Stat Generation Methods Poll

[imported_Rasyr]

Okay, I have recently been seeing a lot of animosity against the random rolling of stats, so I was wondering if perhaps I should to change up the included stat generation methods, removing the current random generation method and replacing it with a point buy method

Proposed Point Buy Method -- have 100 points to spread amongst the 8 stats. Every stat must have a minimum of 6, and no stat can be given more than 20 points.

Current Random Method - Roll 2d10, and keep the higher of the two numbers. Repeat 7 more times until you have 8 numbers. Add 10 to 2 of those numbers and put them in the prime stats. Add 5 to each of the remaining 6 numbers and place them however you want.

Current Pregenerated Stats -- place the following numbers however you want them -- (17, 16, 15, 14, 13, 12, 10, 9), note that this totals to 106 points. I was thinking of perhaps raising this total to 110 points (18, 16, 16, 14, 14, 12, 10, 10). This obviously gives more points than the  Point Buy method would allow, but the Point Buy method would allow for the player to place his stats as he wanted, even having some higher than an 18.

And if you have other ideas for stat generation, please feel free to post them...

[AlCook]

Voted for the point buy method.  While I'm a fan of random rolling of stats, I think the game might be better served if the standard stat generation was point based.  You could always have as an addendum in the back of the book alternative stat generation systems.  I'm not a huge fan of pre-generated stats and placing them where you like.  Makes it seem like all PC's are cut from the same cloth.  With point buy, you can have greater variation in the characters you're making.  Just my two cents.

[brianbloom]

I like choice, and I think a point buy and a random roll method are the best two options to offer... 
however..
I have problems with the implementation of both of them as currently written.

Rolling
I love to roll dice for making characters and would love this to be an option for all players (with GM's approval). The current 2d10, keep-higher-and-add-5/10 roll method isn't a bell curve, but instead a ramp that favors a lot of 13,14,15s (for the stats you are adding 5 to).  You are more likely to get higher numbers than lower ones (the median is 13, in fact).  And for the +10 option, you are more likely to have a 17 than a 16!  16s end up being a small "trough" in the distribution.  In order to keep it as a bell curve, we might want to look at either doing:

• 2d10, added together, then perhaps giving a small pool of extra points to bump up a few stats
• 3d10, drop lowest (sorta like the old 4d6 drop lowest), which gives a similar average of 13.5 as the existing "ramp" method, but with a bell curve instead.

Perhaps a compromise would be 4 stats of 2d10 and 4 stats of 3d10 (drop lowest).  That would give you a few average stats and a few exceptional ones (I kinda like this since we're already willing to have 6 stats with +5 and 2 with +10, we already have a precedent for splitting the method among the various stats).  This would give you an overall average of 12.24, which sounds about right for an adventurer, but still a shot at a couple impressive numbers for your key stats.

Point buy
I like point buys, but I also like them to have a premium for buying really high stats.  For example, in a certain popular RPG, point buy stats of 8-14 cost 1 point per stat point.  A 15 or 16 cost 2 points each, while raising to a 17 or 18 cost 3 points each (and that scale was capped at 18).  That way, if you wanted that really juicy high number, it came with a price.  I'd like to see something like that here.  It keeps exceptional scores as "exceptional".  So, yes, please sanction a point buy as a character gen method, but let's calibrate it a little.

One way we could do so, that fits pretty closely to the current "Stat bonuses"  table, would be for all stats to start at a 6 (in keeping with your desire for that to be the min), all stats 7-15 costs 1 point each, each step from 16-18 costs 2 points, and 19 or 20 costs 3 points each.  A player would get 60 points to allocate among their stats.  (The pregenerated stats amount to 61 points by this system, so you could almost make that same lineup with this)  You could also buy up to some pretty high values, but you'd have to balance those with some average or lower ones.  (I personally think it's fun to role play having at least one low stat in the mix).

