I agree, you can't wait forever. We can micro-step the process...and once we have interrupts, we MUST subdivide the process:

1. MAGE declares action: Fireball.
2. MAGE declares target(s) for action.
3. MAGE casts spell.
4. MAGE rolls to hit targets.
5. MAGE rolls damage, which is applied simultaneously to all targets (based on hit rolls)

Done. Note that steps 1, 2, and 3 are the CASTING part of the action; 4 and 5 are the RESOLUTION part. IMO, the RANGER declares "Disruptive Strike!!!" after step 2. That answers a). For b)...it could impact step 3 or step 4...but the rolling part is usually step 4 in 4E, so that's where it makes sense to apply the penalty. And, yes, I'd apply it to all attack rolls made *as part of step 4.* I probably would NOT have it apply to secondary attack rolls made as a result of the power, when those are of the type "if you hit, make a secondary attack." The point is, the interrupt was trying to make you NOT hit. It didn't work.

I would draw a contrast with powers like Shield, which have the trigger "you are hit by an attack."

That's why you explicitly write out all the micro-steps. "Hit by an attack" is the end of step 4. The trigger interrupts before you move to step 5.

On the subject, I have an issue with your 5-step process. It looks like Step 3 would be the moment when opportunity attacks would occur.

That is, in fact, why I made it a step in itself, for 3E AoOs, where the language specifically discussing the act of spellcasting, and for which there are specific ways to avoid this. Casting on the Defensive applies here, for example. It may not matter for 4E, but I don't think it hurts.

In a context where both an opportunity attack and Disruptive Strike occurred in reaction to the same spell, I'd put the OA first ("uses a power") before the Disruptive Strike ("makes an attack").

Define "uses a power" versus "makes an attack" in terms of these steps. Can't do it here, as I have no access to book or PDF. Read it carefully, and think about what they're trying to do with it. Whatever feels right to you, tho...go for it. There's no absolute, formal right or wrong, at least until they formalize this for us, and consistency *for your use* is what you're after. I doubt they will, personally; people will scream it's too much like Magic. Yeah, well, the more ways you can have, where Action A can be affected by Response B, C, or D...and when, now, you have an entire class of Reactions (no, I'm not using game terms, I'm using general concepts)...you can't treat A as an atomic action that starts and completes any more. Thus, you NEED stages. Magic went through this in *extraordinary* complexity, because there were MANY more ways this could happen. Use the experience gained in sorting out those problems...or don't set up this kind of unsolvable rules question in the first place.

No, not tricky...except that the rules fail to properly explain what "is attacked by" means. Especially in comparison to "uses a power" and other language that says "this guy is taking an action, you (or your buddy) are involved, you won't like what it does."

Just step through everything, and we'll use OAs to illustrate the comparison. Fireball is an area attack power, so the OA trigger is "uses a power." Now, to be sure, the definition of "uses a power" is unclear.

1. MAGE declares action: Fireball.
2. MAGE declares target(s) for action.
3. MAGE casts spell.
4. Roll attacks....

There is no reason to say the interrupts happen DURING these steps; it's cleanest to say they happen between them.

So: Can the OA happen between stages 1 and 2? Yes. Between 2 and 3? Yes. Between 3 and 4? Yes. Once step 1 is finished, the power is "in use." After step 4? Logically, no; the power is no longer "in use", it's "used, and being resolved."

What about the DS? Can it be used between steps 1 and 2? NO. Not until targets are declared, because it has to target you or an ally of yours, and we don't know that yet. Can it be used between 2 and 3? Yes, targets are known. Between 3 and 4? Yes. And, again, there's no reason to invoke it later than that.

There's nothing tricky about any of that...so long as you realize the big distinction that's really always been present, between the period where you're creating the effect, and when the effect resolves. 3E Countermagic (via casting the same spell as a counterspell) inherently recognized this; the only real 'tricky' part is they've never come out and made this explicit.

Again, it's not necessarily so much tricky, as a failure to recognize (on WotC's part) that a question exists.

