Iterative Attacks

Wulf Ratbane said:

But-- allowing of course that groups differ-- nobody really finds it "suspenseful" wondering if Bob is going to correctly add 14+9 for once in his friggin' life.

Click to expand...

I like your idea, and I think that I will propose it to my current group, they are all 3rd, almost 4th level now.

As a side note, you must have esp... How did you know that it is Bob (really!) in my group who can't add!!! The man is a genius (really), but he always has trouble with the simple math...

I've arrived very late to the discussion but I think basic 2 attack pattern with decreasing lower penalties is a fine basis for a model...though I'd have to see the math to confirm.

With 2 attacks, there is enough tension and chaoticness to make it fun (hitting with 0,1,or 2 attacks...and of course the various crits possibilities within). Its still quick to run, and still provides the fighter enough big damage.

I've been doing something similar with firearms in my d20 Modern gaming, and bows in D&D. As a DM who uses a lot of humanoid opponents, I would love having to make less attack rolls and not having to add a different amount to each.

Stalker0 said:

I've arrived very late to the discussion but I think basic 2 attack pattern with decreasing lower penalties is a fine basis for a model...though I'd have to see the math to confirm.

Click to expand...

Stalker, I'll be happy to email you my excel sheet if you drop me a line.

It's a mess at this point and perhaps not terribly likely to make a lot of sense without explanation, but you're welcome to it.

EDIT: Alternately, I can print the long form results (without cell calculations) into a PDF and post them. Anything folks would like other than trying to line up tables with vbcode.

Wulf Ratbane said:

Walk me through that in the case of criticals.

If you roll a natural 20, you automatically get the full "iterative attack" multiplier. You also score a critical threat-- does the confirmation roll also enjoy the same benefits on a natural 20? In other words, does a 20 followed by a 20 score the full iterative multiplier, with the "base damage" equal to your critical damage roll?

Click to expand...

Yes, the multiplier is applied after any critical damage has been determined. However, you've made me realize that I neglected to account for the confirmation rolls when I came up with this fix.

To do this, you'd need to apply separate multipliers to the standard damage and the critical damage. The crit damage multiplier would be determined based on how much the confirmation roll beat the AC by (5 for 2x, 10 for 3x, 15 for 4x). Personally though, I think the complication this adds is not worth it, since confirmed crits on iterative attacks are rare enough that they have little effect on the expected damage.

roguerouge said:

Gantros: my DMs typically don't tell me the AC of what I'm attacking. So doesn't this undermine the time-saving practice of pre-rolling?

Click to expand...

The level of success method should still work fine with pre-rolling and secret ACs. Basically you can just look up your attack roll, add up any applicable modifiers, and tell the DM what AC you hit. The DM then compares that against the target's AC, determines how much you beat it by, and then informs you if any multipliers apply. If they do, then you just look up your pre-rolled damage and apply them. It should still be faster than adding up multiple attack modifiers and damage rolls, and the DM need not tell you the opponent's AC (though the multiplier might allow you to approximate it).

Gantros, do you allow the attack multiplier if the combatant has moved?

Click to expand...

Not really. I use the RAW for full attacks, which state that the only movement allowed is a 5-foot step before or after the attack. You can also take the step in between attacks against multiple opponents.

And while I am thinking of it, do you do anything different with multiple attacks? (Claw/Claw/Bite routines and the like.)

Click to expand...

No. Natural weapon attack routines have too many variables to make something like levels of success practical.

I'm worried that levels of success, especially with the extra multipliers on a natural 20, is going to be too swingy in combat.

True, one consequence of rolling less dice is that the results are always swingier. This is why I recommend ignoring the fixes I threw out there and sticking with the original level of success rule.

The problem with any rule that requires a lot of dice rolls is that the more you roll, the lower the probability of getting an extreme result. For example, if a 20th level mage casts a fireball, they could either roll 20d6 for damage, or 4d6x5. Both have the same damage expectation, while the latter is obviously swingier. However, even though the odds of an extreme result (i.e. 120 damage) improve from 1 in 3.6 quadrillion to a mere 1 in 1300, results like that are still rare enough to justify the time savings of rolling and adding 4 dice each time vs. 20.

