Is RuneQuest Broken? Here's Why...

HyrumOWC said:
Nikk said:
atgxtg said:
A lot in going into try to fix the messed up task resoultion system, and skill halving.

Ai, I only realised the problem with skill halving when I read about the terrible brokeness of it here. It's a biggy.

Movement rate? Sounds like it needs a good tweaking (at least that's an easy one to fix!) :D

Personally I'm worried about the high skill levels to begin with. Clearly some part of character generation needs to go, but which?

So fix it. :)

Well he was asking advice with the "what part needs to go bit"?

HyrumOWC said:
Off the top of my head you could do the following:

1. Don't allow freebie points to be spent on skills gained from profession/culture. (Harsh, but doable)

2. Limit the number of freebie points to 75, or 50. (This is very doable, and doesn't "break" anything.)

3. Don't allow more than 10 or 20 freebie points to be spent in any basic skill. (Again, easy to implement)

I'd reccomend something along the lines of option 2 and 3 combined. I am looking at the "advanced character rules" and thinking of using them in a reverse fashion, like so:.

Green Characters
Free Skill Points: A green character recieves 50 free skill points and individual skills may benefit from a maximum of 10 free skill points.

You might want to halve the starting money, but I wouldn't.

HyrumOWC said:
The beauty of MRQ is how modular it is. Unlike d20, you can add and subtract bits without bringing the whole house of cards down.

Hyrum.

Only if the house doesn't have much of a foundation. A lot of thing in RQ are interrelated and changing one thing can have lots of ramifications. Pretty much all the "bugs" that exisist in MRQ are due to Mongoose taking a modular approach rather than an integrated apporach. The whole "revised" spell descriptions are good examples of that. Fireblade went from being an powerful offenseive spell, to something you use to tone down a greataxe or greatsword. THat would make a bit difference of who learns it and why.

So yeah, yuo can still bring the house down. You just might not see it until it collaspes on you.
 
Nikk said:
Personally I'm worried about the high skill levels to begin with. Clearly some part of character generation needs to go, but which?

There's nothing wrong with having a system that scales up cleanly past 100%. Previous versions of RQ mechanics became noticeably constipated for highly skilled characters.

Anyone who has played much Elric, or the later editions of Stormbringer will attest to the fact that it's perfectly possible to come up with RQ compatible mechanics that work well for this sort of thing.

Simon Hibbs
 
OK, some of you are grousing about halving. So I fired up python, and ran some iterations...
Code:
===============================================================================================================================================
N= 10000
Skills: 	Normal;     	Crits;       	Halved;     	Hlvd Cr Pre; 	Hlvd Crt Pst; 	Decrement;    	Dec, crit pre; 	dec crit post. 
Skills: 	Normal;     	Crits;       	Halved;     	Hlvd Cr Pre; 	Hlvd Crt Pst; 	Decrement;    	Dec, crit pre; 	dec crit post. 

a 10 %: 	4005 40.0 %; 	4006 40.0 %; 	4005 40.0 %; 	4006 40.0 %; 	4006 40.0 %; 	4005 40.0 %; 	4006 40.0 %; 	4006 40.0 %
b 20 %: 	5905 59.0 %; 	5905 59.0 %; 	5905 59.0 %; 	5905 59.0 %; 	5905 59.0 %; 	5905 59.0 %; 	5905 59.0 %; 	5905 59.0 %

a 30 %: 	4005 40.0 %; 	4006 40.0 %; 	4005 40.0 %; 	4006 40.0 %; 	4006 40.0 %; 	4005 40.0 %; 	4006 40.0 %; 	4006 40.0 %
b 40 %: 	5905 59.0 %; 	5905 59.0 %; 	5905 59.0 %; 	5905 59.0 %; 	5905 59.0 %; 	5905 59.0 %; 	5905 59.0 %; 	5905 59.0 %

a 50 %: 	4005 40.0 %; 	4006 40.0 %; 	4005 40.0 %; 	4006 40.0 %; 	4006 40.0 %; 	4005 40.0 %; 	4006 40.0 %; 	4006 40.0 %
b 60 %: 	5905 59.0 %; 	5905 59.0 %; 	5905 59.0 %; 	5905 59.0 %; 	5905 59.0 %; 	5905 59.0 %; 	5905 59.0 %; 	5905 59.0 %

