
![]() |

Perfect Strike (Combat)
When wielding a monk weapon, your attacks can be extremely precise.
Prerequisites: Dex 13, Wis 13, Improved Unarmed Strike, base attack bonus +8.
Benefit: You must declare that you are using this feat before you make your attack roll (thus a failed attack roll ruins the attempt). You must use one of the following weapons to make the attack: kama, nunchaku, quarterstaff, sai, and siangham. You can roll your attack roll twice and take the higher result. If one of these rolls is a critical threat, the other roll is used as your confirmation roll (your choice if they are both critical threats). You may attempt a perfect attack once per day for every four levels..(etc)
Grammar boil-down:
Benefit: If you declare you are making a Perfect Strike before attacking with a kama, nunchaku, quarterstaff, sai, or siangham,
roll your attack die twice and take the higher result. You may attempt a perfect attack once per day for every four levels..(etc)
- If you crit, it has to be confirmed, so you're always rolling another die anyway to check. IOW, all the wording about critical hits in the text of the feat is completely meaningless.

james maissen |
- If you crit, it has to be confirmed, so you're always rolling another die anyway to check. IOW, all the wording about critical hits in the text of the feat is completely meaningless.
Incorrect.
Without the wording you would think that you would roll 2 dice to see if you hit, then on a threat you would roll a 3rd die to confirm even if the unused of the first two die rolls were a hit.
This is not the case. If you get a threat using perfect strike, the confirmation roll is your other die not a separate and subsequent roll.
-James

Mahorfeus |

James hit the nail on the head. The ability gives you the advantage of having a greater chance of striking the opponent.
The trade-off is that you have to pay for that second roll somehow; it would be unfair if you got say, a 20 and a 3, yet you are able to ignore the 3 and reroll for confirmation.
Otherwise, asides from limited use, there is no disadvantage to using Perfect Strike all the time.

Xraal |

Mike Schneider wrote:
- If you crit, it has to be confirmed, so you're always rolling another die anyway to check. IOW, all the wording about critical hits in the text of the feat is completely meaningless.Incorrect.
Without the wording you would think that you would roll 2 dice to see if you hit, then on a threat you would roll a 3rd die to confirm even if the unused of the first two die rolls were a hit.
This is not the case. If you get a threat using perfect strike, the confirmation roll is your other die not a separate and subsequent roll.
-James
+1
Absolutely correct James.

![]() |

This is not the case. If you get a threat using perfect strike, the confirmation roll is your other die not a separate and subsequent roll.
-James
Correct but this part also makes perfect strike not as good in most of my players eyes because it gives twinlinked but then makes you take a potential bad roll for your confirm. We house rule the crit gets rolled seperately since I couldnt figure out a reason to penalize a monk for this ability

eXaminator |

Correct but this part also makes perfect strike not as good in most of my players eyes because it gives twinlinked but then makes you take a potential bad roll for your confirm. We house rule the crit gets rolled seperately since I couldnt figure out a reason to penalize a monk for this ability
I don't see a penalty either. Its just less of an advantage. Normally you would get two rolls if you crit. You still get those two, but you just roll them together. You said he'd to go with potential bad roll... well, yes. Just like he would have to if he didn't have this ability (but in this case he might at least hit while without this ability he might have rolled the 3 first and would probably not hit at all.

Kaiyanwang |

james maissen wrote:Correct but this part also makes perfect strike not as good in most of my players eyes because it gives twinlinked but then makes you take a potential bad roll for your confirm. We house rule the crit gets rolled seperately since I couldnt figure out a reason to penalize a monk for this abilityThis is not the case. If you get a threat using perfect strike, the confirmation roll is your other die not a separate and subsequent roll.
-James
It's a good houserule, but my guess is that Perfect Strike is suited for maneuvers attacks delivered after a move action.
A Trip (Kama) or Disarm (Nunchaku) rolled twice at full BAB is very, very likely to land.
another idea is the quarterstaff used two handed. Less DPR than using a temple sword normally, but rerolling a middle-to-low BAB iterative attack could increase the reliability of the full attack itself.
In other words, is for reliable attacks, not for criticals. I like your houserule indeed.