Applying this logic to the level of success method, you can see that even though it can theoretically reduce the expected damage by up to 15% vs. the RAW against high AC opponents, this number is being skewed by those very rare cases where you roll natural 20s on multiple iterative attacks in a single round. If you look at the math, it will actually result in an identical amount of damage at least 96% of the time, and that's for the worst case where you need to roll a 16 or higher to hit. In most other cases, level of success gives you the same expected damage as the RAW closer to 100% of the time.

Add in the fact that high-level fighters are not going to encounter too many situations where they need such high rolls to hit with their primary attack, and you can see that the difference becomes largely irrelevant in practice.

Incidentally, this reminds me of an attempt I made a while back to estimate how many rolls are made in a typical game session. The goal was to figure out how frequently you could expect to see a highly unlikely result over a given period of time. I don't remember all the assumptions I used, but here's roughly what I came up with:

1 in 10 - at least once per encounter
1 in 100 - at least once per adventure
1 in 1,000 - at least once per campaign
1 in 10,000 - at least once in a lifetime

In other words, if you're rolling more than 5 d6s/d8s, 4 d10s/d12s, or 3 d20s to determine any outcome, you're likely never going to see an extreme result in a lifetime of gaming. This helped put in perspective how many dice get rolled in a typical game that have no meaningful bearing on the outcome.

At the end of the day, d20 combat is about:
- delivering initial sequence of hits via iterative and/or natural attacks
- delivering secondary sequence of hits via confirmed criticals (confirming of a critical is a duplicated initial hit which results in delivery of usually multiple additional hits)

In order to replace this with a single roll, one would have to create a system for replicating this.

Having given this a bit of thought, I'd say that in order to prevent swinginess of combat, the system would have to be reworked so that each combat takes a larger number of rounds or...

Or... change the steps for success are calculated. Basically, allow the basic success (i.e. a single hit) to occur below target AC value, and multiple hits to take place above. For example (the step values are not relevant, I have just made the numbers up to make a point):

Whenever you miss target AC by 5 or less, you score single hit.
Whenever you score target AC or exceed by up to 10, you score average number of hits (two attacks mean still single hit, three attacks mean 2 hits).
Whenever you exceed target AC by 11 or more, you score maximum number of hits (two attacks mean two hits, four attacks mean four hits).

To compensate for low number of hits between target AC and target AC +10, the following mechanic is added for additional critical hits:

Whenever you score a critical threat, you automatically deal one additional hit. If you confirm critical threat, add another hit. If the confirmation roll was a natural twenty, multiply the number of hits scored so far by 1.5.

----

The problem with this system (aside from made up step values) are the attacks which deal different amounts of damage (different natural attacks, off-hand weapons). It would be great if the system simply allowed to score a number of hits to be distributed among targets. However, for this one would have to be using Storyteller/Exalted game.

Given complexity of current system, I find it hard to replace with a reliable one-roll task resolution. It's possible to work out probabilities (as Wulf showed us), but it's hard to preserve certain averages (d20 produces flat distribution, with all numbers being given equal probability) without taking a dip of GURPSish multiple d6s (i.e. roll multiple d6 and then add them up to meet a target number).

It would be possible to create a table with various success levels (d20 vs [AC - attack]: find a match to learn the number of hits)... hmm, maybe such table would be the best way. Who knows?

Regards,
Ruemere

Gantros said:

Applying this logic to the level of success method, you can see that even though it can theoretically reduce the expected damage by up to 15% vs. the RAW against high AC opponents, this number is being skewed by those very rare cases where you roll natural 20s on multiple iterative attacks in a single round. If you look at the math, it will actually result in an identical amount of damage at least 96% of the time, and that's for the worst case where you need to roll a 16 or higher to hit. In most other cases, level of success gives you the same expected damage as the RAW closer to 100% of the time.

Click to expand...

Can we agree to a minimum probability below which we throw out any damage expectation that would affect the average?

1 in 400? (That's two consecutive natural 20's.)

That's an interesting question, and I'm sure people would have differing opinions. Myself, I would set the bar higher, since I'd gladly give up a single chance to do double damage in return for 400 fewer dice rolls. I don't play lotteries either, because I don't think the time and money spent buying tickets justifies the tiny chance of winning, but clearly many other people do.

My gut feel says I'd need at least a 1 in 100 chance, maybe even 1 in 50, of doing significant bonus damage before I'd consider it worthwhile to bother checking for it. Sure it might have a big impact on the game if it came up at just the right moment, but it could just as easily come up in an irrelevant situation (e.g. against a mook that was already almost dead), or be negated by a low damage roll.