a 80 %: 	4005 40.0 %; 	4006 40.0 %; 	4005 40.0 %; 	4006 40.0 %; 	4006 40.0 %; 	4005 40.0 %; 	4006 40.0 %; 	4006 40.0 %
b 90 %: 	5905 59.0 %; 	5905 59.0 %; 	5905 59.0 %; 	5905 59.0 %; 	5905 59.0 %; 	5905 59.0 %; 	5905 59.0 %; 	5905 59.0 %

a 90 %: 	4005 40.0 %; 	4006 40.0 %; 	4465 44.6 %; 	4421 44.2 %; 	4510 45.1 %; 	4465 44.6 %; 	4421 44.2 %; 	4421 44.2 %
b 100 %: 	5905 59.0 %; 	5905 59.0 %; 	5440 54.4 %; 	5485 54.8 %; 	5395 53.9 %; 	5440 54.4 %; 	5485 54.8 %; 	5440 54.4 %

a 100 %: 	4465 44.6 %; 	4421 44.2 %; 	4465 44.6 %; 	4466 44.6 %; 	4490 44.9 %; 	4465 44.6 %; 	4426 44.2 %; 	4421 44.2 %
b 110 %: 	5440 54.4 %; 	5485 54.8 %; 	5440 54.4 %; 	5440 54.4 %; 	5415 54.1 %; 	5440 54.4 %; 	5480 54.8 %; 	5440 54.4 %

a 130 %: 	4465 44.6 %; 	4436 44.3 %; 	4465 44.6 %; 	4466 44.6 %; 	4431 44.3 %; 	4465 44.6 %; 	4441 44.4 %; 	4421 44.2 %
b 140 %: 	5440 54.4 %; 	5470 54.7 %; 	5440 54.4 %; 	5440 54.4 %; 	5475 54.7 %; 	5440 54.4 %; 	5465 54.6 %; 	5440 54.4 %

a 150 %: 	4465 44.6 %; 	4446 44.4 %; 	4465 44.6 %; 	4466 44.6 %; 	4426 44.2 %; 	4465 44.6 %; 	4451 44.5 %; 	4421 44.2 %
b 160 %: 	5440 54.4 %; 	5460 54.6 %; 	5440 54.4 %; 	5440 54.4 %; 	5480 54.8 %; 	5440 54.4 %; 	5455 54.5 %; 	5440 54.4 %

a 250 %: 	4560 45.6 %; 	4565 45.6 %; 	4656 46.5 %; 	4670 46.7 %; 	4674 46.7 %; 	4465 44.6 %; 	4501 45.0 %; 	4421 44.2 %
b 260 %: 	5344 53.4 %; 	5340 53.4 %; 	5247 52.4 %; 	5234 52.3 %; 	5229 52.2 %; 	5440 54.4 %; 	5405 54.0 %; 	5440 54.4 %

a 350 %: 	4656 46.5 %; 	4665 46.6 %; 	4656 46.5 %; 	4675 46.7 %; 	4594 45.9 %; 	4278 42.7 %; 	4434 44.3 %; 	4252 42.5 %
b 360 %: 	5247 52.4 %; 	5239 52.3 %; 	5247 52.4 %; 	5229 52.2 %; 	5310 53.1 %; 	5629 56.2 %; 	5474 54.7 %; 	5629 56.2 %

a 10 %: 	1770 17.7 %; 	1780 17.8 %; 	1770 17.7 %; 	1780 17.8 %; 	1780 17.8 %; 	1770 17.7 %; 	1780 17.8 %; 	1780 17.8 %
b 50 %: 	8170 81.7 %; 	8164 81.6 %; 	8170 81.7 %; 	8164 81.6 %; 	8164 81.6 %; 	8170 81.7 %; 	8164 81.6 %; 	8170 81.7 %

a 30 %: 	1770 17.7 %; 	1780 17.8 %; 	1770 17.7 %; 	1780 17.8 %; 	1780 17.8 %; 	1770 17.7 %; 	1780 17.8 %; 	1780 17.8 %
b 70 %: 	8170 81.7 %; 	8164 81.6 %; 	8170 81.7 %; 	8164 81.6 %; 	8164 81.6 %; 	8170 81.7 %; 	8164 81.6 %; 	8170 81.7 %

