For all those that say the 4d6 method of char creation is totally legit

Quote from 6thLyranGuard >>

Quote from Pantagruel666 >>

Quote from DMman69 >>

However, I have found that players who get some low stats don't often ask to be killed off. 

 The problem isn't some low stats. The problem is when it's things like "your best stat is a 13".

 So why not let them have a reroll or just bump two of their stats up to 15+?

 I think I remember that being a rule in 3e. Something like , if you had no stat above a 13, or if the sum of all your modifiers was 0 or lower, it was considered an unplayable character and you got to re-roll. 

Quote from 6thLyranGuard >>

 So why not let them have a reroll or just bump two of their stats up to 15+?

Because then you aren't actually using 4k3? The general solution to rolling stats is to not roll stats. There are methods of having randomized character creation that's still fair, but they don't look a lot like rolling for stats individually.

Quote from Xalthu >>

Quote from 6thLyranGuard >>

Quote from Pantagruel666 >>

Quote from DMman69 >>

However, I have found that players who get some low stats don't often ask to be killed off. 

 The problem isn't some low stats. The problem is when it's things like "your best stat is a 13".

 So why not let them have a reroll or just bump two of their stats up to 15+?

 I think I remember that being a rule in 3e. Something like , if you had no stat above a 13, or if the sum of all your modifiers was 0 or lower, it was considered an unplayable character and you got to re-roll. 

 Yes.  It was right there in the PHB.

Quote from Pantagruel666 >>

Quote from 6thLyranGuard >>

 So why not let them have a reroll or just bump two of their stats up to 15+?

Because then you aren't actually using 4k3? The general solution to rolling stats is to not roll stats. There are methods of having randomized character creation that's still fair, but they don't look a lot like rolling for stats individually.

 Or you could roll stats but with guardrails to insure that someone doesn't get stuck with an unplayable characters.  There's no need to insist on some sort of silly "purity" of rolls that's completely unnecessary.

The correct way to play is whatever is fun for everyone at the table (yes, this includes the Dungeon Master).  Whether or not your stats are generated with point buy or dice isn't important.

I will say that if I start a new game I'm tempted by my own variant:

1. Attributes are rolled on 3d6, in order (no placing them where you want them to be)
2. Background does not grant any ability scores (those rolls are what you have)
3. You get to roll 100 characters (tool will be provided).

An example of doing this, with the 'best' 6 (by a custom rating algorithm that seems... reasonable to me)

Str  5 Dex 17 Con 14 Int 12 Wis 13 Cha  8
Str 13 Dex 15 Con 14 Int 15 Wis 17 Cha  3
Str 14 Dex 13 Con 14 Int 16 Wis 15 Cha 11
Str 14 Dex 14 Con 18 Int 11 Wis  8 Cha 11
Str  9 Dex 10 Con 15 Int 16 Wis  7 Cha 16
Str 14 Dex 16 Con 11 Int 17 Wis 12 Cha 11

I'm not 100% sure which of those I'd pick, but there's certainly something viable there, and they do have more character than you see from point buy or standard array.

Quote from Pantagruel666 >>

I will say that if I start a new game I'm tempted by my own variant:

1. Attributes are rolled on 3d6, in order (no placing them where you want them to be)
2. Background does not grant any ability scores (those rolls are what you have)
3. You get to roll 100 characters (tool will be provided).

An example of doing this, with the 'best' 6 (by a custom rating algorithm that seems... reasonable to me)

Str  5 Dex 17 Con 14 Int 12 Wis 13 Cha  8
Str 13 Dex 15 Con 14 Int 15 Wis 17 Cha  3
Str 14 Dex 13 Con 14 Int 16 Wis 15 Cha 11
Str 14 Dex 14 Con 18 Int 11 Wis  8 Cha 11
Str  9 Dex 10 Con 15 Int 16 Wis  7 Cha 16
Str 14 Dex 16 Con 11 Int 17 Wis 12 Cha 11

I'm not 100% sure which of those I'd pick, but there's certainly something viable there, and they do have more character than you see from point buy or standard array.

The most fun I've ever had rolling up characters was doing this:

• The DM has everyone roll up a set of six numbers using 4k3, keeping them in order. If there were fewer than six people in the party, the DM rolled extra sets until we had six
• Those stats were then laid out in a 6x6 grid
• We then determined a random order and each player picked a column, row or diagonal to use, with each set only being used once, from a grid that looked something like this:

13 11 17 12  8 10
10 11  9 12 18 13
 7 16 16 13  9 12
14 14 12  9  8 14
12  7 13 16 14  7
 9 12  9  12 13 15

That way, if someone rolled really well, it could benefit multiple people, and if someone rolled poorly it wasn't an issue. It also made it easier to pick a set more tailored to your build or character concept, if it was MAD/SAD or whatever

Quote from AntonSirius >>

Quote from Pantagruel666 >>

I will say that if I start a new game I'm tempted by my own variant:

1. Attributes are rolled on 3d6, in order (no placing them where you want them to be)
2. Background does not grant any ability scores (those rolls are what you have)
3. You get to roll 100 characters (tool will be provided).