Rubia |
laurence lagnese wrote:I don't see a penalty either. Its just less of an advantage. Normally you would get two rolls if you crit. You still get those two, but you just roll them together. You said he'd to go with potential bad roll... well, yes. Just like he would have to if he didn't have this ability (but in this case he might at least hit while without this ability he might have rolled the 3 first and would probably not hit at all.
Correct but this part also makes perfect strike not as good in most of my players eyes because it gives twinlinked but then makes you take a potential bad roll for your confirm. We house rule the crit gets rolled seperately since I couldnt figure out a reason to penalize a monk for this ability
Here are some numbers to show that it's strictly better, even for critting. Let's suppose here (for simplicity) that you would hit on anything 15 or higher for this example, and that only a 20 is a critical threat.
Possible rolls with a perfect strike (two dice) are:
(20,20), (20,19), (20,18), (20,17), (20,16), (20,15), (15,20), (16,20), (17,20), (18,20), (19,20)
Clearly, any of these sets of rolls would result in a confirmed crit, whereas without perfect strike, only the first 6 would. It nearly doubles the chance of earning a critical. If you consider pairs of numbers like (20, 3) and (3, 20), perfect strike results in a 100% chance of hitting, as opposed to a 50% chance of hitting without perfect strike.
Perfect strike is a great feat, and though it may seem terrible that you have to take the second roll, taking the second roll is strictly better than not having it. Moreover, it is quite powerful as written. If you ever get three dice later, it is even more powerful. If I were you, I wouldn't houserule a third die.
Rubia

![]() |

> If you crit, it has to be confirmed, so you're always rolling another
> die anyway to check. IOW, all the wording about critical hits in the
> text of the feat is completely meaningless.Incorrect. Without the wording you would think that you would roll 2
dice to see if you hit, then on a threat you would roll a 3rd die to
confirm even if the unused of the first two die rolls were a hit.This is not the case. If you get a threat using perfect strike, the confirmation roll is your other die not a separate and subsequent roll.
It does not matter how, in terms of probability, the confirmation die to a critical hit is rolled.
The extra text in the wording of the feat pertaining to confirming critical hits is superfluous and obfuscatory padding, because there is no probabilistic difference between the player confirming a crit with his leftover "worst" die, or rolling a third one (in the hypothetical case of him having to disregard the leftover one on the table) -- The odds he confirms are the same.

Mojorat |

pretty sure with perfect strike you are statistically more likely to crit.
let's say we know the next two rolls are going to be a 15 and an 18.
the normal attack never gets to confirm at all. the perfect strike crits.
I'm sure someone with better math skills than I can tell me that the percent is. but really the bug difference is it allows you to crit when you wouldn't have and with combat maneuvers doubles your chance of hitting.
the lvl 10 weapon master just becomes a co,bat maneuver monster in my view. let alone what 3 tries does for crits.

Are |

The extra text in the wording of the feat pertaining to confirming critical hits is superfluous and obfuscatory padding
No, it isn't. The extra text means less time is spent rolling dice at the table, since instead of rolling those two dice and then potentially rolling a third, you only roll those two dice.
The text doesn't affect the probability of the critical threat being confirmed, but that doesn't make it superfluous.