a 50 %: 	1770 17.7 %; 	1780 17.8 %; 	1770 17.7 %; 	1780 17.8 %; 	1780 17.8 %; 	1770 17.7 %; 	1780 17.8 %; 	1780 17.8 %
b 90 %: 	8170 81.7 %; 	8164 81.6 %; 	8170 81.7 %; 	8164 81.6 %; 	8164 81.6 %; 	8170 81.7 %; 	8164 81.6 %; 	8170 81.7 %

a 70 %: 	2415 24.1 %; 	2355 23.5 %; 	3160 31.6 %; 	3170 31.7 %; 	3153 31.5 %; 	2080 20.8 %; 	2095 20.9 %; 	2060 20.6 %
b 110 %: 	7515 75.1 %; 	7579 75.7 %; 	6760 67.6 %; 	6754 67.5 %; 	6769 67.6 %; 	7855 78.5 %; 	7844 78.4 %; 	7855 78.5 %

a 90 %: 	4005 40.0 %; 	3745 37.4 %; 	3160 31.6 %; 	3170 31.7 %; 	3153 31.5 %; 	2080 20.8 %; 	2165 21.6 %; 	2060 20.6 %
b 130 %: 	5905 59.0 %; 	6169 61.6 %; 	6760 67.6 %; 	6754 67.5 %; 	6769 67.6 %; 	7855 78.5 %; 	7774 77.7 %; 	7855 78.5 %

a 110 %: 	4465 44.6 %; 	4150 41.5 %; 	3160 31.6 %; 	3170 31.7 %; 	3153 31.5 %; 	2080 20.8 %; 	2235 22.3 %; 	2060 20.6 %
b 150 %: 	5440 54.4 %; 	5759 57.5 %; 	6760 67.6 %; 	6754 67.5 %; 	6769 67.6 %; 	7855 78.5 %; 	7704 77.0 %; 	7855 78.5 %

a 130 %: 	4465 44.6 %; 	4160 41.6 %; 	3160 31.6 %; 	3170 31.7 %; 	3153 31.5 %; 	2080 20.8 %; 	2305 23.0 %; 	2060 20.6 %
b 170 %: 	5440 54.4 %; 	5749 57.4 %; 	6760 67.6 %; 	6754 67.5 %; 	6769 67.6 %; 	7855 78.5 %; 	7634 76.3 %; 	7855 78.5 %

a 230 %: 	4560 45.6 %; 	4278 42.7 %; 	4005 40.0 %; 	4017 40.1 %; 	3999 39.9 %; 	2080 20.8 %; 	2655 26.5 %; 	2060 20.6 %
b 270 %: 	5344 53.4 %; 	5630 56.3 %; 	5905 59.0 %; 	5897 58.9 %; 	5912 59.1 %; 	7855 78.5 %; 	7284 72.8 %; 	7855 78.5 %

a 130 %: 	4465 44.6 %; 	3305 33.0 %; 	2080 20.8 %; 	2192 21.9 %; 	2095 20.9 %; 	45 0.4 %; 	949 9.4 %; 	35 0.3 %
b 270 %: 	5440 54.4 %; 	6614 66.1 %; 	7855 78.5 %; 	7757 77.5 %; 	7843 78.4 %; 	9945 99.4 %; 	9042 90.4 %; 	9945 99.4 %

a 80 %: 	3160 31.6 %; 	3163 31.6 %; 	3570 35.7 %; 	3533 35.3 %; 	3611 36.1 %; 	3570 35.7 %; 	3533 35.3 %; 	3533 35.3 %
b 100 %: 	6760 67.6 %; 	6759 67.5 %; 	6345 63.4 %; 	6384 63.8 %; 	6305 63.0 %; 	6345 63.4 %; 	6384 63.8 %; 	6345 63.4 %

a 80 %: 	3160 31.6 %; 	3163 31.6 %; 	4005 40.0 %; 	4008 40.0 %; 	4006 40.0 %; 	3486 34.8 %; 	3459 34.5 %; 	3367 33.6 %
b 101 %: 	6760 67.6 %; 	6759 67.5 %; 	5905 59.0 %; 	5904 59.0 %; 	5905 59.0 %; 	6430 64.3 %; 	6459 64.5 %; 	6430 64.3 %

