BettaGeorge
2021-03-06, 06:21 AM
Am I the only one who feels that the way 3.5 handles initiative is flawed? What I mean is the following: say you have a non-minmaxed rogue with reasonably high DEX who spends one of their precious feats on Improved Initiative. In my experience, that character still regularly goes after all enemies in combat, denying them that sweet sweet first-round sneak attack. The reason being that while you may have a modifier of 7 to 9 on your initiative (provided you don't build your character explicitly to have high ini and nothing else), the outcome still depends on a die roll with a range of 20, more than double that modifier. If you roll on the low side and your opponent on the high side, your modifier is worth nothing.
Suspecting myself to be a victim of Negativity Bias, I decided to run the numbers. (Math PhD for the win, I guess.) We pit a combatant with low initiative against one with high initiative. Note that since only the difference in initiative modifiers affects the probabilities, I simply assume the low initiative fighter to have an ini bonus of 0. (The probabilities for ini bonus 0 vs 5 are the same as those for 1 vs 6, 2 vs 7, and so on.)
Bonuses of 0 vs 0 give us probabilities of 47.5000% win, 47.5000% lose, 5.0000% draw.
Bonuses of 0 vs 1 give us probabilities of 42.7500% win, 52.5000% lose, 4.7500% draw.
Bonuses of 0 vs 2 give us probabilities of 38.2500% win, 57.2500% lose, 4.5000% draw.
Bonuses of 0 vs 3 give us probabilities of 34.0000% win, 61.7500% lose, 4.2500% draw.
Bonuses of 0 vs 4 give us probabilities of 30.0000% win, 66.0000% lose, 4.0000% draw.
Bonuses of 0 vs 5 give us probabilities of 26.2500% win, 70.0000% lose, 3.7500% draw.
Bonuses of 0 vs 6 give us probabilities of 22.7500% win, 73.7500% lose, 3.5000% draw.
Bonuses of 0 vs 7 give us probabilities of 19.5000% win, 77.2500% lose, 3.2500% draw.
Bonuses of 0 vs 8 give us probabilities of 16.5000% win, 80.5000% lose, 3.0000% draw.
Bonuses of 0 vs 9 give us probabilities of 13.7500% win, 83.5000% lose, 2.7500% draw.
Bonuses of 0 vs 10 give us probabilities of 11.2500% win, 86.2500% lose, 2.5000% draw.
Bonuses of 0 vs 11 give us probabilities of 9.0000% win, 88.7500% lose, 2.2500% draw.
Bonuses of 0 vs 12 give us probabilities of 7.0000% win, 91.0000% lose, 2.0000% draw.
Bonuses of 0 vs 13 give us probabilities of 5.2500% win, 93.0000% lose, 1.7500% draw.
Bonuses of 0 vs 14 give us probabilities of 3.7500% win, 94.7500% lose, 1.5000% draw.
Bonuses of 0 vs 15 give us probabilities of 2.5000% win, 96.2500% lose, 1.2500% draw.
Bonuses of 0 vs 16 give us probabilities of 1.5000% win, 97.5000% lose, 1.0000% draw.
Bonuses of 0 vs 17 give us probabilities of 0.7500% win, 98.5000% lose, 0.7500% draw.
Bonuses of 0 vs 18 give us probabilities of 0.2500% win, 99.2500% lose, 0.5000% draw.
Bonuses of 0 vs 19 give us probabilities of 0.0000% win, 99.7500% lose, 0.2500% draw.
Bonuses of 0 vs 20 give us probabilities of 0.0000% win, 100.0000% lose, 0.0000% draw.
Adjusted for the fact that on a draw, the higher initiative modifier goes first, we get:
Bonuses of 0 vs 1 give us probabilities of 42.7500% win, 57.2500% lose.
Bonuses of 0 vs 2 give us probabilities of 38.2500% win, 61.7500% lose.
Bonuses of 0 vs 3 give us probabilities of 34.0000% win, 66.0000% lose.
Bonuses of 0 vs 4 give us probabilities of 30.0000% win, 70.0000% lose.
Bonuses of 0 vs 5 give us probabilities of 26.2500% win, 73.7500% lose.
Bonuses of 0 vs 6 give us probabilities of 22.7500% win, 77.2500% lose.
Bonuses of 0 vs 7 give us probabilities of 19.5000% win, 80.5000% lose.
Bonuses of 0 vs 8 give us probabilities of 16.5000% win, 83.5000% lose.
Bonuses of 0 vs 9 give us probabilities of 13.7500% win, 86.2500% lose.
Bonuses of 0 vs 10 give us probabilities of 11.2500% win, 88.7500% lose.
Bonuses of 0 vs 11 give us probabilities of 9.0000% win, 91.0000% lose.
Bonuses of 0 vs 12 give us probabilities of 7.0000% win, 93.0000% lose.
Bonuses of 0 vs 13 give us probabilities of 5.2500% win, 94.7500% lose.
