On the Dodge Action

Ovinomancer said:

I'm an electrical engineer who's made a few posts on this topic and I cannot follow what the chart in that post is doing. It's right at the ends, but very badly wrong in the middle. Whatever it's showing isn't a 'chance to roll this number or higher' which is the critical question of advantage (lower for disadvantage) and it's not chance to roll that exact number (which isn't interesting at all) so...

Click to expand...

Thanks, I realised last night what my mistake was on the earlier post. The one I linked to shows the probability of rolling exactly that number, not greater than that number. I know you say it's not that but if you add all the % under the DC you want you get the same answer in both tables.

Now, please keep in mind that the equivalent bonus is a bad way to think about advantage/disadvantage because you can't reduce a normal distribution to a bonus to a flat distribution, but that's nerd math talk. Just know you're wrong to do it, but it's still kinda handy for a general rule of thumb.

I'm have a Bachelor's of Electrical Engineering (BSEE) from an ABET accredited institution with 10+ years experience in the field of communications technology and a particular interest in statistical analysis. Advantage/disadvantage is not a flat +/- 5, although it resembles such (if you squint and are okay being wrong) between 8 and 12. Given many rolls are in this range for bounded accuracy, it's probably why the designers chose to shorthand it as +5/-5 for passive checks. Makes it easy.

Click to expand...

It's kinda handy to do quick on the fly maths mid-session, as long as you understand how inaccurate it is. A flat bonus is also closer to =/-3 rather than 5. I think the idea of a +/-5 bonus comes passive skill checks getting +5 for advantage and -5 for Advantage.

But going back to the OP's question requires us to look at specific situations, and squinting at highly inaccurate approximations of a bonus won't help with that

TallIan said:

Thanks, I realised last night what my mistake was on the earlier post. The one I linked to shows the probability of rolling exactly that number, not greater than that number. I know you say it's not that but if you add all the % under the DC you want you get the same answer in both tables.

It's kinda handy to do quick on the fly maths mid-session, as long as you understand how inaccurate it is. A flat bonus is also closer to =/-3 rather than 5. I think the idea of a +/-5 bonus comes passive skill checks getting +5 for advantage and -5 for Advantage.

But going back to the OP's question requires us to look at specific situations, and squinting at highly inaccurate approximations of a bonus won't help with that

Click to expand...

I wonder how much heavy lifting “all the time” is doing when we’re evaluating the sentiment “I dodge all the time.”

Once per fight?
Literally every action?
Often enough that it’s saved my bacon a handful of memorable times?

Hand-in-hand I wonder whether the emotional impact of success is a stickier experience than those times someone dodges but then nothing really happens.

I know that’s not a mathematical line of thinking - just seems to me if someone maybe dodged a couple of clutch times and triumphed, they may come out of that with the perspective “I dodge all the time” and “I love it.”

I’ve used the dodge action in most every fight we have had. That’s a lot. Excessively too much would be trying to use it nearly every turn.

smbakeresq said:

If a PC takes the Dodge action and you as DM alter your attack plans against them right away you are meta-gaming the PCs and I don’t think that’s proper.

Click to expand...

That goes back to your interpretation of how you correlate actions in the game to actions within the narrative. It can be hard to reach consensus on what, exactly, is observable to characters in the world; and which information, if acted upon, would be meta-gaming.

I've heard too many horror stories about GMs who wouldn't let a player know how many HP their own character had left, under the theory that HP are meta-game information, for me to risk making the same kind of mistake. In the name of clear communication between DM and players, and in giving characters the benefit of the doubt about what they can observe about the world they actually live in, I assume that everything in the game mechanics correlates to some manner of phenomenon that is directly observable to the characters. So in my games, if someone is so incredibly focused on dodging that they actually get some mechanical effect out of it, then that is in-narrative information which is observable to everyone nearby.

Obviously, if that's not the case at your table, then the Dodge action would be less useful as a deterrent.

Bawylie said:

I wonder how much heavy lifting “all the time” is doing when we’re evaluating the sentiment “I dodge all the time.”

Once per fight?
Literally every action?
Often enough that it’s saved my bacon a handful of memorable times?

Hand-in-hand I wonder whether the emotional impact of success is a stickier experience than those times someone dodges but then nothing really happens.

I know that’s not a mathematical line of thinking - just seems to me if someone maybe dodged a couple of clutch times and triumphed, they may come out of that with the perspective “I dodge all the time” and “I love it.”

Click to expand...

I would refer to seeing something "all the time" in this context of it was basically seen in most every significant vombst, not every action, not every trivial fight etc.

I dont see it that often myself but it is frequent enough it's not forgotten.

