Sage Advice Compendium Update 1/30/2019

[FrogReaver]

What's the point? [MENTION=16814]Ovinomancer[/MENTION] and I have been responding to this using the words in the PHB. There is no text that says "due to the way Sanctuary works, the Attack action is separate from the actual attacks". There is no text that talks about the duration of an action. You can perform the Cast a Spell action and not actually cast a spell, due to it being Counterspelled. You seem to have latched onto Sanctuary as the proof that your interpretation is correct, but I fundamentally disagree and have posted at length about how I believe the Attack action and actions in general work (i.e. the Attack action is making an attack, Extra Attack gives you multiple attacks, it's all part of the action, there's a specific rule that says you can insert movement between attacks, etc etc etc).

JEC has talked at length about how spells in 5E work, specifically that all you need to know about a particular spell is the words in that spell alone. You don't need to refer to other spells or other features of the game, you simply do what the spell says. Sanctuary says that if you try and attack a creature protected by the spell and fail your Wisdom save, that attack is lost. It does not say you get to go back in time and choose a different action, or that you can Shield Master shove any time you like, it simply says that if you try and attack a creature protected by this spell and fail your save, you can't attack the target. We don't need to read anything more from the text of the spell, the effect is quite simple and quite clear: if you fail your save, you can't attack the target protected by the spell. Just do what it says. No need to take these words and infer that some other portion of the game must behave in a way that is not in the text of those rules, this spell simply provides an exception to the general rules of the game.

How about actually responding to my post? See I can repeat myself too....

 

[Asgorath]

Logical Inferences are integral to being about to understand complex topics. The Rules don't have to spell absolutely everything out as we can reason. If reasoning is applied to a given interpretation and it results in something you disagree with you can not reasonably use the notion that the rules would have to state that. What's being provided is a logical deduction from your stated interpretation (the facts you've presented about your interprestion) and what the rules actually say. The whole idea is that given your truths and the truth of the rules we will be able to reason out other facts. If your arguing against that reasoning based on your interpretation "facts" and raw that leads to some conclusion you don't agree with then at least present a reasonable argument as to why the reasoning fails. Saying there's no rule doesn't cut it in such a situation.

Or, perhaps the rules of the game are designed to be simple and straight forward, and not require a ton of reasoning and interpretation. As [MENTION=16814]Ovinomancer[/MENTION] so eloquently put it, just do what it says on the tin.

Occam's razor: the Attack action means making attacks. Sanctuary means the attack fails if you fail your saving throw. "If you X, you can Y" means you have to actually do X before you can do Y. There's a rule that says you can split your movement before and after your action, as well as between attacks in the Attack action.

If I was a game designer and trying to build a rule system that was easy for new players to understand, then I'd lean towards my previous paragraph and not something that relies on knowledge of every single word in the PHB and that the text of the Sanctuary spell radically changes the way the entire action system works (with no words about this in the actual text of that action system itself) and thousands of other obscure rules interactions. There's a really simple solution here, and my position is that this is the correct one. Which also just happens to be confirmed by the Sage Advice Compendium.

 

[Asgorath]

As long as the attack action is comes before the actually attacking (as the sanctuary example helped reveal) then you can take the attack action, shield master shove, attack then attack again. Nothing in our interpretation goes against this quote.

Your interpretation is wrong, because the action is not separate from the attacks. The PHB says "With this action, you make one melee or ranged attack." If you disagree with this, please quote the PHB text that says the Attack action is separate from the attacks.

I agree. But the attack action happens before your actual attacks. That's the key to piecing it back together.

No it doesn't, per the PHB text I just quoted above.

Because the sanctuary discussion revealed that you can take the attack action and never actually attack.
Because the disengage discussion revealed that you must take the disengage action and have it end before you can move again.

The individual evidences may not be wholly convincing. I get that. But the evidences all considered as a whole together definitely make a strong case that the actions in general and more importantly, the attack action all happen before the attacks it provides.

The Disengage action provides a temporary buff, much like the Shield spell. Once you have taken the action, your movement no longer provokes OAs. Nowhere in the PHB does it say the Disengage action lasts for the same duration as the effect it applies, in fact the PHB does not talk about action duration at all. Thus, the logical conclusion is that action duration isn't a thing and has no meaning or relevance.

Sanctuary proves otherwise. You don't make an attack with sanctuary, you lose your attack.

