 20 to save the day?

'Cuz that one time Dave's character was really great. Yeah. Still talk about that one time around the campfire at night. Too bad Dave's character caught a cold and died because he had that 6 con.

</sarcasm>