a 79 %: 	3081 30.8 %; 	2997 29.9 %; 	3486 34.8 %; 	3367 33.6 %; 	3443 34.4 %; 	3486 34.8 %; 	3367 33.6 %; 	3367 33.6 %
b 100 %: 	6840 68.4 %; 	6927 69.2 %; 	6430 64.3 %; 	6552 65.5 %; 	6475 64.7 %; 	6430 64.3 %; 	6552 65.5 %; 	6430 64.3 %
Normal is as described, but no halving for high skills
Crits makes a crit beat a non-crit or lower crit
Halved includes halving only
Decrement is lower same amount until both under 100
Crit pre means crits count, beat a non crit or lower crit, and are based on unreduced (for halving or decrement) skill
Crit post makes crit beat non crit or lower crit, and are based upon adjusted skill (after halving or decrementing).

and here's the source. asi and bsi are the variables for skill input. note Python is indent/tab sensitive
Code:
# RQ Rolls
# William F. Hostman
# AKAramis
listvar=[[10,20],[30,40],[50,60],[80,90]]
print "Skills: \tNormal;     \tCrits;       \tHalved;     \tHlvd Cr Pre;",
print "\tHlvd Crt Pst; \tDecrement;    \tDec, crit pre; \tdec crit post. " 
for lvar in listvar:
	asi = lvar[0]
	bsi = lvar[1]
	aw = 0
	bw = 0
	awc = 0
	bwc = 0
	bwh = 0
	awh = 0
	bwhc = 0
	awhc = 0
	bwch = 0
	awch = 0
	awd = 0
	bwd = 0
	awcd = 0
	bwcd = 0
	awdc = 0
	bwdc = 0
	hm = 5
	rt = 0
	as = asi
	am = 95
	if asi >= 200:
		am= 96
		if asi >= 300:
			am=97
			if asi >= 400:
				am=98
				if asi >= 500:
					am=99
	if asi > 95:
		as = am
	ac = asi/10
	if as < 5:
		as = 5
	bs = bsi
	bm = 95
	if asi >= 200:
		am= 96
		if bsi >= 300:
			bm=97
			if bsi >= 400:
				bm=98
				if bsi >= 500:
					bm=99
	bc = bsi/10
	if bs < 5:
		bs = 5
	#halvings
	ah = asi
	bh = bsi
	while (ah > 100) or (bh > 100):
		ah = ah / 2
		bh = bh / 2
		hm = hm -1
	if ah < hm:
		ah = hm
	if bh < hm:
		bh = hm
	if ah > am:
		ah = am
	if bh > bm:
		bh = bm
	ahc = ah/10
	bhc = bh/10
	# decrement
	ad = asi
	bd = bsi
	while ad >100:
		ad = ad -1
		bd = bd -1
	while bd > 100:
		ad = ad -1
		bd = bd -1
	if ad < 5:
		ad = 5
	if bd < 5:
		bd = 5
	adc = ad /10
	bdc = bd/10
	if ad > am:
		ad = am
	if bd > bm:
		bd = bm	
	for a in range(1,101):
		for b in range(1,101):
			# establish normal rolls
			ar = 0
			br = 0
			if a <= as:
				ar = a
			else:
				ar = as - a
			if b <= bs:
				br = b
			else:
				br = bs - b
			# establish halving rolls
			if a <= ah:
				arh = a
			else:
				arh = ah - a
			if b <= bh:
				brh = b
			else:
				brh = bh - b
			# establish decrement rolls
			if a <= ad:
				ard = a
			else:
				ard = ad - a
			if b <= bd:
				brd = b
			else:
				brd = bd - b
			