yeti1069 |

Quote:> If you crit, it has to be confirmed, so you're always rolling another
> die anyway to check. IOW, all the wording about critical hits in the
> text of the feat is completely meaningless.Incorrect. Without the wording you would think that you would roll 2
dice to see if you hit, then on a threat you would roll a 3rd die to
confirm even if the unused of the first two die rolls were a hit.This is not the case. If you get a threat using perfect strike, the confirmation roll is your other die not a separate and subsequent roll.
It does not matter how, in terms of probability, the confirmation die to a critical hit is rolled.
The extra text in the wording of the feat pertaining to confirming critical hits is superfluous and obfuscatory padding, because there is no probabilistic difference between the player confirming a crit with his leftover "worst" die, or rolling a third one (in the hypothetical case of him having to disregard the leftover one on the table) -- The odds he confirms are the same.
The feat is designed to increase your chance of hitting, not scoring a critical.
His chances DO go up for crits with your 3rd roll.
In the 3 and 20 example.
Normally, you roll to-hit.
If you roll a 3, you miss. If you roll a 20 after this, you're just wasting rolls.
If your roll a 20, you hit and crit threat, then if you then roll a 3, you get a regular hit.
With Perfect Strike, you roll a 3 and a 20. Your chance to-hit jumps up dramatically, but your chance to crit doesn't improve.
If you add in a 3rd roll all it does is improve your chance to land a critical hit, which isn't the point of the feat.

Mojorat |

except I'n the cases where both rolls would have hit. it probably doubles your chance to crit and as I said I'n my previous post allows you to crit when you never would have. the feat does what it does you cannot just ignore part of the text and pretend it isn't a house rule.
but your right a third roll would increase your chance to crit and lvl 10 weapon master would be 4 times as likely to by your rules. but as per the feat you use the dice rolled to confirm not an extra one.

![]() |

The feat is designed to increase your chance of hitting, not scoring a critical.Your chances of scoring either a hit or a crit are doubled -- because you're rolling twice in the attempt.
pretty sure with perfect strike you are statistically more likely to crit.
Of course, because you're rolling two dice -- but that's not the issue.
What you're not more likely to do is confirm a crit if you get one. Thus the feat text regarding crits is meaningless.

Mahorfeus |

Quote:pretty sure with perfect strike you are statistically more likely to crit.Of course, because you're rolling two dice -- but that's not the issue.
What you're not more likely to do is confirm a crit if you get one. Thus the feat text regarding crits is meaningless.
The point is to give you a higher chance of hitting, and by extension, gives you a higher of threatening, not critting.
Bold: You've already pointing out the purpose of the crit text, right here.

Ksorkrax |

Guys, read Rubias post again.
Suppose we have a guy who hits another guy at a 15 and the crit range is the 20.
(the {x,y} indicates the results of the red die x and the blue die y, with perfect strike they are thrown simultaneous, without red is thrown first)
P(critical without perfect strike) =
P({20,15},{20,16},...,{20,20}) =
6/400 =
1,5%
P(critical with perfect strike) =
P({20,15},{20,16},...,{20,20} , {15,20},{16,20},...,{19,20}) =
11/400 =
2,75%
@Mahorfeus: Here the math for threatening with the guy from above:
P(threaten without ps) = 1/20 = 5%
P(threaten with ps) = P({x,20},{20,x} for x arbitrary in 1..20) = 39/400 = 9,75%

![]() |

This line of text:
If one of these rolls is a critical threat, the other roll is used as your confirmation roll (your choice if they are both critical threats).
...is completely meaningless. The odds that a threat confirms are the same: an additional d20 was rolled. In Perfect Strike, you just rolled it before you knew you had a crit.

The Chort |

I'm not that amazing with probability, but here's a fairly simple way to break it down:
Let's say the only way your attack can connect with the opponent is to roll a nat 20. In order to crit on him, you'd have to get a nat 20 twice, meaning you have a 1/400 chance of doing so. (1/20 squared.)
With perfect strike, since you have to roll twice and take the other as the confirmation roll, this stays at a 1/400 chance.
If, however, perfect strike allowed you to roll twice and then after that roll a third die to confirm, your chances would improve slightly.
The probability of getting a nat 20 with two rolls is better than with one roll. Lets say your chance is 1/10 rather than 1/20. (I know this is wrong, because rolling 20 d20s does not mean you have a 100% chance of getting a nat 20, but just "roll with it.") So you'd have a 1/10 chance followed by a 1/20 chance, meaning you'd have a 1/200 chance of confirming if perfect strike worked as you claim rather than the actual 1/400.
Huzzah. Hope this helps?

