What is technically considered an attack?

[Easy_Lee]

I think it would help if we knew why each feature or spell interacts with attacks. Ex: invisibility - why does this break upon attacking or casting a spell? If we knew, we could say what else might break invisibility.

An example of a trick that probably ought not work. A warlock casts Eyebite, then has a friend cast invisibility on him. The warlock can now perform three different Save or Bad Thing effects against foes, one each round. Unless the foes can find him while he's invisible, there's little they can do.

This technically works, to my knowledge. The warlock isn't casting a spell or attacking, just using an already-cast spell to do something hostile. The same could be done with many other spells.

And there are also the breath weapon and limited wand abilities above.

Invisibility doesn't say it breaks if you do something hostile; that's too vague, and would negate setting up traps which is probably an intended valid use. But as written, it enables a variety of hostile actions that it probably ought not.

And that's just one example. Coincidentally, sanctuary wouldn't block eyebite either, even though it arguably should.

[BurgerBeast]

I understand your point, I just don't believe that it will lead to anything particularly useful.

If the point I am making is correct, then it is not meant to bring something new to the table. It is meant to help people see that what they are currently doing/thinking is incorrect, and empower them to stop doing it.

That's the use: to prevent a currently occurring mistake and end the misconception in at least some cases. That's pretty useful.

Medicine is not useful when applied to healthy people. That's a not an argument that medicine is useless.

Arguing that we cannot conclude that taking the Dodge action (to take a more extreme example) is not an attack because it does not explicitly say so is hardly going to lead to a fruitful discussion.

Nobody is going around claiming that dodge counts as an attack. People are going around claiming that every single action that is not explicitly an attack or does not involve an attack roll is definitely not an attack. It is incorrect. It ought to be corrected.

Sure, there are some cases such as those you mentioned where it could be argued either way depending on your stance of English vs. game mechanics, but that's not really a RAW discussion anymore.

This is not about English vs. game mechanics. This is about what the book says. It is about one very specific example in the PHB and how to read it:

If there’s ever any question whether something you’re doing counts as an attack, the rule is simple: if you’re making an attack roll, you’re making an attack.

It comes down to reading comprehension and logic. That's all. People are bringing in other contexts that do not apply and distract from the point (it is not about game language versus everyday language; it is not about general versus specific; it's not about external biases around the truth of the matter; it's not about what the writers meant to write; it's not about JC's opinion).

[BurgerBeast]

I think it would help if we knew why each feature or spell interacts with attacks. Ex: invisibility - why does this break upon attacking or casting a spell? If we knew, we could say what else might break invisibility.

An example of a trick that probably ought not work. A warlock casts Eyebite, then has a friend cast invisibility on him. The warlock can now perform three different Save or Bad Thing effects against foes, one each round. Unless the foes can find him while he's invisible, there's little they can do.

This technically works, to my knowledge. The warlock isn't casting a spell or attacking, just using an already-cast spell to do something hostile. The same could be done with many other spells.

And there are also the breath weapon and limited wand abilities above.

Invisibility doesn't say it breaks if you do something hostile; that's too vague, and would negate setting up traps which is probably an intended valid use. But as written, it enables a variety of hostile actions that it probably ought not.

And that's just one example. Coincidentally, sanctuary wouldn't block eyebite either, even though it arguably should.

None of this is problematic unless you have the incorrect assumption that actions without attack rolls are definitely not attacks. This assumption has no basis. So the problem is solved. It comes down to the DM's judgment about whether the action is an attack or not, in some specific cases where the RAW does not make the determination for the DM.

[coolAlias]

If the point I am making is correct, then it is not meant to bring something new to the table. It is meant to help people see that what they are currently doing/thinking is incorrect, and empower them to stop doing it.

That's the use: to prevent a currently occurring mistake and end the misconception in at least some cases. That's pretty useful.

I guess I'm just not seeing enough people claiming that to be an issue for me, so it seems moot.

However, I agree with you - the test is not proof for a negative, only a positive, otherwise grapple and shove, despite being explicitly called out as attacks, would not be since they fail the general test as it is written.

