Also, in case you're wondering, reflex save DC 14 to avoid crit doesn't result in crazy weirdness, either. (A house rule I convinced my own DM out of).
The best critical paradigm I've seen is
1) Only crit on first attack of full attack sequence AND
2) Reflex save DC 14/some sort of crit confirmation
It makes the odds of failure low and solves much of the numbers messiness.
Also, if I'm a 20th level expert TWF making some 10 attacks a round (which isn't tough), my odds of fumbling are ~50% per round. I fumble every other round. The 1st level incompetent two handed fighter fumbles 1/10th that the 20th level does
Here's some math on the reflex save house rule.
[spoilerbutton=Dextrous Greatsword Wielder]
A relatively dextrous fighter (14 dex) who wields a greatsword, and gets boots of speed at 10, and Cloak of resistance +1 at 5, +2 at 10, +3 at 15 and +4 at 20.
Odds of critting and failing the save in any single round
Base rate (5% per attack) * odds of failing save.
Level 1: 5% --- 55% (+2) = 2.75% odds of critting failing per round
Level 2: 5% --- 55% (+2) = 2.75% odds of critting failing per round
Level 3: 5% --- 50% (+3) = 2.5% odds of critting failing per round
Level 4: 5% --- 50% (+3) = 2.5% odds of critting failing per round
Level 5: 5% --- 45% (+4) = 2.25% odds of critting failing per round
Level 6: 10% --- 40% (+5) = 4% odds of critting failing per round
Level 7: 10% --- 40% (+5) = 4% odds of critting failing per round
Level 8: 10% --- 40% (+5) = 3% odds of critting failing per round
Level 9: 10% --- 35% (+6) = 3.5% odds of critting failing per round
Level 10: 15% --- 30% (+7) = 4.5% odds of critting failing per round
Level 11: 20% --- 30% (+7) = 6% odds of critting failing per round
Level 12: 20% --- 25% (+8) = 5.% odds of critting failing per round
Level 13: 20% --- 25% (+8) = 5.% odds of critting failing per round
Level 14: 20% --- 25% (+8) = 5% odds of critting failing per round
Level 15: 20% --- 15% (+10) = 3% odds of critting failing per round
Level 16: 25% --- 15% (+10) = 3.75% odds of critting failing per round
Level 17: 25% --- 15% (+10) = 3.75% odds of critting failing per round
Level 18: 25% --- 10% (+11) = 2.5% odds of critting failing per round
Level 19: 25% --- 10% (+11) = 2.5% odds of critting failing per round
Level 20: 25% --- 5% (+12) = 1.25% odds of critting failing per round
Our relatively dextrous greatsword wielder is least likely to screw up and stab himself in the eye to most likely in this level ordering:
(Safer) -------------------------------------------------------------- (unsafe)
20 >>> 5 >>> 3, 4, 18, 19 >>> 1, 2 >>> 8, 15 >>> 9 >>> 16, 17 >>> 6, 7 >>> 10 >>> 12, 13, 14 >>> 11
As we can see hear, our greatsword fighter is safest in the low (1-5) or high (18-20) level ranges, and most likely to injure himself at the middle ranges (6-7, 10-14). Level 11 is particularly dangerous, where every minute of fighting, said fighter is likely (60%) to screw up and hurt himself, and basically guaranteed every 100 seconds (1 minute, 40 seconds) to injured himself. Contrast that with a novice of the blade, who in a minute is unlikely (27.5%) to injure himself, and is basically guaranteed to injure himself every 220 seconds (3 minutes, 40 seconds); a rate much less frequent than his expert acquaintance.
[/spoilerbutton]
[spoilerbutton=Dual Wielding Elven Ranger Friend]
Let's instead consider a dual wielding elven ranger friend:
Dex 20, +1 at 4, 8, 12, 16, 20. +2 at 5, 10, 15, and +1 saves at 4, 8, 12, 16, 20. Boots of speed at 10 again.
Base rate (5% per attack) * odds of failing save.
Level 1: 10% * 30% (+7) = 3% odds of critical failure per round
Level 2: 10% * 25% (+8) = 2.5% odds of critical failure per round
Level 3: 10% * 25% (+8) = 2.5% odds of critical failure per round
Level 4: 10% * 20% (+9) = 2% odds of critical failure per round
Level 5: 10% * 10% (+11) = 1% odds of critical failure per round
Level 6: 20% * 5% (+12) = 1% odds of critical failure per round
Level 7: 20% * 5% (+12) = 1% odds of critical failure per round
Level 8: 20% * 5% (+15) = 1% odds of critical failure per round
Level 9: 20% * 5% (+15) = 1% odds of critical failure per round
Level 10: 25% * 5% (+18) = 1.25% odds of critical failure per round
Level 11: 30% * 5% (+18) = 1.5% odds of critical failure per round
Level 12: 35% * 5% (+20) = 1.75% odds of critical failure per round
Level 13: 35% * 5% (+20) = 1.75% odds of critical failure per round
Level 14: 35% * 5% (+21) = 1.75% odds of critical failure per round
Level 15: 35% * 5% (+22) = 1.75% odds of critical failure per round
Level 16: 40% * 5% (+25) = 2% odds of critical failure per round
Level 17: 40% * 5% (+25) = 2% odds of critical failure per round
Level 18: 40% * 5% (+26) = 2% odds of critical failure per round
Level 19: 40% * 5% (+26) = 2% odds of critical failure per round
Level 20: 40% * 5% (+28) = 2% odds of critical failure per round
Breaking down this saves chart:
Dexterity: +5 at 1, +6 at 5, +7 at 8, +9 at 10, +10 at 15, +11 at 16. (+2 item at 5/10/15, +2 at 8/16 from levelling up). Item at 4/8/12/16/20.
Safer ------------------------------------------------------------------------- Unsafe
5, 6, 7, 8, 9 ---> 10 ----> 11 >>>>>>> 12, 13, 14, 15 >>>>>> 4, 6, 7, 16, 17, 18, 19, 20 >>> 2, 3 >>> 1
For this character, their best levels for avoiding crits are early-middle (5-11) At their peak (5-9), they only screw up every 10 minutes of fighting, which is exactly what I'd expect from a swords expert...though that goes down to every 5 minutes at their true expertise. At their first level they are indeed least able to successfully fight, with some sort of accident every 200 seconds (3 minutes, 20 seconds).
[/spoilerbutton]