It really doesn't prove this, though. Sanctuary simply means you either have to make your saving throw, or you fail to attack the target. This is very similar to the Cast a Spell action and Counterspell, if you lose your spell from Counterspell you still took the Cast a Spell action.

Okay, then I'll just shield master shove after choosing the target for my attack. You agree I can take the shove anytime on my turn after I've taken the attack action right? And since I've taken the attack action after choosing my target then I should be able to shield bash before determining modifiers and resolving the attack right?

The PHB says the Attack action is "with this action, you make one melee or ranged attack". The rules for making an attack list 3 steps which must be followed. There is nothing in the PHB about splitting an action and taking a bonus action in between these 3 steps. We shouldn't have to list out every single thing that you aren't allowed to do, we simply just do what the PHB says we can do. Nothing more, nothing less.

Do you have a rule or any evidence for this? Our side has provided evidence that actions come before their effects, including the attack action. What supports your belief that it doesn't?

There are actions that provide buffs that last for a duration. The length of the effect is explicitly listed in the action itself. The Attack action has no such language, and thus it does not work like those other actions (e.g. Disengage, Dodge).

No. As long as you have taken the attack action then whether you actually can attack or not is irrelevant. In our scenario it would be take attack action. shield master shove. get stunned. SHOOT I lost my attacks. DM reviews situation and finds nothing that broke the rules as the attack action was taken before the shield master shove.

Taking the Attack action is making the attack(s), per the text in the PHB.

I can't answer for Arial but I don't want to be able to use it on my turn. I wanted to be right. My original belief had always been that the rules only allow the shield master shove after the attacks. I was thrilled when JC joined my side. So no, I don't want to use shield master after the attack action. I would much rather have been right.

However, I have found compelling evidence in this thread that I was wrong.

You seem to have latched onto the words of one spell and using this to prove your interpretation, but that has just meant you've drawn the wrong conclusion.

Bad arguments deserve that response. The argument in this case is very persuasive.

It's really not persuasive, though.

Yep, even if JC had been right in this case then I would have houserulled it. But my reason was that I'd played it by his original ruling for so long that I wasn't changing it now. It is nice knowing I don't have to houserule it though.

The Sage Advice Compendium is very clear about how this feat is supposed to work. You can obviously play it however you like, but it's a real stretch to claim that JEC is simply wrong and that obscure words from other portions of the game radically change the action system.

Did we really learn anything new here? Did I change your mind? I doubt it, which is why I wasn't really planning on replying to your entire post again.

 

[DND_Reborn]

Okay, so I am about to go to bed and I will make one final attempt to clarify this so Asgorath isn't doing all the work (good job, btw!). From tweets posted by JC and others on May 11, 2018:

#1. Jonathan Ellis
What was it supposed to be? How is using a bonus action to knock someone prone and then attack cheesy?

#2. Jeremy Crawford
It's supposed to be what it is: a way to knock someone prone after your attack. It's essentially a finishing move.

Please note two things in JC's reponse to Mr Ellis's questions.

1. "a way to knock someone prone after your attack." Not after your Attack action, after your attack. If you have not attacked, you have not satisfied the condition. Taking the Attack action means you are attacking. Until you have resolved the attack, you are not "after" it.

2. "It's essentially a finishing move." A finishing move. Not a in-the-middle-of-my-attacks move, finishing move.

A later tweet also from May 11, 2018:

#3. Jeremy Crawford
If taking the Attack action is the condition for something else happening, you must take that action before the other thing can happen, unless the rules state otherwise. The action as a whole is the condition.

We all agree (I believe) with the interpretation of the first part, "If taking the Attack action is the condition for something else happening, you must take that action before the other thing can happen, unless the rules state otherwise." The rule for Shield Master is "If you take the Attack action on your turn, you can use a bonus action to try to shove a creature..." We all agree you must take the Attack action in order to gain the benefit of the bonus action--there is no doubt about that.

Now, the case is being argued that you are taking the Attack action, just not making your attacks yet. This is the point where we deviate so I will continue with the next sentence:

"The action as a whole is the condition."

Since the action as a whole is the condition, you must take it in its entirety, not piece by piece, in order to satisfy the condition. Trying to take it piece-by-piece as some have reasoned violates that "The action as a whole is the condition." ruling. Therefore, you cannot: take the Attack action, bonus action Shove, attack. If you try to do so, you are not taking the action as a whole, which is the condition, thus denying yourself the Bonus action with which to Shove.