[Unoriginal]

None of this is problematic unless you have the incorrect assumption that actions without attack rolls are definitely not attacks. This assumption has no basis. So the problem is solved. It comes down to the DM's judgment about whether the action is an attack or not, in some specific cases where the RAW does not make the determination for the DM.

So, just to be clear, you don't think that "If there's ever any question whether something you're doing counts as an attack, the rule is simple: if you're making an attack roll, you're making an attack" means that whenever the question "is this an attack?" is asked, you have to look if there is an attack roll, and that the answer is "it is an attack" if the attack roll is here?

[BurgerBeast]

I guess I'm just not seeing enough people claiming that to be an issue for me, so it seems moot.

However, I agree with you - the test is not proof for a negative, only a positive, otherwise grapple and shove, despite being explicitly called out as attacks, would not be since they fail the general test as it is written.

I'm not sure you've been reading the same threads as I have, then. It is exceedingly rare to find anyone who does not apply the reasoning presented by Unoriginal. Jeremy Crawford himself appears to apply the same misconception (this may not be true because he may not be speaking of RAW, but rather of RAI - however it appears to be true in some of his tweets).

Spoiler: The Underlying Fallacy is Context-Free

Show

The logical fallacy itself, in a different context, was the reason behind a marathon thread about how mirror image works, as well, in my opinion.

The misapplication of the expression: "RAW mean what they say, no more and no less" is both rampant and egregious. People tend to bring in a multitude of assumptions that bias their readings.

[MaxWilson]

Except, again, that the text EXPLICITLY refer to the special melee attacks as being attacks AS PER GAME RULES.

The same way that the authors took care of NEVER saying that the giants were humanoids, preasumably because they knew that it would lead to confusions like the hypothetical exemple you put above and so avoid the situation of people saying "but the text call them humanoid."

You are asserting that there are many things that are attacks that are not attacks in game term, like a dragon's Breath. And you could have a point. But the thing is, the PHB NEVER refers to dragon's breath or any of those hostile and potentially damaging actions that are not attacks as attacks. Never. Not even colloquially.

The PHB doesn't have any dragons in it, but the MM does. Try page 9:

Monsters with a challenge rating of 0 are insignificant except in large numbers; those with no effective attacks are worth no experience points, while those that have attacks are worth 10 XP each.

Are you claiming that a 1 HP micro-dragon with a 1d6 (DC 10 Dex save for half) 5' cone fire breath attack ought to provide zero experience points because it has no "attacks"?

There aren't all that many monsters in the MM that rely primarily on save-or-die effects instead of actual attacks, but one of them that does, a beholder, is said to "attack" at range, according to the MM, even though it has only eyestalks at range.

Adult dragons all have a "Wing Attack" which forces a Dex saving throw in an AoE. Creatures which fail that save take bludgeoning damage and are knocked prone. Is a Wing Attack an "attack" from the perspective of Sanctuary or Invisibility? Of course not. It's an attack, but it's not an attack.

[BurgerBeast]

So, just to be clear, you don't think that "If there's ever any question whether something you're doing counts as an attack, the rule is simple: if you're making an attack roll, you're making an attack" means that whenever the question "is this an attack?" is asked, you have to look if there is an attack roll, and that the answer is "it is an attack" if the attack roll is here?

1. Do the rules say that this is an attack?

a. YES -> It is an attack. Stop here.

b. NO -> Go to 2

2. Does it involve an attack roll?

a. YES -> It is an attack. Stop here.

b. No -> Undetermined. It is not explicitly an attack. It does not require an attack roll. The rules are silent.

[BurgerBeast]

Is a Wing Attack an "attack" from the perspective of Sanctuary or Invisibility? Of course not. It's an attack, but it's not an attack.

I would say it's an "attack." I understand that many will disagree (edit: and they have that right, because the answer is not clearly one way or the other), but it's not obvious in any case.

[coolAlias]

Are you claiming that a 1 HP micro-dragon with a 1d6 (DC 10 Dex save for half) 5' cone fire breath attack ought to provide zero experience points because it has no "attacks"?

There aren't all that many monsters in the MM that rely primarily on save-or-die effects instead of actual attacks, but one of them that does, a beholder, is said to "attack" at range, according to the MM, even though it has only eyestalks at range.