I would expect to see it more when tactics are built around it, like the spirit guards spirit wpn dodgey cleric or the AO tanky character. Iirc tunnel fighter from UA gives a fighter unlimited OA as bonus action and can be combo with sentinel if your game allows playtest material - I could easily see a tunnel fighter going fudgey and then BA defensive stance plus sentinel eith glaive bring a thing one would see a lot.

If either of those are in play you could expect to see dodge round after round for many significant fights. (Anyth8ng worth spending resources on).

Like many thing perspective drives perception - someone biased against dodge who sees it as mostly useless or the weaker choice is gonna make choices that in fact create just that reality.

It's funny how much folks will do yo get advantage but do much less to give disadvantage.

Just for interest sake ... here is the math for advantage and disadvantage as a function of the number you need to roll to hit.

If the number you need to roll to hit is X ... e.g. X + to hit modifier = AC then the formulae are (X has a minimum value of 2 since a 1 always misses and a maximum value of 20 since 20 always hits).

Advantage probability to hit is: 1.0 - the odds that both die rolls miss.

Advantage % to hit X = 100 * ( 1 - [(X-1)/20]^2)

e.g. AC=15, to hit modifier is +5, to hit = 10
Advantage % to hit = 79.8%

Disadvantage probability to hit is: probability that both die rolls hit

Disadvantage % to hit X = 100 * [(20-X+1)/20]^2

e.g. AC=15, to hit modifier is +5, to hit = 10
Disadvantage % to hit = 30.6%

The base probability to get a 10 or greater result without advantage or disadvantage is 55%

Thus in this case, disadvantage reduces the to hit by about 25% while advantage increases it by about 25%.

THIS is why passive skills are modified by +/-5 for advantage and disadvantage since for the average case advantage/disadvantage incur a +/-25% success probability change which is roughly equivalent to a static +/-5.

(so can folks stop talking about +/-3?)

P.S. For to hit dice rolls, a target number X below 2 is capped at 1 since a 1 is always a miss while a target number X above 20 is capped at 20 because criticals always hit. However, this is not true for saving throws or skill checks.

So I just thought I would add a table showing the % to hits for normal/advantage/disadvantage vs target number along with the difference from the base case.



Notes:

This shows that over the range of the most common target numbers from about 8 to 14 (due to bounded accuracy) the effect of advantage/disadvantage varies from +/-21% to +/-25% (+4 to +5).

At the very extreme of the target numbers ... like needing a natural 20 to hit ... the effect is closer to that of a +1. However, the extremes do not come up as often as the middle of the distribution ... the game is balanced around typical target numbers in a standard encounter around 11. AC16 with +5 to hit at level 3 or maybe a typical AC20 with +9 to hit at level 11 ... sometimes the AC's are much easier or much harder to hit but then the creatures likely have varied hit points or other compensating abilities (like resistances).

Due to this, ascribing a static +/-3 to advantage/disadvantage isn't an accurate assessment.

The effect of elven accuracy "trivantage" is also interesting since for target numbers between 12 and 17 it is the equivalent of +2 to almost +3 to hit compared to regular advantage.

Finally, by comparing the target numbers for a base hit against target numbers with a +5 applied for advantage and trivantage you can assess the impact of using feats like GWM and SS. (eg. if the target number is normally 10 ... it will be 15 when using SS/GWM)


P.S. Keep in mind that the 1 line is only for skill checks (and saving throws?) ... to hit rolls auto miss on a 1 so it uses the "2" line in the table.

TallIan said:

Thanks, I realised last night what my mistake was on the earlier post. The one I linked to shows the probability of rolling exactly that number, not greater than that number. I know you say it's not that but if you add all the % under the DC you want you get the same answer in both tables.

It's kinda handy to do quick on the fly maths mid-session, as long as you understand how inaccurate it is. A flat bonus is also closer to =/-3 rather than 5. I think the idea of a +/-5 bonus comes passive skill checks getting +5 for advantage and -5 for Advantage.

But going back to the OP's question requires us to look at specific situations, and squinting at highly inaccurate approximations of a bonus won't help with that

Click to expand...

It's not the odds of rolling that number. Those odds are the same for any given number at 9.75%. Hence why I said I can't follow what that table is doing.

Ovinomancer said:

It's not the odds of rolling that number. Those odds are the same for any given number at 9.75%. Hence why I said I can't follow what that table is doing.

Click to expand...

Roll 2d20 take highest causes you to have a 1/400 chance of rolling a 1.

Ovinomancer said:

It's not the odds of rolling that number. Those odds are the same for any given number at 9.75%. Hence why I said I can't follow what that table is doing.

Click to expand...

That can’t be right (or I’m missing sometimes again) as that would give you 20 * 9.75% = 195% chance of getting a result.

FrogReaver said:

Roll 2d20 take highest causes you to have a 1/400 chance of rolling a 1.

Click to expand...

That wasn't the question.