Therefore you must do the following:
1. Take the Attack action since that is required for the Bonus action as its condition. (No arguments here.)
2. Since the action as a whole is the condition, you must take it and complete it, not break it apart, before you have satisfied the condition.

Once you take the Attack action as a whole (that's the condition from #3 above), you then gain the benefit of the bonus action to Shove. If you don't take the Attack action as a whole, you don't get the bonus action to Shove.

You don't agree with JC's rulings, fine, as always it is up to the DM and the table to play how they see fit. If you do so, however, hopefully you will now understand why it is against the official rules and a house-rule. I don't see why anyone has an issue with that, since many of us play with at least some house-rules, after all.

Good night!

 

[epithet]

Or, perhaps the rules of the game are designed to be simple and straight forward, and not require a ton of reasoning and interpretation. As [MENTION=16814]Ovinomancer[/MENTION] so eloquently put it, just do what it says on the tin.

Occam's razor: the Attack action means making attacks. Sanctuary means the attack fails if you fail your saving throw. "If you X, you can Y" means you have to actually do X before you can do Y. There's a rule that says you can split your movement before and after your action, as well as between attacks in the Attack action.

If I was a game designer and trying to build a rule system that was easy for new players to understand, then I'd lean towards my previous paragraph and not something that relies on knowledge of every single word in the PHB and that the text of the Sanctuary spell radically changes the way the entire action system works (with no words about this in the actual text of that action system itself) and thousands of other obscure rules interactions. There's a really simple solution here, and my position is that this is the correct one. Which also just happens to be confirmed by the Sage Advice Compendium.

I think most designers building rules that are easy for new players to understand try to keep in mind that while the game might not be emulating a fantasy world, it is modelling it. Thus, trying to have your rules make sense and carry a level of verisimilitude helps the new player get into character and understand the range of what's possible in the game.

If this new player plays a champion, for example, he'll pick up Extra Attack at level 5, and might have taken the Shield Master feat at level 4. It's totally straightforward to understand that "so if you attack this turn, you get to make two attacks plus a bonus shield shove." Even the newest player can understand that. What doesn't make any sense at all is "yeah, you can't make that shove until after the other two attacks. I mean, you can totally shove as one of those attacks, but then you still have the shove at the end." It doesn't make any sense, it wrecks immersion because you have to get out of the moment to parse the text in nit-picking gamist terms which are obviously ambiguous.

To me, one of the obvious clues that Jeremy is just pulling this out of his posterior is the whole "finishing move" assertion. This is pretty clearly not intended to be a finishing move, because you don't have to shove someone you've already attacked. Even according to Jeremy's new and not-at-all-improved advice on the feat, you can cut down an enemy before you move across the room, open a door, find a new target, and shove it with your shield without doing any damage. The only way that's a finishing move is that it finishes your turn. Either Jeremy has not idea what a finishing move is (not likely,) or he's just throwing turds at the wall to see which ones stick. Obviously the "finishing move" fewmet found a couple of people willing to repeat it (apparently with sincerity and not ironically) in this thread, but it is a ridiculous claim on its face.

Wizards has gotten its messaging under control, so we shall apparently never again see Mike or Chris give opinions on rules interpretations. That's unfortunate in this instance, because I seriously doubt that Jeremy's revised Advice on Shield Master matches the original intention for the feat when it was written. Could I be wrong? Of course, it happens all the time, but it seems very unlikely to me that Jeremy simply forgot that he imagined that if means after, and that there was supposed to be a timing requirement. For two years!

The issue here isn't really the Shield Master feat and its bonus action shove. That's easily fixed with either a different interpretation of the rules or a house rule like mine that eliminates the need for the Attack Action altogether. "Bash and dash," I call it. The real issue for me is the direction that this change in the Sage Advice represents, specifically a willingness to sacrifice modeling believable action in the service of gamist overspecificity and hyper-literal parsing. I don't want to see the playability of future content compromised by the attitude this type of "official ruling" represents, an attitude that moves away from the improvisation and in-the-moment inspiration that makes D&D immeasurably superior to board games or computer RPGs.