For those who like a visual depiction, here it is in chart form:

Stat	Point cost
6	0
7	1
8	2
9	3
10	4
11	5
12	6
13	7
14	8
15	9
16	11
17	13
18	15
19	18
20	21

[/list]

[imported_Rasyr]

Rolling --

The current method generates 6 stats that range from 6 to 15, so the median of that would be about 10-11. The reason for rolling 2 dice and dropping the lowest is that sometimes you can get a bad roll and this allows for some minor adjustment. However, since each die roll is looked at separately, that still means that the median for each die is still going to be 5-6. This means that our median for the Prime Stats is going to be 15-16 on average, with the occasional being higher. -- Just trying to explain my reasoning here...

The 2 things that NEED to remain the same here, regardless of any potential changes, is that 6 is the lowest possible stat value for Humans and 20 is the highest possible.


Point Buy --

Instead of starting at 5 points and then going up from there, why don't we start at a base of 10, and then the first 5 points cost 1 point each, and the next 5 cost 3 points each (thus, still within the same scale for purchasing Favored Ranks as you suggested.

However, we could allow a player to sell off up to 4 points from each stat (meaning a minimum stat of 6) on a 1 for 1 basis.

And with the combination of these 2 rules, we only give the player 35 points to spend on stats, and the costs per stat would be as follows:

6   = -4
7   = -3
8   = -2
9   = -1
10 = 0
11 = 1
12 = 2
13 = 3
14 = 4
15 = 5
16 = 8
17 = 11
18 = 13
19 = 16
20 = 19

So, if we look at my proposed Stat Block (18, 16, 16, 14, 14, 12, 10, 10), that would cost a total of  39 (13 + 8 + 8 + 4 + 4 + 2 +0 + 0) points
So, if we look at my current Stat Block(17, 16, 15, 14, 13, 12, 10, 9), that would cost a total of  32 (11 + 8 + 5 + 4 + 3 + 2 +0 + -1) points

Therefore, by giving them 35 points to spend on stats, using the costs outlined here, that would allow for a couple of good stats and some slightly better than average stats.... Afterall, adventurers are supposed to be better than average. But it doesn't rule out average or worse than average stats (perhaps we should include a rule that converts 3 unused stat points to 1 extra Character Point that may be spent on skills, talents, etc...

[brianbloom]

The current method generates 6 stats that range from 6 to 15, so the median of that would be about 10-11.

Actually, the median is 13. (An 8 from the dice plus 5 added to it)  It would be 10-11 if all the numbers 1-10 were equally likely, but since you're taking the higher of them, you get a distribution like the attached picture...

The reason for rolling 2 dice and dropping the lowest is that sometimes you can get a bad roll and this allows for some minor adjustment. However, since each die roll is looked at separately, that still means that the median for each die is still going to be 5-6. This means that our median for the Prime Stats is going to be 15-16 on average, with the occasional being higher. -- Just trying to explain my reasoning here...

For the prime stats, the median is 18 (An 8 from the dice plus 10 added to it).  Yes, the median of each die is 5-6 as you describe, but since you're always taking the better, you get a net median of 8. (and an average of 7.15).  When you add 10, the median prime stat jumps to 18.  The chart here shows the "dip" where you can see that 20s will be more common than 16s!

(sorry for geeking out on this, but the entire reason I am interested in this system is that I didn't like the flat distribution of the d20 system.  I did a lot of numerical analysis to examine ways of making it better, and whipping up distribution charts like this was part of that)



[attachment[/attachment]

[imported_Rasyr]

Hmm... it is sounding like 3d10, drop the lowest (rerolling anything below a 6) may be the way to go for a random roll determining stats...

Brian - what did you think of my variation of the point buy method that you suggested?

[imported_Rasyr]

efiscrow responded to the poll notification via email and he made a suggestion. I am including it here (in the quote below), so that it can be included in the discussions as well.

efiscrow wrote:

OK. Perhaps you could incorporate the Talent purchase points with the Stat purchase point, so that it all revolves around one system.