The DS can't occur between 4 and 5. "Resolve the attack" means "do damage"...rolling the attack gives the answer "did it hit?" The DS has to occur before step 4 for it to be a factor in that. Effects that modify the damage are what happen after steps 4 and 5.

But...while, yes, this breakdown suggests it's a modifier to the caster's attack roll that should be applied, it's NOT explicit that this should be the case...because they don't have the process adequately defined.

An interrupt *generally* could happen after step 4, yes...but the trigger doesn't say that for DS.

4. roll to hit targets
5. determine hit/miss
6. determine damage

Ehhhh...no. I can't buy including step 5. The start of 4 is roll the die, the end is "did you hit." This feels atomic, and I don't see any reason separate this.

And WOULD Shield force the re-roll you suggest? It's NOT clear that it would. The attack hit you already; IMO, it's too late to change that...and as I read it, I wouldn't give it to you. There is nothing in the errata about it. My objection to it, tho, has nothing to do with including a separate "determine hit/miss". Your ruling says the spell has to roll back time, to a very slight degree. There's no other way for me to interpret the re-roll that's forced to happen after the fact. (An ability that gives you a re-roll is something completely different.)

And, for that matter: it could create a paradox. If the +4 to AC is enough that the attack would miss, then the trigger for Shield is no longer satisfied...so what happens? I detest paradoxes, and I will ALWAYS rule in such a way as to avoid them. And that's easy to do here: the attack still hits.

Now, note that the Shield AC bonus *would* kick in for something like Rain of Blows, if the wizard was the target of the secondary attack. That doesn't require the "time warp" effect or create the paradox.

But...ok, what you're saying is that if a hypothetical interrupt created a situation where it *was* necessary to check whether a given power was gonna hit or not...you don't automatically go all the way back to the start of step 4. Going back to step 4 is forcing a second die roll; going back to step 5 is not, it's just adjusting the parameters that determine hit or miss, but using the same die roll. OK, I'll buy that. 99 times out of 100, 4 and 5 are atomic, but the potential should exist for the interrupt...if there isn't one already, in the Warlock powers list. Unluck...triggers when a power is about to hit you, forces attack re-roll...take the worse result. (So if the 2nd roll is a 20, sorry, no crit.) Another one...one that does a better job of avoiding paradox...would be a two-part effect involving a tiny bit of prescience. Standard action to set up; the effect is to know the exact result of any attack against you, including how much damage it'd do, for the remainder of the encounter. As an interrupt, you can choose to try to intercept the power...say, gain +4 to the targeted defense, only against this attack, and/or force the attack to be re-rolled (thus giving you the chance to dodge/negate a crit.) Choosing to do so ends the spell.

This respects the steps.

So now...in the 6-step process...where does DS kick in? If we go absolutely literally, you're probably right: the literal "attacked by" is step 4. This doesn't much help decide the original question...because this part:

while (targets remaining) {
4: Roll attack against one remain target
5: Resolve that attack
} repeat

most commonly *can't* logically be interrupted. If I cast a burst, everyone I choose to affect, is affected at the same time. There can be attacks that have a sequential nature...say, "you shoot a ray of fire at 3 different targets...." That could all be at once, or could be read as "shoot at A, then B, then C in rapid succession." Typically, tho, I wouldn't read it that way.

So...if the attack is like a burst, where the 4-5 steps are simultaneous (the general case)...the DS kicks in if any of the targets is an ally of yours, and if successful, the penalty (because of the timing) applies to ALL the attacks.

Mind: give an example. It'll be useful to analyze it.

I should've said recompute, not reroll.

And it covers the paradox as well.

When you use Shield, you use it after step 5, and it occurs chronologically between steps 4 and 5. It doesn't force a reroll, as it occurs before the hit resolves, but after it is rolled. This is why I inserted step 5 in there. Unluck does the same thing, forcing a second roll between steps 4 and 5, and then moving on to resolving the attack once again.