A hundred pages of discussion in this thread alone make it abundantly clear that there are several ways to read the Shield Master feat and the rules with which it interacts. Many of you seem to hold the belief that (despite the fact that he insists that he was wrong before) Jeremy Crawford's statements regarding the interpretation of this rule are The One True Way. As I see it, he's either wrong now or he was wrong before, and I'll take whichever one makes more sense to me. When you have more than one way to reasonably interpret a rule of D&D, the way that leads to arbitrarily locking a player character into an invariable pattern and restricting a player's ability to determine what his character can do on its turn is the wrong damn interpretation.

 

[Ovinomancer]

I think most designers building rules that are easy for new players to understand try to keep in mind that while the game might not be emulating a fantasy world, it is modelling it. Thus, trying to have your rules make sense and carry a level of verisimilitude helps the new player get into character and understand the range of what's possible in the game.

If this new player plays a champion, for example, he'll pick up Extra Attack at level 5, and might have taken the Shield Master feat at level 4. It's totally straightforward to understand that "so if you attack this turn, you get to make two attacks plus a bonus shield shove." Even the newest player can understand that. What doesn't make any sense at all is "yeah, you can't make that shove until after the other two attacks. I mean, you can totally shove as one of those attacks, but then you still have the shove at the end." It doesn't make any sense, it wrecks immersion because you have to get out of the moment to parse the text in nit-picking gamist terms which are obviously ambiguous.

To me, one of the obvious clues that Jeremy is just pulling this out of his posterior is the whole "finishing move" assertion. This is pretty clearly not intended to be a finishing move, because you don't have to shove someone you've already attacked. Even according to Jeremy's new and not-at-all-improved advice on the feat, you can cut down an enemy before you move across the room, open a door, find a new target, and shove it with your shield without doing any damage. The only way that's a finishing move is that it finishes your turn. Either Jeremy has not idea what a finishing move is (not likely,) or he's just throwing turds at the wall to see which ones stick. Obviously the "finishing move" fewmet found a couple of people willing to repeat it (apparently with sincerity and not ironically) in this thread, but it is a ridiculous claim on its face.

Wizards has gotten its messaging under control, so we shall apparently never again see Mike or Chris give opinions on rules interpretations. That's unfortunate in this instance, because I seriously doubt that Jeremy's revised Advice on Shield Master matches the original intention for the feat when it was written. Could I be wrong? Of course, it happens all the time, but it seems very unlikely to me that Jeremy simply forgot that he imagined that if means after, and that there was supposed to be a timing requirement. For two years!

The issue here isn't really the Shield Master feat and its bonus action shove. That's easily fixed with either a different interpretation of the rules or a house rule like mine that eliminates the need for the Attack Action altogether. "Bash and dash," I call it. The real issue for me is the direction that this change in the Sage Advice represents, specifically a willingness to sacrifice modeling believable action in the service of gamist overspecificity and hyper-literal parsing. I don't want to see the playability of future content compromised by the attitude this type of "official ruling" represents, an attitude that moves away from the improvisation and in-the-moment inspiration that makes D&D immeasurably superior to board games or computer RPGs.

A hundred pages of discussion in this thread alone make it abundantly clear that there are several ways to read the Shield Master feat and the rules with which it interacts. Many of you seem to hold the belief that (despite the fact that he insists that he was wrong before) Jeremy Crawford's statements regarding the interpretation of this rule are The One True Way. As I see it, he's either wrong now or he was wrong before, and I'll take whichever one makes more sense to me. When you have more than one way to reasonably interpret a rule of D&D, the way that leads to arbitrarily locking a player character into an invariable pattern and restricting a player's ability to determine what his character can do on its turn is the wrong damn interpretation.

Dude, lower your voice; hitpoints are right there.

More seriously, your opinions about what constitutes fictional fidelity with actions is your opinion. It's not the design basis for the game designers (who will be using thier own). D&D is chock full of mechanics that require resolution before being described in the fiction; you're just used to them and no longer notice.

Personally, I don't have any problem with narrating the shield bash after the attack. The attack sets up the foe by unbalancing them enough for a skilled warrior to take advantage with a well-timed shield bash that sends the foe staggering or knocks them down. Ta-da!

I think the problem here is that you already have a fiction in mind and want the gane to model that instead of seeing what the gane models and then narrating that.


In another ongoing thread, I made the observation that there are players that want the fictional state to drive the mechanics, ie they consider all of the fictional inputs going into the action and choose the appropriate mechanical test that fits those inputs. Call this resolution at the end. This method tends toward identifying the outcome and then resolving if that happens or not.