Let's say, for example, all stats begin at 12. A class gives you +2, or maybe one stat at 14.

Then you can have 'Gifted' or 'Higher Attribute', costing say 3 points. You can buy this each time for a new stat. It raises it by +3, or to 16.

Then, you can have 'Highly Gifted', or 'Exceptional Attribute', costing say 10 points. Again, you can buy this for each new stat. It raises it by a further +2 point, or up to 18. You need the pre-requisite of the 'Gifted.' talent.

You may not be allowed to buy these once the game begins.

Finally, you can have a weakness. This gives you back say 4 points, and reduces a stat by 3. This opens up a further can of worms, where you can start including weaknesses as a means for additional points during character creation.

Just an idea. The inherent problem with the 100 points freely distributed amongst the stats, is that many players will dump 20 straight into their most favored. This means that for the most part, all PC warriors will start out with STR 20, all PC wizards will start out with INT 20, etc. This invalidates the worth of any other score, because the warrior with STR 17 will be looked upon as a poor cousin.

Of course you could get rid of all these problems by dumping the scores, and just keep the bonuses. Thus, with say 6 stats, you have the bonuses from -3 to +5. Have them all start out as +0, and allocate 4 points. If you want more, you need to reduce other stats to gain additional points - thus a +5 comes at a cost. OR, you allocate 10 points, and the cost is equal to the bonus given - per point given. Thus, +1 cost 1, +2 costs 2, so gaining +2 costs 1+2 = 3. A +5 costs 15 points. A -3 can give you -6 at a time.

[imported_Rasyr]

Personally, I think that efiscrow has an interesting idea. However, my issue with it is that it radically changes stat generation, and radical changes something I am trying to avoid at the moment - especially since I am hoping that version 0.7 of the Open Beta will be the last Open Beta version and that the next version will be the finished product and I can then focus on other products for Novus.

[brianbloom]

Hmm... it is sounding like 3d10, drop the lowest (rerolling anything below a 6) may be the way to go for a random roll determining stats...

Brian - what did you think of my variation of the point buy method that you suggested?

First of all, I think there might be a small error in the point buy numbers you quoted.  There's only a 2 point jump from 17 (11pts) to 18 (13pts), when I think that should be 3 points according to your description.  But I used your chart as written for the following. I can correct the numbers as needed, but it will mostly just bump up the point counts of the more powerful characters below...


So I just experimented with both the rolling method you suggest, plus the point buy values.  I wrote a dice roller that implements those rolls (3d10, top 2, stat >=6 ) and ran 100 random "characters" through it.  Here are the highest, middle, and lowest lineups for that:

#1) 20    19  18  18  18  11  11  11 (avg: 15.8, and worth 77 points using proposed point buy values)
#50) 20    17  15  14  13  10  10  9 (avg: 13.5, and worth 41 points using proposed point buy values), the typical character
#100)15  15  14  11  10  9   7   6 (avg: 10.9, and worth 7 points using proposed point buy values)

Even that lowest character, the 1% worst case scenario, is still a very playable one with a couple of 15s, and the top one is a monster with 5 stats of 18 or higher, which makes think that we may still be a little "high" in the overall scale.

Two ways to trim it down a bit:
If we do the 4 stats with 3d10 (drop lowest) and 4 stats 3d10 (drop middle, which is effectively a 2d10) we get:
#1) 19   18  18  17  16  13  10  10 (avg: 15.2, and worth 64 points using proposed point buy values)
#50) 20    13  13  13  12  10  9   9  (avg: 12.4, and worth 28 points using proposed point buy values), the typical character
#100) 16    13  12  12  9   9   6   6  (avg: 10.4, and worth 5 points using proposed point buy values)

or perhaps a simpler/more elegant method of 2d10 but rerolling any 1s or 2s (which also ensures a minimum stat of 6):
#1) 19 18  18  18  16  15  10  9 (avg: 15.4, and worth 67 points using proposed point buy values)
#50) 17 15  14  14  13  13  12  9 (avg: 13.4, and worth 31 points using proposed point buy values), the typical character
#100) 16 14  12  10  8   7   7   6 (avg: 10, and worth 2 points using proposed point buy values)