Adult dragons all have a "Wing Attack" which forces a Dex saving throw in an AoE. Creatures which fail that save take bludgeoning damage and are knocked prone. Is a Wing Attack an "attack" from the perspective of Sanctuary or Invisibility? Of course not. It's an attack, but it's not an attack.

In these cases (assuming they are not explicitly called out as attacks in their text - AFB), the rules do not say one way or the other with 100% certainty whether they are or are not attacks. I would expect DMs to make rulings in these cases using their best judgment.

[Unoriginal]

Are you claiming that a 1 HP micro-dragon with a 1d6 (DC 10 Dex save for half) 5' cone fire breath attack ought to provide zero experience points because it has no "attacks"?

What are you even talking about?

Even the Faerie Dragon doesn't have 1 HP.


EDIT:

Assuming that you mean an hypothetical monster created as per the Monster Creation rules:

DMG p. 275: Use the Experience Points by Challenge Rating table to determine how much XP the monster is worth. A monster of challenge rating 0 is worth 0 XP if it poses no threat. Otherwise, it is worth 10 XP

So yes, I am claiming that micro-dragon with 1 HP that needs to be in melee range to use its breath, with a breath that has a DC of 10 and cause at maximum 6 and in average 4 damages if the save is failed, is no threat to a lvl 1 PC, especially if the breath has to recharge (as dragon breaths usually do), and lacking any attacks is a significant part of this.

Such a fight would end the instant the dragon is attacked, most likely, even a mage is going to have something like 50% chances to get half damage, and even if it succeeds with its breath it's still likely to be doomed. In any case, that micro-dragon is far weaker than a Commoner, who are considered enough of a threat to warrant 10 XPs.

a beholder, is said to "attack" at range, according to the MM, even though it has only eyestalks at range.

The beholder is more than able to use ranged attacks (attack rolls and everything) by picking up objects with Telekinetic Ray and throwing them at enemies, making the description accurate. Furthermore, in game term, "attack" as a verb is not the same as "attack" the noun (the rules generally say "do an attack" or the like, not "attack").

Adult dragons all have a "Wing Attack" which forces a Dex saving throw in an AoE. Creatures which fail that save take bludgeoning damage and are knocked prone. Is a Wing Attack an "attack" from the perspective of Sanctuary or Invisibility? Of course not. It's an attack, but it's not an attack.

The Wing Attack is basically a super-Shove, so I don't see why it can't be an "attack". Especially when the Adult Dragon also has the Tail Attack legendary action just above Wing Attack, and it's described as "an attack".

[MaxWilson]

In these cases (assuming they are not explicitly called out as attacks in their text - AFB), the rules do not say one way or the other with 100% certainty whether they are or are not attacks. I would expect DMs to make rulings in these cases using their best judgment.

The DM can always make a judgment call no matter what the rules say. The point here is that there are some people here who are trying to argue that "But the thing is, the PHB NEVER refers to dragon's breath or any of those hostile and potentially damaging actions that are not attacks as attacks. Never. Not even colloquially," and yet the MM does exactly that. Sometimes, 5E's writers use "attack" colloquially or in a non-technical sense; often they mean it in a jargony sense. If you're trying to distinguish the two, you look at the guidelines explicitly provided: the presence or absence of the attack structure, which revolves around the attack roll.

If there's ever any question whether a dragon's Wing Attack can affect someone under Sanctuary per RAW, check for the presence of an attack roll. A Wing Attack has no target (unlike an attack) and has no attack roll (unlike an attack), so it's not an attack, and the dragon's Wing Beat doesn't have to target another creature instead of the one under Sanctuary. (Which is good, because a Wing Beat doesn't target creatures anyway, unlike an attack, so we aren't being required by the spell to do something nonsensical.)

A lich can use its lair action to cause spirits to "materialize and attack" a creature within its lair, for 15d6 necrotic damage (Con save for half). Does Sanctuary prevent this? Can Uncanny Dodge cut this damage in half? No, it's an attack but it's not an attack.