![]() |

This line of text:Quote:If one of these rolls is a critical threat, the other roll is used as your confirmation roll (your choice if they are both critical threats)....is completely meaningless. The odds that a threat confirms are the same: an additional d20 was rolled. In Perfect Strike, you just rolled it before you knew you had a crit.
Except, without that line, when someone rolled a 3 on the first die, and a 20 on the second, he has to roll a third die. Or, if he rolled a 15, and then a 20, he might start an argument that he should be able to use the 15 for his confirmation roll. Having it in the rules precludes that possibility. It's also important if you have an ability that activates on a 20, and roll a 19 followed by a 20.

Are |

This line of text:Quote:If one of these rolls is a critical threat, the other roll is used as your confirmation roll (your choice if they are both critical threats)....is completely meaningless. The odds that a threat confirms are the same: an additional d20 was rolled. In Perfect Strike, you just rolled it before you knew you had a crit.
It's only meaningless if you assume that the text has anything whatsoever to do about crit probability. Which it doesn't. The text is there solely to prevent a third die from being rolled unnecessarily. And the text accomplishes that purpose perfectly.

BryonD |

Except, without that line, when someone rolled a 3 on the first die, and a 20 on the second, he has to roll a third die.
Yep.
They also could have also solved by simply declaring that a third die must be rolled. But I like the way they did it. Not only does it run a little more efficiently, it also puts more interest in both dice. Regardless of the fact that one of them MAY end up being pointless.
And this may just be me, but in an abstract way it captures a the feel a little better with it all happening at once. That may be crazy and I don't know if I can really explain it. But it is fun this way.

![]() |

The text is there solely to prevent a third die from being rolled unnecessarily. And the text accomplishes that purpose perfectly.
It doesn't matter if the threat is confirmed by a third die or the "leftover" die already on the table, because the odds would be the same. The text is meaningless.

The Chort |

Quote:The text is there solely to prevent a third die from being rolled unnecessarily. And the text accomplishes that purpose perfectly.It doesn't matter if the threat is confirmed by a third die or the "leftover" die already on the table, because the odds would be the same. The text is meaningless.
Are you trolling? xP I broke it down fairly simply in my post. Rolling a 3rd die DOES matter.

james maissen |
The odds that a threat confirms are the same: an additional d20 was rolled. In Perfect Strike, you just rolled it before you knew you had a crit.
Semantics here. It depends upon the givens. Otherwise we're talking lies, damn lies and statistics here.
Imagine if you always had to roll 2 dice at the same time. The 2nd die (distinguished ahead of time) was for threat confirmations should the 1st die merit such.
Perfect strike is merely saying that you can relabel these 2 dice after the roll.
Now consider the case that when a PC would hit on a 12 but only threaten on a 20. You roll two dice and get a 13 and a 20. There are two ways for this to occur.
Either you rolled the 13 first and hit, meaning the 2nd die is useless.
Or you rolled the 20 first and not only hit but made a threat, meaning the 2nd die confirms it.
If this was a perfect strike you could choose to make it the second case rather than the first.
Seems like it can give you more critical hits to me.
-James

Are |

Quote:The text is there solely to prevent a third die from being rolled unnecessarily. And the text accomplishes that purpose perfectly.It doesn't matter if the threat is confirmed by a third die or the "leftover" die already on the table, because the odds would be the same. The text is meaningless.
Did you even read my post before you quoted a part of it?
The odds are completely irrelevant, because the text simply doesn't have anything to do with odds.
The text is only there to help keep die-rolling to a minimum. So, instead of rolling two dice, and then a third, you never roll more than two dice.
That is the only purpose of that text, and for that purpose, the text is not meaningless. In fact, to accomplish that purpose, the text must be there.
If you realize that the text was never meant to affect probabilities, it will make more sense to you.