			# establish crit rolls and crit before halving rolls
			arc = ar
			arch = arh
			arcd = ard
			brc = br
			brch = brh
			brcd = brd
			if a <= ac:
				arc = arc + 100
				arch = arch +100
				arcd = arcd + 100
			if b <= bc:
				brc = brc + 100
				brch = brch+100
				brcd = brcd + 100
			# establish crit after halving rolls
			arhc = arh
			brhc = brh
			ardc = ard
			brdc = brd
			if a <= ahc:
				arhc = arhc +100
			if b <= bhc:
				brhc = brhc + 100
			if a <= adc:
				ardc = ardc +100
			if b <= bdc:
				brdc = brdc + 100
			#check wins
			# - normal
			if ar > br:
				aw = aw + 1
			if br > ar:
				bw = bw + 1
			# - crit
			if arc > brc:
				awc = awc + 1
			if brc > arc:
				bwc = bwc + 1
			# - Halved no crit
			if arh > brh:
				awh = awh + 1
			if brh > arh:
				bwh = bwh + 1
			# - Crit on unmodified, halved skill
			if arch > brch:
				awch = awch + 1
			if brch > arch:
				bwch = bwch + 1
			# - Crit on halved
			if arhc > brhc:
				awhc = awhc + 1
			if brhc > arhc:
				bwhc = bwhc + 1	
			# Decrement wins
			if ard > brd:
				awd = awd + 1
			if brd > ard:
				bwd = bwd + 1
			# Decrement wins, crit pre
			if arcd > brcd:
				awcd = awcd + 1
			if brcd > arcd:
				bwcd = bwcd + 1
			# Decrement wins, crit post
			if ardc > brdc:
				awdc = awdc + 1
			if brd > ard:
				bwdc = bwdc + 1
				
			
			rt=rt+1
			
	print
	print "a",asi,"%: \t",aw,(1000*aw/rt)*0.1,"%; \t", awc,(1000*awc/rt)*0.1,"%; \t",
	print awh,(1000*awh/rt)*0.1,"%; \t", awch,(1000*awch/rt)*0.1,"%; \t", awhc,(1000*awhc/rt)*0.1,"%; \t",
	print awd,(1000*awd/rt)*0.1,"%; \t", awcd,(1000*awcd/rt)*0.1,"%; \t", awdc,(1000*awdc/rt)*0.1,"%"
	print "b",bsi,"%: \t",bw,(1000*bw/rt)*0.1,"%; \t", bwc,(1000*bwc/rt)*0.1,"%; \t",
	print bwh,(1000*bwh/rt)*0.1,"%; \t", bwch,(1000*bwch/rt)*0.1,"%; \t", bwhc,(1000*bwhc/rt)*0.1,"%; \t",
	print bwd,(1000*bwd/rt)*0.1,"%; \t", bwcd,(1000*bwcd/rt)*0.1,"%; \t", bwdc,(1000*bwdc/rt)*0.1,"%"

edit: revised source code to implement failure range, automate processing. Redid stat runs with revision.[/i]
 
I've spotted an anomaly that i don't understand try two runs with these numbers, and check the 'Decrement' column:

A=80; B=100 then compare this to
A=80, B=101

It actualy swaps over from B winning 67.6% of the time to B losing 54% of the time. That can't be right, bearing in mind the two character's skill have only been adjusted down by 1%.


Simon Hibbs
 
Typo an h instead of a d on the b line. Regenerating data, correcting code via edit.

Thanks!

This should be better.

on the listvar = [[10,20],[30,50]]
Each pair is [a,b] and the larger brackets force it to be a single list.

More I work with python, better I like it for brute force stats work.

One other note: I am aware the totals do not add up to 100%; ties are ties!
 
Can you give me an example... Lets say I have 200% and I'm rolling against someone with 100%. That would bring it down to 50% against 25% right? I don't understand (I really am bad with math). Won't the 50% skill beat the 25% twice as often? What is skewed? I must not be understanding where the problem lies (is it in the criticals or something?)
 
Don Allen said:
Can you give me an example... Lets say I have 200% and I'm rolling against someone with 100%. That would bring it down to 50% against 25% right? I don't understand (I really am bad with math). Won't the 50% skill beat the 25% twice as often? What is skewed? I must not be understanding where the problem lies (is it in the criticals or something?)

While the percentage difference between the two skills stays the same, your statistical chance to hit goes down. Mongoose sacrificed 100% accurate math for easy of play.

Hyrum.
 
So what is the difference according to that chart up there between someone with a 99% skill vs a 49%, and someone who has a 100% skill vs a 50%?
 