An example of doing this, with the 'best' 6 (by a custom rating algorithm that seems... reasonable to me)

Str  5 Dex 17 Con 14 Int 12 Wis 13 Cha  8
Str 13 Dex 15 Con 14 Int 15 Wis 17 Cha  3
Str 14 Dex 13 Con 14 Int 16 Wis 15 Cha 11
Str 14 Dex 14 Con 18 Int 11 Wis  8 Cha 11
Str  9 Dex 10 Con 15 Int 16 Wis  7 Cha 16
Str 14 Dex 16 Con 11 Int 17 Wis 12 Cha 11

I'm not 100% sure which of those I'd pick, but there's certainly something viable there, and they do have more character than you see from point buy or standard array.

 

The most fun I've ever had rolling up characters was doing this:

• The DM has everyone roll up a set of six numbers using 4k3, keeping them in order. If there were fewer than six people in the party, the DM rolled extra sets until we had six
• Those stats were then laid out in a 6x6 grid
• We then determined a random order and each player picked a column, row or diagonal to use, with each set only being used once, from a grid that looked something like this:

13 11 17 12  8 10
10 11  9 12 18 13
 7 16 16 13  9 12
14 14 12  9  8 14
12  7 13 16 14  7
 9 12  9  12 13 15

That way, if someone rolled really well, it could benefit multiple people, and if someone rolled poorly it wasn't an issue. It also made it easier to pick a set more tailored to your build or character concept, if it was MAD/SAD or whatever

 

 There was this system I heard of but never used. I want to say it’s from 13th Age. You roll 6 d6’s and give each one a letter A through F. Then you start with 6 13’s. To the first 13 you add dice A and subtract B. Then add B and subtract C and so on until you add F and subtract A. 
It always seemed like an interesting idea. But as I said, I’ve never actually tried it.

I've read a bit in this thread. I don't care for the Coville method, or other methods which include "roll stats in order," because I like being able to choose my own character, with created backstory hooks (not answers, hooks) instead of having to improvise one on the spot. That might be fine for the old-school dungeon crawl type games, but if I want to do a dungeon crawl with a minimal, irrelevant backstory, I have my pick of videogames which are better suited for that. The appeal of playing at the tabletop is a unique, interactive story.

Something I don't think anyone else has said is that "roll stats in order" also could create problems in party composition if the stat spread for one of the columns is uniformly poor, or another is overrepresented.

I don't have a problem with point buy or standard array, but I do think rolling can lead to more interesting variations for classes. The argument that rolling is the generation method for min-maxers is a bit silly; point buy and standard array INSIST on min-maxing. I'm also going to agree with the sentiment that if someone is the kind of person who wants to roll stats and then whines and complains if they don't get a perfect, amazing stat line, they would surely whine and complain about a lot of things during the game no matter what stat generation method they use.

The method one of my DM's used for rolling stats was that each player rolled two sets of 4d6dl. If the total of each set was below 72 (sum of standard array), you reroll both sets, but if at least one set was 72 or higher, you have to use one of them. But the method I'm thinking of, if I DM for another game, is to roll up (player count + 1) sets of 4d6dl stats (min total of 72), and the group as a whole chooses which set of stats everyone uses. That solves the issue of a discrepancy between someone who gets a poor set and someone who gets a great set. (I guess if someone really REALLY wants, they could pick a different set.)

Quote from Stabbey_TC >>

but if I want to do a dungeon crawl with a minimal, irrelevant backstory, I have my pick of videogames which are better suited for that. The appeal of playing at the tabletop is a unique, interactive story.

I don't think rolling for stats/improv and engaging storytelling with integrated backstories are mutually exclusive. Perhaps that's been your experience, and if so, that sucks. Or, perhaps you just don't like improv, which is totally fair.

I do get the impression that the older DnD crowd tend to enjoy that more than the "modern" DnD player, but that's a discussion for another thread.

Quote from Stabbey_TC >>

I've read a bit in this thread. I don't care for the Coville method, or other methods which include "roll stats in order," because I like being able to choose my own character, with created backstory hooks (not answers, hooks) instead of having to improvise one on the spot.