![]() |

The text is only there to help keep die-rolling to a minimum
Heaven forbid the typical gamer overly-strain his cholesterol-laden left-ventricle by power-curling one more d20 than is absolutely necessary to meet the minimal prerequisites for entry into the P90X Plus Extreme Fitness prestige program!
Brevity is a virtue: no rule should contain extraneous text. The presence or absence of the line of text in question alters nothing in terms of game-mechanic effects, hence it is extraneous.

Kirth Gersen |

Brevity is a virtue: no rule should contain extraneous text. The presence or absence of the line of text in question alters nothing in terms of game-mechanic effects, hence it is extraneous.
I almost always agree -- the only things that trump rules brevity are (a) clarity and (b) streamlining play. If this language does both of those, as TOZ points out, then I'm perfectly willing to add some verbiage.

![]() |

> Except, without that line, when someone rolled a 3 on the first die,
> and a 20 on the second, he has to roll a third die.
It doesn't make a difference, as the odds are the same.
There are a number of abilities which permit rerolling missed attacks; none of them suggest using a first "missed" die to confirm crits. Perfect Strike is just a slightly better mechanic versus a reroll ability in that it gives you two chances to get a crit, whereas in a reroll ability you don't bother to reroll if you get a hit on the first throw (so you miss out on a few crits because you're not going to reroll a good-enough 17 hoping to get a 20).

![]() |

Quote:> Except, without that line, when someone rolled a 3 on the first die,
> and a 20 on the second, he has to roll a third die.It doesn't make a difference, as the odds are the same.
Rolling two dice isn't something different from rolling three dice?
Odds mean nothing to the question of 'does this line mean anything?'
It changes what happens, therefore it means something. You may feel the change is unimportant, but it is still there.

![]() |

Rolling two dice isn't something different from rolling three dice?
Situation:
Player: <rolls Perfect Strike> "Yippee! Take that!" <number crunch, number crunch>....
DM: "Didn't you have a Scabbard of Keen Edges for your kama? You rolled a 19. Confirm your crit."
Player: "Oh, Uhm....I already moved the dice."
DM: "Well, roll another one."
-- The odds the player confirmed are the same regardless of the number of dice, because only one of the additional ones will be used.
The elimination of unneeded die rolls has been a primary objective in Pathfinder since its development.
I sincerely hope that's a facetious statement. (I can think of about a hundred more important objectives, and heading the list are: promoting a user-sustained regional convention system to equal the old Living Greyhawk, better bindings for the books, consistent artwork, retooling the website so a single click takes me to my Society PCs' accomplishments page, reconsidering the "3 mods = level" Society scheme at mid-level so characters don't retire so fast, an official stamp-of-approval offline character generator, etc.)

![]() |

Here's why you're both wrong:The Chort wrote:Rolling a 3rd die DOES matter.Just quoting The Chort here since the post seems to have been rather unfairly overlooked by the OP. As his example suggested the text is not meaningless and does impact the odds.
If, however, perfect strike allowed you to roll twice and then after that roll a third die to confirm, your chances would improve slightly.
-- There isn't a DM alive who is going to let a player get away with having two die to confirm his crit, text or no text.
In the absence of any text, EITHER the DM says "Just use the other one which you already rolled," OR he says "Roll another one." In zero percent of cases is he going to say, "Use the leftover die to confirm, unless it stinks; in which case I'll let you roll another."

james maissen |
It doesn't make a difference, as the odds are the same.
The odds of what exactly?
Be specific here as in many ways the odds are not the same.
The odds of getting a critical using perfect strike are higher than getting a critical when not. Simply consider a pair of die rolls where one is a threat and the other is a hit. In one order for a normal attack it's a simple hit, while in another it is a critical.
-James

Heaven's Agent |

I sincerely hope that's a facetious statement.
Hate to disappoint you, but it's not. It was one of the primary design goals, and one applauded by most of the community when it was announced.
Scratch that, I don't hate disappointing you in this matter. You are complaining about something without understanding the background of the situation, and object when that background is presented to you. I'd have to agree with The Chort; it feels as if all your doing is trolling. I'm going to stop feeding you now.