There's a lot of people here relying on a nonsensical reading of the text, despite the PHB giving you the only plain definition it ever gives: attack === "something with an attack roll."

[coolAlias]

The DM can always make a judgment call no matter what the rules say. The point here is that there are some people here who are trying to argue that "But the thing is, the PHB NEVER refers to dragon's breath or any of those hostile and potentially damaging actions that are not attacks as attacks. Never. Not even colloquially," and yet the MM does exactly that. Sometimes, 5E's writers use "attack" colloquially or in a non-technical sense; often they mean it in a jargony sense. If you're trying to distinguish the two, you look at the guidelines explicitly provided: the presence or absence of the attack structure, which revolves around the attack roll.

If there's ever any question whether a dragon's Wing Attack can affect someone under Sanctuary per RAW, check for the presence of an attack roll. A Wing Attack has no target (unlike an attack) and has no attack roll (unlike an attack), so it's not an attack, and the dragon's Wing Beat doesn't have to target another creature instead of the one under Sanctuary. (Which is good, because a Wing Beat doesn't target creatures anyway, unlike an attack, so we aren't being required by the spell to do something nonsensical.)

A lich can use its lair action to cause spirits to "materialize and attack" a creature within its lair, for 15d6 necrotic damage (Con save for half). Does Sanctuary prevent this? Can Uncanny Dodge cut this damage in half? No, it's an attack but it's not an attack.

There's a lot of people here relying on a nonsensical reading of the text, despite the PHB giving you the only plain definition it ever gives: attack === "something with an attack roll."

I agree in general, but arguing that grapple and shove are not attacks in the rules sense is not something I can get behind - it is clear from the PHB that they are the specific case that beats the general rule.

[Unoriginal]

Also, just to say, but Wing Attack would bypass Sanctuary because it's an AoE attack and "This spell doesn't protect the warded creature from area effects"

[MaxWilson]

Guys,

On reflection, I'm uncomfortable with how heated discussion is getting here. I just wrote a reply to Unoriginal and realized that it was waaaaay more snippy than I want it to be. I recognize that you guys all have your own opinions, and while I disagree with those opinions, I am glad to know you and glad that you're having fun playing 5E.

I reflect on this quote: "If I esteem mankind to be in error, shall I bear them down? No. I will lift them up, and in their own way too, if I cannot persuade them my way is better; and I will not seek to compel any man to believe as I do, only by the force of reasoning, for truth will cut its own way."

I've presented my reasons for my opinion (Shadow Jump, Wing Attack, lich lair actions, my view of general vs. specific and the rules text) and I won't seek compel anyone else's opinion further. I'm sorry if I hurt any feelings with my tone in recent posts. Have a good discussion!

Peace,
Maximilian

[Millstone85]

This is a pretty strong argument against, but I still don't think it succeeds. That part of the text is preceded by: "Whether you're striking with a melee weapon, firing a weapon at range, or making an attack roll as part a spell, an attack has a simple structure." So, if you are doing neither of these three things, the text that follows does not apply.

Nobody is arguing that making an attack roll means you are making an attack. (Edit: nobody is arguing against the notion. Sorry for confusion.) We're arguing the converse. There is still no blue to green.

There is blue to green, only now you are saying it is but one shade of blue leading to green. And to you, RAW doesn't forbid a dragon's breath to count as a different shade of blue.

Or maybe these three things in bold are now violet, and violet to blue to green is not the same as blue to green.

Furthermore, in game term, "attack" as a verb is not the same as "attack" the noun (the rules generally say "do an attack" or the like, not "attack").

They are the same. See the imp's Invisibility and the spell of the same name. Those end when the creature "attacks".

[PhoenixPhyre]

It comes down to reading comprehension and logic. That's all. People are bringing in other contexts that do not apply and distract from the point (it is not about game language versus everyday language; it is not about general versus specific; it's not about external biases around the truth of the matter; it's not about what the writers meant to write; it's not about JC's opinion).

Note: the rules of formal logic do not apply here. Rule statements are definitions and are this tautologies. They are not propositions. Natural language does not follow formal logic anyway. So your points about logical fallacies are irrelevant here.

