G
Guest
Guest
Archived from groups: rec.games.frp.gurps (More info?)
I've been thinking about something J. Verkuilen said about probability
distributions: that we're too addicted to roll-n-dice-and-sum, which
leads to normal distributions, which are far from universal in real
life. So I started thinking about negative binomial distributions,
which reminded me...
The automatic success at 17+/automatic success at 4- mechanic has
always bugged me. It's worse where I originally came from in AD&D2,
where both a 10th level fighter (THAC0 11) using a +3 longbow (+3 to
hit) at short range in broad daylight and a drunk (-2 to hit) 0-level
beggar (THAC0 20) using a staff sling at long range (-4) which he is
not proficient with(-4) in complete darkness (-4) have the exact same
chance to hit a great wyrm shadow dragon (AC -12)--the fighter needs a
20 or better, the beggar needs a 46(!), but since a natural 20 always
hits they both hit 5% of the time, which if realistic for the fighter
is ridiculously often for the beggar. GURPS has a more normal
distribution, and the auto-success comes only 2% of the time so it's
easier to sweep it under the rug, but the real problem IMO is that at
some point bonuses/penalties become irrelevant because they "fall off
the edge of the dice," so to speak. I realize that this is to keep some
drama in the game, but there must be a better way to deal with this.
GULLIVER has a suggestion about using "Long Rounds" to deal with
too-high defenses. That is, if everybody parries on 17+, abstract out
the combat round to length 2^n and roll dice at -n to defense rolls,
once for each extended round instead. That is, you can have one
attack/defense response per second with parry at 17, which is roughly
equivalent to one attack/defense once every 2 seconds w/ parry 16, or
one attack/defense per minute at parry 11. (Effectively, you have a
~60% chance of parrying everything that minute, and a ~40% chance of
letting one through.) ESCARGO talks in a somewhat-similar way about
amalgamating skills at a point-cost penalty, so a x1.5 cost-multiplier
either buys you knowledge of x1.5 as much knowledge or a +1 to skill,
so depth and breadth trade off. Okay, so here's my proposal:
Granular Skill Checks:
On any skill roll (/ability roll?), the GM/player can either roll it
normally OR take one or more -4 modifiers for "granularity's sake." For
each modifier so taken, one extra skill attempt is made. A success on
ANY attempt means overall skill success. Penalties may not take skill
below 12. A 17+ still means auto-failure.
Example A: driving to work in a small town (Driving-11, routine +5) can
be considered either as a single monolithic skill roll where failure
results in e.g. a minor accident, or a number of routine maneuvers
where each failure brings you closer to danger. E.g. one failed roll
means you forgot to look both ways at the stop sign, but a success
means you still managed to avoid the car coming the other way once you
did notice it. Roll once at 16 (~98% chance of success) or twice at 12
(91% chance of success). (Granular mods turns out to only be beneficial
at 18+ skill level.)
[AFB, so the next modifiers are guesstimates]
Example B: For some bizarre reason, Dead-Eye Dai (Rifles-24) is aiming
his custom-made sniper rifle (Acc 6+3) at a 1-foot-square practice
target (SZ -3) across the his room (30', -4). He needs a 26 or better
to hit. He could choose to roll once (2% chance of 17+ failure) or pay
full attention and roll 3 times at an extra -8. He would need to roll
17/18 three times in a row to miss this target, which he would do
roughly 8 times in a million tries. That seems pretty reasonable.
Remember, multiple skill rolls must be predeclared if they are to be
used (it's not a Luck advantage).
I haven't figured out how to make the math work out right for long-shot
successes like the aforementioned fighter/beggar. I'm also unsure
whether a Luck use should apply to one roll or the whole series of
granular rolls. Finally, I'm not sure how critical successes work into
this--if any roll is a critical success they all are? Same for critical
failures?
Is there anyone else who thinks this (extreme cumulative
penalties/bonuses) is an issue worth dealing with?