![]() |

Mike Schneider wrote:It doesn't make a difference, as the odds are the same.The odds of what exactly?
James, I've already answered that many times. (And it's right there in the OP.)
-- The odds of confirming a crit.
The odds of getting a critical using perfect strike are higher than getting a critical when not.No one was arguing over that.
Simply consider a pair of die rolls where one is a threat and the other is a hit. In one order for a normal attack it's a simple hit, while in another it is a critical.
No one was arguing over that.
----
Hate to disappoint you, but it's not. It was one of the primary design goals, and one applauded by most of the community when it was announced. Scratch that, I don't hate disappointing you in this matter.Disappoint the customer -- what an excellent policy! (I have ranks in Sarcasm too. Don't we all....)
You are complaining about something without understanding the background of the situation, and object when that background is presented to you.I find the assertion that most of the gaming community considers elimination of die-rolls a primary priority to be, frankly, credulous.
I'd have to agree with The Chort; it feels as if all your doing is trolling. I'm going to stop feeding you now.
Except that Chort's argument, and now yours by extension for having sided with him, was wrong. The subconscious desire to evade that recognition (by not responding to my rejoinder to Chort) with a foray into this other "primary design" claim is a moving goalpost fallacy.
Truce: If you'll stipulate I was correct in that the line of text in question alters nothing in terms of game-machines, then I'll stipulate you are correct when you say reducing the amount of dice rolled during game play was the primary reason the text was introduced.
(I'll still dislike it, as I'll take rule brevity over rule bloat every time.

lobachevskii |
1 person marked this as a favorite. |
-- There isn't a DM alive who is going to let a player get away with having two die to confirm his crit, text or no text.
In the absence of any text, EITHER the DM says "Just use the other one which you already rolled," OR he says "Roll another one." In zero percent of cases is he going to say, "Use the leftover die to confirm, unless it stinks; in which case I'll let you roll another."
According to my back of the envelope calculations these two scenarios give (slightly) different probabilities of a confirmed crit.
1) Perfect Strike as written:
number of ways of getting a confirmed crit: 2*(20-X)+1 (where X is the target to hit, X not equal to 1).
total number of ways to roll 2 D20: 400
P(confirmed crit) = (2*(20-X)+1)/400
Brief explanation: You get a confirmed crit if you roll a 20 and the other die hits. You can roll two 20s (that's the 1) or a 20 and any number less than 20 down to X in one of two ways (that's the 2*(20-X) term).
2) Perfect strike but confirm the critical separately:
number of ways of getting a 20 on 2D20: 2*19+1 = 39.
number of ways of hitting on the confirming die = 21-X (X is target number, X not equal to 1)
P(confirmed crit) = 39*(21-X)/(400*20)
Brief explanation: It doesn't matter what you roll on the first pair of dice as long as one's a 20. You can roll two 20s (that's the 1) or a 20 and any number less than 20, there are nineteen of those, in one of two ways (that's the 2*19 term). Probability of confirming a crit is (21-X)/20. We can multiply the two probabilities since they are independent.
Comparing these two distributions we find that 39*(21-X) > 20*(2*(20-X)+1) for all X > 1. The difference becomes increasingly large as X increases.
TLDR: The statement is necessary. It specifies a game mechanic to use which produces a slightly different probability distribution to that which would arise under the standard rules.

james maissen |
james maissen wrote:Mike Schneider wrote:It doesn't make a difference, as the odds are the same.The odds of what exactly?James, I've already answered that many times. (And it's right there in the OP.)
-- The odds of confirming a crit.
Specifically the odds of confiming a crit given that you've threatened a critical correct?
And the odds of this happening is different between an attack with perfect strike and one without as the chance to threaten is not independent of the chance to confirm there.
You're miscounting the chance that both dice threaten criticals in your thinking is my guess.
-James