The only good reason for rolling stats is to come up with unexpected characters, so if you don't want that... use point buy.

Quote from sabin76 >>

I don't think rolling for stats/improv and engaging storytelling with integrated backstories are mutually exclusive. Perhaps that's been your experience, and if so, that sucks. Or, perhaps you just don't like improv, which is totally fair.

I do get the impression that the older DnD crowd tend to enjoy that more than the "modern" DnD player, but that's a discussion for another thread.

They aren't mutually exclusive, but at the start of Session 1, a concept you've been thinking of for a while and had a chance to think about, and possibly talk about with a GM is going to be more fleshed out than one which you come up with during Session 0.

Quote from Pantagruel666 >>

The only good reason for rolling stats is to come up with unexpected characters, so if you don't want that... use point buy.

I don't agree that is the ONLY good reason for rolling stats.

There's also a difference between coming up with characters some amount of time before Session 0 and coming up with them DURING session 0. A person being able to think of a concept for an unexpected character by piecing it together over time does not mean that same person has the skill to come up with the same quality of character on the spot. They certainly can't come up with one with the same depth in an instant.

Quote from Pantagruel666 >>

Quote from Stabbey_TC >>

I've read a bit in this thread. I don't care for the Coville method, or other methods which include "roll stats in order," because I like being able to choose my own character, with created backstory hooks (not answers, hooks) instead of having to improvise one on the spot.

The only good reason for rolling stats is to come up with unexpected characters, so if you don't want that... use point buy.

The best reason for rolling for stats as opposed to array/point buy is that rolling dice is fun

Some say rolling dice also makes party's stats less samey.

Quote from AntonSirius >>

The best reason for rolling for stats as opposed to array/point buy is that rolling dice is fun

And on a one-shot that lasts three hours, that five minutes of fun might make up for the problems random stats cause later in the game, but when those five minutes have a major controlling effect on a hundred hours or more of gameplay...

Quote from Pantagruel666 >>

Quote from AntonSirius >>

The best reason for rolling for stats as opposed to array/point buy is that rolling dice is fun

And on a one-shot that lasts three hours, that five minutes of fun might make up for the problems random stats cause later in the game, but when those five minutes have a major controlling effect on a hundred hours or more of gameplay...

 Then you can learn how to actually play with a character who's stats aren't 100% optimal.

Quote from Pantagruel666 >>

And on a one-shot that lasts three hours, that five minutes of fun might make up for the problems random stats cause later in the game, but when those five minutes have a major controlling effect on a hundred hours or more of gameplay...

I've been in a 5e campaign that went from level 1 to level 20. There were no such "problems" or "major controlling effect"

It's not about being optimal, though. I think the Devs have recognised that too much depends on those stats, and have started to decouple them from abilities. Eg the Wizard used to prepare a number of spells boosted by their Int modifier, now the number is not affected by it.

If it were simply a case of "your spell fails in one in twenty rolls, which you otherwise would have succeeded on", it would be a case of "not being 100% optimal". I've not done a careful search on this with 2024e, but in 2014e, there was a lot of pressure to go for the highest score possible because of these extra tie ins to features. A 13 Int Wizard was a bad Wizard. Not only would their spells early work, but they had few spells to pick from in the moment.

Quote from Linklite >>

It's not about being optimal, though. I think the Devs have recognised that too much depends on those stats, and have started to decouple them from abilities. Eg the Wizard used to prepare a number of spells boosted by their Int modifier, now the number is not affected by it.

If it were simply a case of "your spell fails in one in twenty rolls, which you otherwise would have succeeded on", it would be a case of "not being 100% optimal". I've not done a careful search on this with 2024e, but in 2014e, there was a lot of pressure to go for the highest score possible because of these extra tie ins to features. A 13 Int Wizard was a bad Wizard. Not only would their spells early work, but they had few spells to pick from in the moment.

 I think that’s true but only to a point and really in theory. In practice, if you’re playing a 13 int wizard (even in ‘14), you’ve almost certainly made a conscious choice as part of your character concept to be a wizard who’s bad at wizarding, and you’ve gotten what you wanted. 

Quote from Plaguescarred >>

Some say rolling dice also makes party's stats less samey.

 Not sure I'd agree statistically.  

Quote from Linklite >>

I never said it was. Read the posts. I said it was easier to minmax with Point Buy - that's semantically very different to saying it is minmaxxing and vital to the discussion.

 

 Except there isn't at least in how the base builder works. There is really no way to max over anyone else. Looking at the 'max' there is very little difference from an average character. Which isn't really min-maxing. And if anything will be more detrimental with the negatives. 
You are basically ignoring the examples I presented.  You have a preference that is biasing you to actual data.