Instead, we need to use the rules of conversational implicature, which are much looser. The most natural reading of this text is that the game term "attack" is defined to cover the set of interactions that use the attack roll resolution method + those interactions that are specified as attacks. As a definition, that is exhaustive and all alternatives are excluded.

[smcmike]

Note: the rules of formal logic do not apply here. Rule statements are definitions and are this tautologies. They are not propositions. Natural language does not follow formal logic anyway. So your points about logical fallacies are irrelevant here.

Instead, we need to use the rules of conversational implicature, which are much looser. The most natural reading of this text is that the game term "attack" is defined to cover the set of interactions that use the attack roll resolution method + those interactions that are specified as attacks. As a definition, that is exhaustive and all alternatives are excluded.

Yeah, I can sign on fully to this statement.

[Millstone85]

Yeah, I can sign on fully to this statement.

That's what it pretty much boils down to for me as well. The PHB has a two pages long section called Making an Attack, with an introductory text that lists situations where you are making an attack and the simple three-step process that follows. You are going to assume the game is defining an attack before going into further details (Attack Rolls, Unseen Attackers and Targets, etc.) and Crawford's tweets confirm that was the expected response. It takes effort to realize that an attack, as a game term, hasn't been formally defined. But at this point you are trying to defeat the PHB, not understand it.

Somewhere inside of me, a modron is crying because it looked at a mirror and saw a slaad. On the bright side, I now have a pretty good idea of what the Infernal language must be like, with every logical connection precisely defined forward and backward, like "when-and-only-when-but-not-conversely".

[BurgerBeast]

There is blue to green, only now you are saying it is but one shade of blue leading to green. And to you, RAW doesn't forbid a dragon's breath to count as a different shade of blue.

Or maybe these three things in bold are now violet, and violet to blue to green is not the same as blue to green.

They are the same. See the imp's Invisibility and the spell of the same name. Those end when the creature "attacks".

I'm not sure how this got confused. Here it is:

TRUE: attack roll -> attack

NOT TRUE: attack -> attack roll

That's it.

Note: the rules of formal logic do not apply here. Rule statements are definitions and are this tautologies. They are not propositions. Natural language does not follow formal logic anyway. So your points about logical fallacies are irrelevant here.

We don't need formal logic. You can err logically when using it informally. You can read plain English in an illogical way. The only purpose for bringing formal logic into it is to illustrate that. If you're saying that formal logic doesn't apply, then you're saying logic doesn't apply, and the conversation is basically over. I'm not interested in illogical conversation.

At the end of the day you're just avoiding the issue, in a different way. If you think the RAW say:

attack -> attack roll

then you're wrong. Formally and informally and also by the rules of conversation (insofar as it is logical).

Instead, we need to use the rules of conversational implicature, which are much looser. The most natural reading of this text is that the game term "attack" is defined to cover the set of interactions that use the attack roll resolution method + those interactions that are specified as attacks. As a definition, that is exhaustive and all alternatives are excluded.

You can say this, but it has no bearing. The "rules of conversational implicature," whatever those may be, do not mean you get to make up what you want. The book says what it says. The words have meanings.

This is more or less woo-woo. You're waving your hands and saying you don't want the words to say what they say. I'm sorry, there's no magic to be found in the words "conversational implicature."

Yeah, I can sign on fully to this statement.

That's bizarre. This is a much bigger can of worms than even MaxWilson's contrivance. PheonixPhyre has basically decided that "the rules of conversational implicature" are a license to declare that the book says whatever he wants it to say (because it is apparently exempt from logic).

[BurgerBeast]

And now three of the people on this thread are going to invent and hide behind vagaries because they don't want to deal with the cognitive dissonance presented in the truth of the matter.

@smcmike and Millstone85: you'e both obviously very intelligent. Don't buy the woo-woo. Think it through. It's difficult enough to find a point in there, never mind agree with it.

Would you let me get away with saying that you guys don't understand the "conversational vagaries" of the PHB, or that you're too concerned with logic when reading a rulebook, and that's why your arguments do not apply? I should hope not.