Max Wilson
nee Hemlock
I've been thinking about something J. Verkuilen said about probability
distributions: that we're too addicted to roll-n-dice-and-sum, which
leads to normal distributions, which are far from universal in real
life. So I started thinking about negative binomial distributions,
which reminded me...
The automatic success at 17+/automatic success at 4- mechanic has
always bugged me. It's worse where I originally came from in AD&D2,
where both a 10th level fighter (THAC0 11) using a +3 longbow (+3 to
hit) at short range in broad daylight and a drunk (-2 to hit) 0-level
beggar (THAC0 20) using a staff sling at long range (-4) which he is
not proficient with(-4) in complete darkness (-4) have the exact same
chance to hit a great wyrm shadow dragon (AC -12)--the fighter needs a
20 or better, the beggar needs a 46(!), but since a natural 20 always
hits they both hit 5% of the time, which if realistic for the fighter
is ridiculously often for the beggar. GURPS has a more normal
distribution, and the auto-success comes only 2% of the time so it's
easier to sweep it under the rug, but the real problem IMO is that at
some point bonuses/penalties become irrelevant because they "fall off
the edge of the dice," so to speak. I realize that this is to keep some
drama in the game, but there must be a better way to deal with this.
GULLIVER has a suggestion about using "Long Rounds" to deal with
too-high defenses. That is, if everybody parries on 17+, abstract out
the combat round to length 2^n and roll dice at -n to defense rolls,
once for each extended round instead. That is, you can have one
attack/defense response per second with parry at 17, which is roughly
equivalent to one attack/defense once every 2 seconds w/ parry 16, or
one attack/defense per minute at parry 11. (Effectively, you have a
~60% chance of parrying everything that minute, and a ~40% chance of
letting one through.) ESCARGO talks in a somewhat-similar way about
amalgamating skills at a point-cost penalty, so a x1.5 cost-multiplier
either buys you knowledge of x1.5 as much knowledge or a +1 to skill,
so depth and breadth trade off. Okay, so here's my proposal:
Granular Skill Checks:
On any skill roll (/ability roll?), the GM/player can either roll it
normally OR take one or more -4 modifiers for "granularity's sake." For
each modifier so taken, one extra skill attempt is made. A success on
ANY attempt means overall skill success. Penalties may not take skill
below 12. A 17+ still means auto-failure.
Example A: driving to work in a small town (Driving-11, routine +5) can
be considered either as a single monolithic skill roll where failure
results in e.g. a minor accident, or a number of routine maneuvers
where each failure brings you closer to danger. E.g. one failed roll
means you forgot to look both ways at the stop sign, but a success
means you still managed to avoid the car coming the other way once you
did notice it. Roll once at 16 (~98% chance of success) or twice at 12
(91% chance of success). (Granular mods turns out to only be beneficial
at 18+ skill level.)
[AFB, so the next modifiers are guesstimates]
Example B: For some bizarre reason, Dead-Eye Dai (Rifles-24) is aiming
his custom-made sniper rifle (Acc 6+3) at a 1-foot-square practice
target (SZ -3) across the his room (30', -4). He needs a 26 or better
to hit. He could choose to roll once (2% chance of 17+ failure) or pay
full attention and roll 3 times at an extra -8. He would need to roll
17/18 three times in a row to miss this target, which he would do
roughly 8 times in a million tries. That seems pretty reasonable.
Remember, multiple skill rolls must be predeclared if they are to be
used (it's not a Luck advantage).
I haven't figured out how to make the math work out right for long-shot
successes like the aforementioned fighter/beggar. I'm also unsure
whether a Luck use should apply to one roll or the whole series of
granular rolls. Finally, I'm not sure how critical successes work into
this--if any roll is a critical success they all are? Same for critical
failures?
Is there anyone else who thinks this (extreme cumulative
penalties/bonuses) is an issue worth dealing with?
Max Wilson
nee Hemlock