On the other side, if there's a question mechanics are used and the results determine what some of the fictional inputs must have been to generate that outcome. Call this resolution in the middle. This method leaves outcomes more open ended to accommodate fiction.

D&D cribs from both. Ability checks are endian because the current situation determines the check type and DC. Hitpoints are middlish because the fictional outcome adapts to the resolution. Shield Master is between these two ends. It's inputs are not based on the fiction, but it's outputs are fixed in the fiction. This causes issues with players that prefer the first path.

4e is chock full of this kind of middlish resolution, and that was/is one of the major fronts of the edition war.

 

[DND_Reborn]

I think most designers building rules that are easy for new players to understand try to keep in mind that while the game might not be emulating a fantasy world, it is modelling it. Thus, trying to have your rules make sense and carry a level of verisimilitude helps the new player get into character and understand the range of what's possible in the game.

If this new player plays a champion, for example, he'll pick up Extra Attack at level 5, and might have taken the Shield Master feat at level 4. It's totally straightforward to understand that "so if you attack this turn, you get to make two attacks plus a bonus shield shove." Even the newest player can understand that. What doesn't make any sense at all is "yeah, you can't make that shove until after the other two attacks. I mean, you can totally shove as one of those attacks, but then you still have the shove at the end." It doesn't make any sense, it wrecks immersion because you have to get out of the moment to parse the text in nit-picking gamist terms which are obviously ambiguous.

To me, one of the obvious clues that Jeremy is just pulling this out of his posterior is the whole "finishing move" assertion. This is pretty clearly not intended to be a finishing move, because you don't have to shove someone you've already attacked. Even according to Jeremy's new and not-at-all-improved advice on the feat, you can cut down an enemy before you move across the room, open a door, find a new target, and shove it with your shield without doing any damage. The only way that's a finishing move is that it finishes your turn. Either Jeremy has not idea what a finishing move is (not likely,) or he's just throwing turds at the wall to see which ones stick. Obviously the "finishing move" fewmet found a couple of people willing to repeat it (apparently with sincerity and not ironically) in this thread, but it is a ridiculous claim on its face.

Wizards has gotten its messaging under control, so we shall apparently never again see Mike or Chris give opinions on rules interpretations. That's unfortunate in this instance, because I seriously doubt that Jeremy's revised Advice on Shield Master matches the original intention for the feat when it was written. Could I be wrong? Of course, it happens all the time, but it seems very unlikely to me that Jeremy simply forgot that he imagined that if means after, and that there was supposed to be a timing requirement. For two years!

The issue here isn't really the Shield Master feat and its bonus action shove. That's easily fixed with either a different interpretation of the rules or a house rule like mine that eliminates the need for the Attack Action altogether. "Bash and dash," I call it. The real issue for me is the direction that this change in the Sage Advice represents, specifically a willingness to sacrifice modeling believable action in the service of gamist overspecificity and hyper-literal parsing. I don't want to see the playability of future content compromised by the attitude this type of "official ruling" represents, an attitude that moves away from the improvisation and in-the-moment inspiration that makes D&D immeasurably superior to board games or computer RPGs.

A hundred pages of discussion in this thread alone make it abundantly clear that there are several ways to read the Shield Master feat and the rules with which it interacts. Many of you seem to hold the belief that (despite the fact that he insists that he was wrong before) Jeremy Crawford's statements regarding the interpretation of this rule are The One True Way. As I see it, he's either wrong now or he was wrong before, and I'll take whichever one makes more sense to me. When you have more than one way to reasonably interpret a rule of D&D, the way that leads to arbitrarily locking a player character into an invariable pattern and restricting a player's ability to determine what his character can do on its turn is the wrong damn interpretation.

Well said! Of course, I don't know so much about the reference to Jeremy pulling something out of his posterior...

Otherwise, I think you are right. While my posts have been vehemently in favor of understanding the official stance, I don't agree with it either. Like you I agree the Shove attack works narratively coming first, in the middle, or last. It makes no difference to me personally and I would play with a DM who runs it either way without any qualms. I feel it is more useful allowing the Shove to happen first, but not essential. Other aspects of the feat are still very good given the proper situations.

Writing Shield Master as "If you take the Attack action on your turn, you may take an additional attack that you can use to shove a creature within 5 feet of you with your shield."

Now, this doesn't cost action economy since you aren't spending your Bonus action, but it shouldn't be OP even given that and further testing would prove it out either way. It could be worded that making the additional attack to shove deprives you of taking any bonus action on that turn.