[BurgerBeast]

Also, just to say, but Wing Attack would bypass Sanctuary because it's an AoE attack and "This spell doesn't protect the warded creature from area effects"

I thought that by your definition, AOEs are not attacks. Am I missing something? Are you saying it's an attack because the word "attack" is found in the name?

[smcmike]

That's bizarre. This is a much bigger can of worms than even MaxWilson's contrivance. PheonixPhyre has basically decided that "the rules of conversational implicature" are a license to declare that the book says whatever he wants it to say (because it is apparently exempt from logic).

Nah. I agree, as you argue, that the text doesn't explicitly rule out the possibility that an action may be an attack without being named an attack by the text or meeting the requirements of the test set forth in the text.

I also agree with PhoenixPhyre that the most reasonable interpretation of the text is that for an action to be an attack, it must either meet the test set forth in the text to determine if something is an attack, or be labeled an attack by the text.

I admit that this interpretation goes beyond literal RAW, and am unconcerned, because literal RAW always requires interpretation, and the intended meaning of the text is very clear to me.

As you post out, this is similar to past disagreements, though this one is likely to be less contentious, as you have not decided that the RAW rules out my interpretation this time.

[PhoenixPhyre]

We don't need formal logic. You can err logically when using it informally. You can read plain English in an illogical way. The only purpose for bringing formal logic into it is to illustrate that. If you're saying that formal logic doesn't apply, then you're saying logic doesn't apply, and the conversation is basically over. I'm not interested in illogical conversation.

At the end of the day you're just avoiding the issue, in a different way. If you think the RAW say:

attack -> attack roll

then you're wrong. Formally and informally and also by the rules of conversation (insofar as it is logical).

You can say this, but it has no bearing. The "rules of conversational implicature," whatever those may be, do not mean you get to make up what you want. The book says what it says. The words have meanings.

This is more or less woo-woo. You're waving your hands and saying you don't want the words to say what they say. I'm sorry, there's no magic to be found in the words "conversational implicature."

BurgerBeast, first and foremost,
humans are not logical creatures and neither is human language. Ambiguity and implications abound. Idiom and non-literal utterances are probably more common than literal utterances.

Implicature is the term for understanding what is implied by a statement. Since we do not speak (or write) according to the axioms of formal logic (and thus invoking logical fallacies is not dispositive), there is much more implied than is entailed by the words we use. Thus we need to use implicature to understand natural language.

Formal logic (and statements about propositions, truth values, etc) is only useful when the preconditions are met. Definitions (like are found in the rules) are tautological in nature--they either (depending on your point of view) have no truth value (which is not the same as being false) or are always true. You have to use different rules for understanding informal writing (such as game rules) than for understanding formal statements. Natural language cannot be analyzed using the mechanisms of formal logic--if it could, it would be trivial to implement a formal grammar for english (and thus make it possible for computers to understand and translate natural language without error, a thing that is known to be impossible).

Reading strictly by the words written and ignoring implications leads to absurd results. It is the cause of most of the dysfunctions of previous editions and is a tradition that should be abandoned. The rules were never intended or written to be read in that fashion. Nothing in the PHB or DMG (or MM for that matter) would survive such a reading intact. So don't. Take the common face meaning and go with it, making rulings as needed. RAW should really be RACU (Rules as Commonly Understood). Everyone would be better off and there would be much less fighting about trivialities.

We should apply heuristics similar to those used in the legal system, called canons of construction. Game canons are necessarily different than the legal canons, but will have strong similarities. One of those canons is "the words mean what they say they mean, not more and not less." This includes abilities and spells--they do what they say they do. Importing "real world logic" or other such things makes a mess. Don't do it (please?). Another would be "the simplest explanation that makes things playable is probably the right one." Yet another is "Hard cases make bad law," (meaning don't make the rules based on the exceptions. Exceptions are exceptional, after all. Handle them separately instead of trying to shove them into the general case).

Spoiler: legal side note

Show

As a side note, the "but it doesn't say it isn't" excuse (which is what you're pulling here) would not fly in a court of law. Logic or no logic. There are no magic words--courts (and DMs) rule based on what "reasonable people" (a term of art in the legal world) understand the words to mean. The antics of a RPG rules lawyer would get that person sanctioned by the court and would result in their client losing the case, probably on summary judgement.