Let me see if I can give you a very quick idea of what is happening with halving. I am going to repeat some things others have done before but in a very quick way.

During an opposed contests a character can either succeed or fail a roll (yeah I'm going very basic, but stick with me). On any roll a character either succeeds or fails (or fumbles, but I'm keeping it simple). Now this result is sort of matrixed with what the other character gets for a result to detmine a winner. This gives you four basic results.

1) Success vs Failure: Easy to determine the outcome

2) Faillure vs. Success: Ditto

3) Success vs Success: High roll wins

4) Failure vs. Failure: Low roll wins


NOw if we look at the first two outcomes, it is cut and dried as to who wins-the one character that succeeded. In the latter two outcomes (both succeed or fail) the winner varies based on who rolled the highest or lowest. This is practically a coin toss, as either side has nearly a 50-50 chance of rolling higher or lower than the other.

SO if we take two characters, Rusk Runerapier @100% and Groggy the Guardsman at 50% we can see that:

Rusk's chance of rolling a success is almost 100% (there is a minimum 5% chance of failure). This reduces the chances of outcomes #2 (Failure vs Success) and #4 (Failure vs Failure) to 5%. So we won't see them too often. So most of our results will be Outcome 1 (Rusk wins) or Outcome 3 (both succed, high roll wins).

Now if Rusk improves his skill to 101%, he gets halved to 50%. This cuts the chance of condtions 1 & 3 in half, and thus increases the chances of condtion 2 & 4, thus reducing Rusks chance of winning.

So by rasing his skill from 100% to 101% Rusk actually reduced his chance of success.

How's that?
 
Don Allen said:
Can you give me an example... Lets say I have 200% and I'm rolling against someone with 100%. That would bring it down to 50% against 25% right? I don't understand (I really am bad with math). Won't the 50% skill beat the 25% twice as often? What is skewed? I must not be understanding where the problem lies (is it in the criticals or something?)

Here's a simplified atttempt to explain why it doesn't work:

The way the opposed skill checks work, the higher of the two numbers wins if both roll under their skill chance. The lower of the two numbers wins if neither of them roll under their skill chance. The problem is that when you compare the relative values of the two skills, you're only comparing what I call the "underskill" component (the relative size of the range at which you roll "under" your skill). That's fine and dandy and works great. The problem is that the second part of the skill opposition rules involves what I call the "overskill" range. It's the relative size of the range that one can roll that is "over" their skill.

Prior to halving a 100% skilled person had a very small overskill range (5%), compared to the 50% skilled person having a largish one (50%, right?). When you halve them, the relative sizes of the underskill range stays the same, but the overskill sizes change to 75% and 50% respectively. Thus a 10-1 changes to a 3-2. That's a *huge* shift of odds. Remember, that if both roll in either the underskill range or the overskill range, it's that resulting ratio that's going to determine the odds of winning.

Additionally, by halving the skill, you increase the odds of both rolling in the overskill range in the first place (dramatically in the case of the higher skilled person). Which means that the ratio I mentioned above becomes a larger determinant of overall success then it would otherwise have been.

What this causes is a situation where one can become "worse" at an opposed skill check by increasing his skill, if that increase pushes him over a boundary at which he'll have to halve his skill. Actually, not "can become worse", but "will become worse". It's a mathmatical certainty that you're better off having a 95% then a 105% (assuming you have the higher skill of course!).

Does that make a bit more sense?
 
Yes, I get it now. I rolled 10 times each for 2 combatants and the first set of 10 (with 100% skill) that combatant won ALL of the tests vs. the 50% guy. But keeping all of the same rolls but halfing the percents (50% and 25%) the 25% guy won 3 out of the 10 tests. Yikes.

Why even have over 100% in the first place. Why not cap it at 100%. I mean if you have a hundred percent chance of doing something then you automatically do it right?

Either that or have the Active participant subtract the passive (defender etc) participants skill level from his own and roll against that. 150% vs 75% = 75% chance (150-75).

I dont care much for mathematical gymnastics during a game so whatever fix I use it needs to be pretty simple to work out in the heat of battle. I like to keep things fast and loose.
 
Let's see...