For comparison, the traditional D&D method of 4d6, drop low, gives you:
#1) 18  17  16  15  15  14  14  13 (avg: 15.3, and worth 53 points, although the max stat is 18)
#50) 17   15  13  13  13  11  8   8 (avg: 12.3, and worth 22 points)
#100)12 11  11  10  8   7   7   7 (avg: 9.125, and worth -7 points)

So I will leave these numbers up for everyone's consideration.  I personally have never been the kind of player to desire super-high stats, because any good DM will simply throw super-hard baddies at you.  I like a mix of a few good stats, maybe one exceptional one, and a few flaws.  Forces you to have to defer to other players when something challenges your weaknesses.

I think I myself lean toward the "2d10, reroll 1s/2s" as they keeps the namesake 2d10 mechanic, and everyone is ensured of having 2 d10s, but not always 3...

(disclaimer/footnote:  The above numbers are just sample sets run 100 at a time.  They do not necessarily reflect average probability.  But if you want those, I can compute those too

[imported_Rasyr]

First of all, I think there might be a small error in the point buy numbers you quoted.  There's only a 2 point jump from 17 (11pts) to 18 (13pts), when I think that should be 3 points according to your description. 

Ack!! You are correct...

It should be:

6   = -4
7   = -3
8   = -2
9   = -1
10 = 0
11 = 1
12 = 2
13 = 3
14 = 4
15 = 5
16 = 8
17 = 11
18 = 14
19 = 17
20 = 20

So, how about this for a proposed Stat Block(17, 16, 15, 14, 14, 12, 11, 10), that would cost a total of  35 (11 + 8 + 5 + 4 + 4 + 2 +1+ 0) points -- it raises the low end slightly and makes it equal 35 points, which is what I propose giving to somebody assigning their own stats using the point buy method. Thus the stat block method then becomes simply a short hand of the point buy, allowing for slightly faster chargen...

Hmm... I like your idea of rolling 2d10 and rerolling any 1s or 2s on the dice. (almost the same thing as saying reroll any results less than 6), could even say:

Random Method: Roll 2d10 eight times, rerolling any results that are less than 6. Record each result on a piece of scrap paper. Once you have all 8 numbers, then assign them to your stats in any order that you wish.

And if we do that, that would eliminate the need for the example that is there, and I would include the new point buy method as well and leave in the stat block (i.e. go back to 3 methods).

[brianbloom]

Hmm... I like your idea of rolling 2d10 and rerolling any 1s or 2s on the dice. (almost the same thing as saying reroll any results less than 6), could even say:

Almost, but not the same.  Rolling 2d10, and redoing any stat less than 6 still gives the standard 2d10 numbers, peaking at 11, just without anything under 6.  Rolling 2d10, but rerolling 1s and 2s, gives higher numbers (since you are eliminating all the 1+8, 1+9, 1+10, etc) and results in a distribution that peaks at 13.

You'll get "average" characters with redoing stats under 6s, and you'll get "buff" characters with the reroll 1/2s.

The attached image shows both distributions.

Prerolled stat block
For the prerolled stats, I personally would like to see at least one stat that is a handicap.  I mean, the average "tank" fighter almost certainly deserves some lower Int or Wis, similarly the average wizard is probably not muscular and strong.  So I would prefer to see an 8 or 9 included in the "prerolled" numbers.  Yes overall, they represent above average individuals, but to not have any weaknesses below average feels kinda boring to me.  Part of the challenge is for the wizard to figure out how to compensate for being a weakling, or clumsy, or whatever.  Otherwise every player becomes "an army of one" and doesn't really need other players to cover their gaps.