[bid]

I'm not sure how this got confused. Here it is:

TRUE: attack roll -> attack

NOT TRUE: attack -> attack roll

That's it.

Yep.

TRUE: all attack rolls are attacks
TRUE: all non-attacks are non-attack-rolls
UNKNOWN: All attacks are attack rolls
UNKNOWN: All non-attack-rolls are non-attacks

So, it is possible that a certain attack does not use an attack roll, we can't know this from p194 generic rule.

[Unoriginal]

I thought that by your definition, AOEs are not attacks. Am I missing something? Are you saying it's an attack because the word "attack" is found in the name?

Basically, yes.

Wing Attack is a legendary action, and it's basically a super-shove, so to speak. Tail Attack is also a legendary action of the same creature.

I've never said that AoEs can't be attacks, I'm saying that in the vast majority of cases they are not. Fireball isn't an attack, for exemple. But if something is called an attack by the rules, then it's an attack regardless of how it works.


But what do I know? I'm just an idiot whose contribution doesn't interest nor help anyone.

UNKNOWN: All attacks are attack rolls
UNKNOWN: All non-attack-rolls are non-attacks

Those two are known. Not all attacks are attack rolls (Grapple and Shove are attacks but don't involved attack roll) and not all non-attack rolls are non-attacks (Grapple and Shove don't involve attack roll yet are attacks)

However, those exceptions are explicitly said in the rules.

[PhoenixPhyre]

Yep.

TRUE: all attack rolls are attacks
TRUE: all non-attacks are non-attack-rolls
UNKNOWN: All attacks are attack rolls
UNKNOWN: All non-attack-rolls are non-attacks

So, it is possible that a certain attack does not use an attack roll, we can't know this from p194 generic rule.

Actually we know that at least two attacks are not resolved with attack rolls: grappling and shoving. Those are specifically called out as attacks (specific beats general) but use the "contested ability check" resolution method. That means that the 3rd statement is false: at least two attacks are not attack rolls. This also makes the fourth statement false (as @Unoriginal pointed out above).

We don't need to apply formal logic, though--the general rule covers everything.

GENERAL: Attacks have a) targets, b) modifiers, c) attack rolls (PHB 194). Note the first sentence of the subsection titled "Attack Rolls":

When you make an attack, your attack roll determines whether the attack hits or misses.

Combined with the preceding subsection, this is a general definition for the term "attack". The presence of an attack roll is necessary and sufficient to indicate an attack. No other general rule for attacks is present anywhere in the text. Thus, this is the general rule and applies everywhere except for explicit exceptions.

SPECIFIC 1: Grappling. As pointed out elsewhere, PHB 195 calls this out as an attack. Since specific beats general, it's an attack despite not using an attack roll for resolution. Note the statement (same page): "you try to seize the target by making a grapple check instead of an attack roll." This is a specific replacement for this instance only.

SPECIFIC 2: Shoving. "Instead of making an attack roll, you make a Strength (Athletics) check..." Same considerations apply. Also an attack that uses the contested ability check instead of an attack roll. Still an attack.

Others may exist, but they have to be specifically called out as exceptions, otherwise the general rule applies. That goes for every rule in the book. Exceptions are exceptional and must be explicitly mentioned. Otherwise the general rule applies. Any other understanding makes the game harder to run and only advantages munchkins looking to break things.

[smcmike]

You in the law, Phoenix?

[PhoenixPhyre]

You in the law, Phoenix?

No, but my dad is and I've done lots of reading on the topic. I also grew up with an example of the worst stereotypes of the high-school debater (sophistry, twisting definitions, bad faith arguing, "winning" as the goal of conversations, verbal abuse, etc). This made me particularly sensitive to such tactics.

[bid]

Actually we know that at least two attacks are not resolved with attack rolls: grappling and shoving. Those are specifically called out as attacks (specific beats general) but use the "contested ability check" resolution method.

Yes, but you're going beyond p194. Nothing on p194 says that grappling cannot be attacks. You'd have to use other pages of the PHB to reach that conclusion.

Even if it is questionable that an "attack" is an "attack", p194 doesn't help.