Bonuses of 0 vs 14 give us probabilities of 3.7500% win, 96.2500% lose.
Bonuses of 0 vs 15 give us probabilities of 2.5000% win, 97.5000% lose.
Bonuses of 0 vs 16 give us probabilities of 1.5000% win, 98.5000% lose.
Bonuses of 0 vs 17 give us probabilities of 0.7500% win, 99.2500% lose.
Bonuses of 0 vs 18 give us probabilities of 0.2500% win, 99.7500% lose.
Bonuses of 0 vs 19 give us probabilities of 0.0000% win, 100.0000% lose.
Bonuses of 0 vs 20 give us probabilities of 0.0000% win, 100.0000% lose.
So if my DEX equals my opponent's but I have spent a feat on Improved Initiative, that raises my chances of going first from 57% to 70%. Not sure that is worth a feat. It even gets worse as my DEX goes up – the higher my DEX is, the fewer percentage points I get out of Improved Initiative.
If my initiative is 10 higher than my opponent's (which requires quite a few feats or items, or an opponent with abysmal DEX), I still have a higher than 10% chance of going last.
All in all, those numbers aren't quite as bad as I had imagined them in my head, but they still confirm that I never ever want to take Improved Initiative again.
Now, I'm not saying that there shouldn't be some randomness to the initiative order. I just feel like the game designers didn't quite think through how much randomness they introduced there.
One idea I had was to just roll a d10 instead of a d20 (ignoring for the sake of argument that initiative is actually a DEX check, not just a die roll). The resulting probabilities are a lot more favorable for high-initiative builds:
Bonuses of 0 vs 1 give us probabilities of 45.0000% win, 55.0000% lose.
Bonuses of 0 vs 1 give us probabilities of 36.0000% win, 64.0000% lose.
Bonuses of 0 vs 2 give us probabilities of 28.0000% win, 72.0000% lose.
Bonuses of 0 vs 3 give us probabilities of 21.0000% win, 79.0000% lose.
Bonuses of 0 vs 4 give us probabilities of 15.0000% win, 85.0000% lose.
Bonuses of 0 vs 5 give us probabilities of 10.0000% win, 90.0000% lose.
Bonuses of 0 vs 6 give us probabilities of 6.0000% win, 94.0000% lose.
Bonuses of 0 vs 7 give us probabilities of 3.0000% win, 97.0000% lose.
Bonuses of 0 vs 8 give us probabilities of 1.0000% win, 99.0000% lose.
Bonuses of 0 vs 9 give us probabilities of 0.0000% win, 100.0000% lose.
Bonuses of 0 vs 10 give us probabilities of 0.0000% win, 100.0000% lose.
What are your thoughts?
Edit: I pasted the wrong numbers like a doofus. They are correct now.
Suspecting myself to be a victim of Negativity Bias, I decided to run the numbers. (Math PhD for the win, I guess.) We pit a combatant with low initiative against one with high initiative. Note that since only the difference in initiative modifiers affects the probabilities, I simply assume the low initiative fighter to have an ini bonus of 0. (The probabilities for ini bonus 0 vs 5 are the same as those for 1 vs 6, 2 vs 7, and so on.)
Bonuses of 0 vs 0 give us probabilities of 47.5000% win, 47.5000% lose, 5.0000% draw.
Bonuses of 0 vs 1 give us probabilities of 42.7500% win, 52.5000% lose, 4.7500% draw.
Bonuses of 0 vs 2 give us probabilities of 38.2500% win, 57.2500% lose, 4.5000% draw.
Bonuses of 0 vs 3 give us probabilities of 34.0000% win, 61.7500% lose, 4.2500% draw.
Bonuses of 0 vs 4 give us probabilities of 30.0000% win, 66.0000% lose, 4.0000% draw.
Bonuses of 0 vs 5 give us probabilities of 26.2500% win, 70.0000% lose, 3.7500% draw.
Bonuses of 0 vs 6 give us probabilities of 22.7500% win, 73.7500% lose, 3.5000% draw.
Bonuses of 0 vs 7 give us probabilities of 19.5000% win, 77.2500% lose, 3.2500% draw.
Bonuses of 0 vs 8 give us probabilities of 16.5000% win, 80.5000% lose, 3.0000% draw.
Bonuses of 0 vs 9 give us probabilities of 13.7500% win, 83.5000% lose, 2.7500% draw.
Bonuses of 0 vs 10 give us probabilities of 11.2500% win, 86.2500% lose, 2.5000% draw.
Bonuses of 0 vs 11 give us probabilities of 9.0000% win, 88.7500% lose, 2.2500% draw.
Bonuses of 0 vs 12 give us probabilities of 7.0000% win, 91.0000% lose, 2.0000% draw.
Bonuses of 0 vs 13 give us probabilities of 5.2500% win, 93.0000% lose, 1.7500% draw.