[attachment[/attachment]

[brianbloom]

So since you have seem interested in the "buff-er" stats, Tim, I rolled up random 10 characters using the 2d10/reroll 1s and 2s method, and have shown them in the attached image.

They are sorted by average stat, so most players will get a character that looks like #4 thru 7, but all of them are certainly possible, and I personally think all would playable, although for #10 I might ask the DM for a reroll.

For each one, I show the avg stat, median stat, how much that character is "worth" via the [corrected] point buy values, and then the net (sum) of their stat bonuses from those stats.

I like that most of the typical characters have a weak stat to deal with.  Everyone else satisfied with numbers like these?

[and my apologies if this technical detour has bored everyone to tears.  I don't think anyone needs to pay this much attention to the math.  But I do think players can sense or will eventually discover when a system is unbalanced.  Plus if we are having 3 different stat creation systems, we should get them to produce comparable characters. So I think it's important to devote this time to get this "right", tuning it to values that should ultimately just feel seamless and invisible.  I hate having my fantasy world jarred out of its magic when some dice mechanics suddenly feel "wrong". ]

[attachment[/attachment]

[imported_Rasyr]

Hmmmm..........

Here is an interesting thought that I just had.....

For the Point buy method, we were already discussing the idea of lowering a stat  from 10 (on a 1 for 1 point basis) down to a minimum of 6, to allow for slightly higher stats in other areas....

What if - what if, we allow this across the board (reduce a stat by 1 point, to get 1 point to be used in purchasing higher stats.

Then the reasons for the methods are as follows (which they basically were before)

Random Roll -- chance of higher overall stats, but also a chance of lower stats overall as well.
Point Buy -- player has to make choices, but can basically decide how he wants his stats to be.
Stat Block -- works as a quick method to get starting stats, for those who don't want to spend the time (or do the math) from either of the other methods.

And by allowing those sorts of adjustments for all three methods, then we still allow for the player to exercise some initiative in his character's design, which in turn makes him more invested in his character and more likely to enjoy the gamer overall....

thoughts? comments?

[brianbloom]

I like the premise, but I see abuse where someone drops all their 11s to 10s (no cost or loss) and bumps other stats (or maybe just 1) up to higher values like a 19 (big gain with a +3 bonus). 
I would prefer either a weighted scale like we've been talking about (where premium stats come at a premium) or at least a simple rule like "2 for 1" (you can surrender 2 points in any stats to get 1 in a given one), so that min-maxing has a price to it.

The underlying premise that has me pushing back on this topic comes from having a system with a curve (the 2d10 system roughly approximates one, esp with exploding dice) and the way bonuses make a difference.  A +1 sword helps a little by shifting your curve against the target's curve.  But by and large they mostly line up still.  But a jump to a +2 sword is more than the jump to +1 was, because it shifts the peak of your curve further into the downward slope of the target's curve, exposing an even larger set of numbers where you will succeed.  And a jump to a +3 sword is a even larger jump than the 1->2 was, making a much larger improvement in your odds than the 1->2 jump did, by shifting your curve into that large empty space above the opponents curve.

What does this mean?  It means that it's not linear.  That incremental bonuses higher up are much more valuable than the same increment closer to the baseline/average.  A +4 or +5 bonus is devastating in a curve system far more than it was in something like d20 (where every increment there was just a flat 5% increase in your odds).   That +3 stat bonus you get from a 19-21 here in 2d10 land is much more valuable than three +1 bonuses from a few 13-15 stats.

So I am simply suggesting that we charge more for that premium.

[imported_Rasyr]

Oh, perhaps I wasn't clear the costs for increasing stats would be the same as for the point buy (1 point for 1 point increase up to 15, then 3 points per 1 point increase from 16 to 20), it is the trade in (-1 point to a stat (any stat value gives 1 point for use in increasing other stats)