![]() |

According to my back of the envelope calculations these two scenarios give (slightly) different probabilities of a confirmed crit.
The math provided does not account for the possibility of 19s threatening with monk weapons (he might have Improved Critical or be the recipient of a spell granting extended threat-range). In any event, see below:
You're miscounting the chance that both dice threaten criticals in your thinking is my guess.
It doesn't matter if both dice threaten because you only get to score one attack. The "goodness" of Perfect Strike is in rolling two dice simultaneously (rather than one after another, as you would when using a "reroll" feat or ability, because with those you don't reroll if the first roll is a hit...unless your opponent is so bad and your weapon so good that you hit on 3s and you threaten on 14s or some such). This "goodness" is front-loaded by the mechanic of rolling two dice simultaneously; it has nothing to do with the gobbledegook of crit-confirmation because, once you've chosen either dice to be the attack die, the odds that the remaining one confirmed a threat are the same as any randomly thrown nth dice, and it does not matter if Schroadinger's box is open (the die is already rolled and on the table) or closed (a 3rd die yet unrolled in the instance the "leftover" die may not be used).
As said previously, no DM is going to let you get away with two chances to confirm a crit -- and that's the only way the odds would be different.

Rubia |
I think what I'm about to write is a clean way to understand the issues involved so that it will be clearly understood that it does matter whether you ignore the lower die(s) of perfect strike or not. Since probability drives people into insane land, I shall instead count how many ways there are to generate a critical hit, which should be easily verifiable for all people.
Clarifying Terms
Let me clarify terms so that it's crystal clear what's being discussed.
Plan A--Use perfect strike as written (i.e., roll two dice, and of those two dice, choose which will be used to hit and threaten, and which shall confirm, if necessary).
Plan B--Use a modified version of perfect strike (i.e., roll two dice, choose which will be used to hit and threaten. Then, ignore the other die entirely and roll a separate third die to confirm).
This post shall establish that Plan A and Plan B do not generate the same probability distribution. This post shall also explain the fallacy in the idea that "all dice are the same" argument from other posters (in short, the visible dice affect the likelihood of more than one event).
Background (Plan A)
It should be clear that Perfect Strike (as written) improves both the chance of hitting as well as critting. See my early post for details about how that was determined. Notice that there I alluded to the fact that a third die is bad (m'kay?), but did not explain why.
We shall use the same example as my earlier post. In that post, there were 11 ways to confirm a critical using Plan A.
Plan B
Let's now count how many way there are to confirm a critical using plan B.
To threaten:
(20,1), (20,2), ..., (20,20) plus (1,20), (2,20), ..., (19,20). There are thus 39 ways to threaten. We then, as per Plan B, ignore the second die, regardless of what it was.
For each threatening attempt, there are six ways to confirm that critical hit---roll a 15, 16, 17, 18, 19, or 20 on our third die. Thus, there are 39*6=234 ways to confirm a critical hit in this method under the assumptions previously made. Notice that none of these ways are repeated events, so I am not overcounting. What is happening is that we're leveraging the power of the non-chosen die to threaten.
Since 11 is not equal to 234, the plans clearly provide different distributions, and Plan B would be far more powerful than Plan A if it were allowed in the game.
Why Not The Cat in the Box?
For those of you watching at home, the reason the "Schrödinger's box" argument doesn't work is because choosing the maximum of two visible dice dramatically affects how many times you'll threaten (and therefore hit). In probabilistic terms, this means that initial two dice are not independent of one another, and you cannot simply consider the effect of the second die's confirming ability without realizing that it had an impact on the threatening part.
Hopefully that clears things up and we understand how things work.
Conclusions
Don't allow a third die, even if you always ignore the effect of the die(s) not originally chosen in the use of Perfect Strike.
No rules bloat, because the phrasing affects mechanics. Use it as written.
Corollary/Fun Fact
Both Plan A and Plan B generate the distribution for the number of hits produced, if we ignore the critical system. They only differ in the effect on criticals.
Rubia
P.S. I also like that Perfect Strike reduces additional dice rolled, as a side bonus.
P.P.S. Modified because of Are's observation that there are 39 ways to threaten.