100 vs 50

Code:
Skills:	Normal;	Crits;	Halved;	HlvdCrPre;	HlvdCrtPst;	Decrement;  	Dec,critpre;	deccritpost.
50/100:	N 14.8/84.6;	C 14.7/84.7;	H 14.8/84.6;	CH 14.7/84.7;	HC 14.7/84.7;	D 14.8/84.6;	CD 14.7/84.7;	DC 14.7/84.6
25/50:	N 27.7/71.5;	C 27.5/71.7;	H 27.7/71.5;	CH 27.5/71.7;	HC 27.5/71.7;	D 27.7/71.5;	CD 27.5/71.7;	DC 27.5/71.5
52/104:	N 15.9/83.4;	C 15.6/83.8;	H 27.0/72.2;	CH 27.1/72.1;	HC 26.8/72.4;	D 13.7/85.6;	CD 13.8/85.6;	DC 12.9/85.6
26/52:	N 27.0/72.2;	C 26.8/72.4;	H 27.0/72.2;	CH 26.8/72.4;	HC 26.8/72.4;	D 27.0/72.2;	CD 26.8/72.4;	DC 26.8/72.2


Shown are the percentages for the cases you've put forth, showing jusr pecrentages of wins.

Don't forget that 96-00 is a fail, so long as skill is under 200...

Given that in 50 vs 96-100, there are 55 possible ties (1 to 50, 96-00)
in 25 vs 50, there are 75 possible ties (01-25,51-00).

I've run the numbers enough to know that I will most likely use Halving, crit pre.

It's "Close Enough" for me. And it handles a far wider range than does decrementing. YMMV.
 
And if anyone is still in doubt, here's the chances for skill 1 vs skill 2 with skill 1 going in increments of 100 and skill 2 being 50

Code:
Skill 1    Skill 2    1 Wins     2 Wins    Tie
-------    -------    ------     ------    -------
100        50         82.35%     17.1%     0.55%
200        50         91.85%     7.85%     0.3%
300        50         77.22%     22.41%    0.37%
400        50         94.32%     5.51%     0.17%
500        50         68.82%     30.74%    0.44%
600        50         77.79%     21.9%     0.31%
700        50         87.57%     12.24%    0.19%
800        50         94.89%     5%        0.11%
900        50         65.31%     34.2%     0.49%
1000       50         68.88%     30.69%    0.43%

One interesting thing here is that the halving rule is valid for skills that resolve down to 100, and will maintain tolerable accuracy near enough to 100 (depending on what you consider "tolerable", of course). But the further below 100 the skill resolves to, the bigger the blip.

Another thing is that the chance of the highest winning is going to be within a few percent of what the highest resolves down to. In the extreme case of 101 vs 1, it's 12 percent.
 
Just a general point on skill comparisons. An 'average' combat character will probably end up having about an 80% weapon skill using standard character generation. That's therefore quite a good base skill to compare against. So to best see the statistical swings, I think comparing a range of abiolities (90%, 100% 110%, 120%...) against an opponent at 80% gives a good indication of relative advancement.

Simon Hibbs
 
simonh said:
Just a general point on skill comparisons. An 'average' combat character will probably end up having about an 80% weapon skill using standard character generation. That's therefore quite a good base skill to compare against. So to best see the statistical swings, I think comparing a range of abiolities (90%, 100% 110%, 120%...) against an opponent at 80% gives a good indication of relative advancement.

As combat doesn't actually use the halving rule, this may not be a valid approach, but I calculated the 100 to 110 stuff anyway, and there's a drop from 62.7% to 56.7%.

One other thought - "both fail lowest wins" is also a flawed mechanic, as it penalises the person with the higher skill. This is, IMO, even worse than halving, as it is far more likely to occur in general situations.

Switching to "both fail highest wins" is actually doubly beneficial as it mitigates the blip from the halving rule quite a bit. 100 vs 1 (99.45% win) and 101 vs 1 (84.6% win) is nowhere near as bad.
 
Don Allen said:
You don't use the halfing rule during combat? Why not? It says dodging etc is an opposed roll vs. attackers weapon skill.
Check out the players guide PDF, it explains all.

Although the main site appears to be broken at the moment. :evil:
 
Back
Top