But that's still going back in time to an earlier stage in the process. That's why I don't buy your argument. You can't use Shield after step 5 to change what happened IN step 5. And note that my Unluck is being used BEFORE step 4. The logical justification is that you, the target, have advance knowledge of the result before it happens, therefore time to invoke something to attempt to change the outcome. The target is at step 6 when the caster is still inside step 4.

I just realized: there's a solution for these sequencing issues. Make the powers Reliable. NOW Shield can be declared in step 4, and take effect between 4 and 5. If the power misses, the trigger doesn't happen, and the fact that it's Reliable means it's not used up. I think the powers that give the re-roll can also be used this way.

And yes, the fireball would hit them all at once in the game world, but I'm talking metagame mechanics here. You don't normally roll all three attack rolls simultaneously. You roll and resolve them in sequence.

I would rather respect the sequence, and the internal, systemic logic. This breaks the systemic logic. Having the penalty apply to all the attack rolls, doesn't. The net effect is to make DS actually a bit more powerful, but I think it's OK.

And Shado, on your last point...look at Crucial Advice (Ranger Utility 2). It gives an ally a reroll, but the trigger says you're doing it during the check. You don't get to know, in advance, whether it'll succeed or not. But, note too that it's an encounter power. My gut feeling is that encounter powers are probably more likely to be of this type...especially with a skill check. Daily powers will more likely let you wait, because they're gone for so much longer. And, of course, you just don't have that many dailies, so one that you have to use...but it may not do you any good? Ugh. Of course, if it's Reliable, that's not a problem.

It's not an argument. It's a quote from the core rules.

The quote is, yes. The assertion that Shield may negate an attack that has hit you, is not. That's an interpretation.

As for your hypothetical Unluck (actually the name of a Bard attack power, oddly), does Unluck trigger when you would be hit? I assumed, from what you said ("triggers when a power is about to hit you") that it did. If so, it triggers off of Step 5, and thus would occur before the resolution of the attack roll, between steps 4 and 5.

Remember what I said. To keep sequencing intact, I'm providing that the spell includes a weeeee bit of precognition. If it didn't do that, then yes, it would have a similar paradox.

This is how interrupts work. They are like readied actions, and can cancel the consequences of the trigger, but this does not mean they never got triggered.

I understand interrupts *extremely* well. Your example is incorrect, because you misstated the trigger. The trigger, properly phrased, would be "the caster starts to cast a spell on me." If it's "casts a spell on me" then you WAIT!!! until after he finishes. There's no paradox; it's simply imprecision of language. Of course, that could be true with Shield. The correct wording could be "would hit you." Not "are hit by" but "would hit you." The RAW language triggers right AFTER it's hit; mine is right BEFORE it hits, and therefore WOULD let the defense bonuses be applied against this attack. "Would hit you" is that tiny little boundary that is the transition from step 4 to step 5; "are hit by" is after step 5.

My assertion is this: no interrupt can, at any time, affect anything that has already happened. It can only affect what happens from the point of trigger forward. That is the ONLY! logical interpretation. You can never go back to before the trigger for the interrupt effect......BUT...it's not necessarily true that you can't declare the *power* before then, *if* it's built with that use in mind, similar in many ways to Contingency. Or, as with Unluck, invoke that the power itself involves precognition.

While I'm not sure I'd call the power overpowered, I believe it is one of the best Ranger powers in the book. A -10 to an attack roll is hard to come by, and is just as effective at level 30 as at level 3. A -10 to 6 attack rolls is very powerful indeed.

Your ranger has a 24 Wis??? The penalty is only 3+WisMod.

As such, if we use the same sequence as above (though some steps will be different), Crucial Advice triggers with Step 4 (make the skill check), and occurs before Step 5 (the resolution of that skill check).

NO. Why MUST it trigger as late as step 4? What is the concept of "making a skill check"? Step 1 is declaration/intent: "I'm going to try to pick the lock on the chest." He is now involved in the standard action of making a skill check. The roll itself is NOT the entire action; it's the end of the action.