Bonuses of 0 vs 14 give us probabilities of 3.7500% win, 94.7500% lose, 1.5000% draw.
Bonuses of 0 vs 15 give us probabilities of 2.5000% win, 96.2500% lose, 1.2500% draw.
Bonuses of 0 vs 16 give us probabilities of 1.5000% win, 97.5000% lose, 1.0000% draw.
Bonuses of 0 vs 17 give us probabilities of 0.7500% win, 98.5000% lose, 0.7500% draw.
Bonuses of 0 vs 18 give us probabilities of 0.2500% win, 99.2500% lose, 0.5000% draw.
Bonuses of 0 vs 19 give us probabilities of 0.0000% win, 99.7500% lose, 0.2500% draw.
Bonuses of 0 vs 20 give us probabilities of 0.0000% win, 100.0000% lose, 0.0000% draw.
Adjusted for the fact that on a draw, the higher initiative modifier goes first, we get:
Bonuses of 0 vs 1 give us probabilities of 42.7500% win, 57.2500% lose.
Bonuses of 0 vs 2 give us probabilities of 38.2500% win, 61.7500% lose.
Bonuses of 0 vs 3 give us probabilities of 34.0000% win, 66.0000% lose.
Bonuses of 0 vs 4 give us probabilities of 30.0000% win, 70.0000% lose.
Bonuses of 0 vs 5 give us probabilities of 26.2500% win, 73.7500% lose.
Bonuses of 0 vs 6 give us probabilities of 22.7500% win, 77.2500% lose.
Bonuses of 0 vs 7 give us probabilities of 19.5000% win, 80.5000% lose.
Bonuses of 0 vs 8 give us probabilities of 16.5000% win, 83.5000% lose.
Bonuses of 0 vs 9 give us probabilities of 13.7500% win, 86.2500% lose.
Bonuses of 0 vs 10 give us probabilities of 11.2500% win, 88.7500% lose.
Bonuses of 0 vs 11 give us probabilities of 9.0000% win, 91.0000% lose.
Bonuses of 0 vs 12 give us probabilities of 7.0000% win, 93.0000% lose.
Bonuses of 0 vs 13 give us probabilities of 5.2500% win, 94.7500% lose.
Bonuses of 0 vs 14 give us probabilities of 3.7500% win, 96.2500% lose.
Bonuses of 0 vs 15 give us probabilities of 2.5000% win, 97.5000% lose.
Bonuses of 0 vs 16 give us probabilities of 1.5000% win, 98.5000% lose.
Bonuses of 0 vs 17 give us probabilities of 0.7500% win, 99.2500% lose.
Bonuses of 0 vs 18 give us probabilities of 0.2500% win, 99.7500% lose.
Bonuses of 0 vs 19 give us probabilities of 0.0000% win, 100.0000% lose.
Bonuses of 0 vs 20 give us probabilities of 0.0000% win, 100.0000% lose.
So if my DEX equals my opponent's but I have spent a feat on Improved Initiative, that raises my chances of going first from 57% to 70%. Not sure that is worth a feat. It even gets worse as my DEX goes up – the higher my DEX is, the fewer percentage points I get out of Improved Initiative.
If my initiative is 10 higher than my opponent's (which requires quite a few feats or items, or an opponent with abysmal DEX), I still have a higher than 10% chance of going last.
All in all, those numbers aren't quite as bad as I had imagined them in my head, but they still confirm that I never ever want to take Improved Initiative again.
Now, I'm not saying that there shouldn't be some randomness to the initiative order. I just feel like the game designers didn't quite think through how much randomness they introduced there.
One idea I had was to just roll a d10 instead of a d20 (ignoring for the sake of argument that initiative is actually a DEX check, not just a die roll). The resulting probabilities are a lot more favorable for high-initiative builds:
Bonuses of 0 vs 1 give us probabilities of 45.0000% win, 55.0000% lose.
Bonuses of 0 vs 1 give us probabilities of 36.0000% win, 64.0000% lose.
Bonuses of 0 vs 2 give us probabilities of 28.0000% win, 72.0000% lose.
Bonuses of 0 vs 3 give us probabilities of 21.0000% win, 79.0000% lose.
Bonuses of 0 vs 4 give us probabilities of 15.0000% win, 85.0000% lose.
Bonuses of 0 vs 5 give us probabilities of 10.0000% win, 90.0000% lose.
Bonuses of 0 vs 6 give us probabilities of 6.0000% win, 94.0000% lose.
Bonuses of 0 vs 7 give us probabilities of 3.0000% win, 97.0000% lose.
Bonuses of 0 vs 8 give us probabilities of 1.0000% win, 99.0000% lose.
Bonuses of 0 vs 9 give us probabilities of 0.0000% win, 100.0000% lose.
Bonuses of 0 vs 10 give us probabilities of 0.0000% win, 100.0000% lose.
What are your thoughts?
Edit: I pasted the wrong numbers like a doofus. They are correct now.