Another option would be this (or along similar lines)

"If you use one of your attacks to shove a creature on your turn, you gain advantage on the Strength (Athletics) check for that attack by using your shield."

Again, word to deprive bonus action or not.

In SA Jeremy comments that the Eldritch Knight ability, War Magic, "That said, a DM would break nothing in the system by allowing an Eldritch Knight to reverse the order of the cantrip and the weapon attack."

Does it break anything by allowing the shove prior to the attack? Probably not. However, apparently enough issues came up that the powers that be felt it was important to reverse his prior ruling. And so now we are stuck with the shove coming after you've completed the Attack action as a whole. People can house-rule it or ignore the reversal if it suits their game, or just play without Feats and then they don't have to worry about it at all!

 

[Hriston]

I see how you're doing this, and that's great, I'm glad it works for you, but it's not how the rules read. You don't get to "you haven't not gained it yet" when you can't get a bonus action until given one. That's not how the rules work, although I see no serious issue to you ruling that way for your table.

But you are given one by virtue of having the feat. That's my point. Saying, "you can't get a bonus action until given one", is a paraphrase of the section I quoted, which is about "class features, spells, and other abilities" letting you take one. It's a clarification that bonus action is not a default part of the action economy that a player should be thinking about filling on every turn, like similar things are in previous editions. If you have Shield Master, then you do have a bonus action to take. And sure, the condition needs to be satisfied, but that's not what this passage is about.

No, because you're then skipping to the end to check the condition to apply the result to the beginning. Not how conditionals work. If X, Y requires X to be true either before or at the same time as Y, not that Y can exist so long as X eventually does.

But X is true at the same time as Y because it's true of your entire turn. You can't both take the Attack action on your turn and not take the Attack action on your turn. It's one or the other.

Possibly, except we have another rule, the one that says you do not have a bonus action until given one. In that case, you cannot go to the bookstore (bonus action) because the bookstore doesn't exist until you take money out of the ATM. Okay, that example got weird, but still, that's how it works.

Okay, I was ignoring the bonus action part of the example because of the weirdness, but what I've been saying about Shield Master still holds. The rule you're citing says you don't have a bonus action to take unless "a special ability, spell, or other feature of the game states that you can do something as a bonus action." Shield Master states (with conditions) that you can shove a creature as a bonus action. Alternatively, [MENTION=6987520]dnd4vr[/MENTION]'s example states (with conditions) that you can go to the bookstore and buy a book as a bonus action.

You cannot take a bonus action to shove until you've taken the Attack action on your turn. If X, Y means X cannot be a future event if Y occurs, it must be a current event.

Right, and my argument is that it's current because of the "on your turn" language. Perhaps an interpolation would help: If you [do] take the Attack action on your turn, you can use...

The fact your interpretation jumps to the end of the turn to check if the Attack action has occurred and then goes back to earlier to allow the bonus action prior to the Attack action. Since you've been clear that declaration isn't how you do this, then you have to be allowing a end-of-turn check to justify the bonus action.

It's different than that, though. Until satisfaction of the condition can be checked for (which, at the latest, is at the end of your turn), all that has happened is a shove-attempt. Once the moment of your turn is reached in which you take the Attack action, then the condition for using a bonus action is met, and the bonus action is assigned to the shove-attempt. Now, that may seem like a retcon, or "going back in time", but to me it isn't because it doesn't change any established events in the fiction.

Did I miss something? What trouble did War Magic run into?

I’m really glad you asked me this. Here’s the story as far as I can reconstruct it. On July 6, 2015, Jeremy Crawford answered this question on twitter:

Does the “when” in the Eldritch Knight’s War Magic feature mean the bonus attack comes after you cast the cantrip, or can it come before?​

This response, an expansion of his earlier tweeted response, appears in the "RULES ANSWERS: JUNE 2016" Sage Advice article:

The intent is that the bonus attack can come before or after the cantrip. You choose when to take a bonus action during your turn, unless the bonus action specifies when it must take place (PH, 189).​

I want to stop right there to note that here we have the statement of RAI for bonus actions like the War Magic weapon attack, of which the Shield Master shove is one. I’d also like to note how highly unlikely it is that Crawford was drunk and in line at Trader Joe’s when he tweeted both this response as well as the one he tweeted on January 21, 2015 about the Shield Master shove.

In the August, 2017 Sage Advice Compendium, however, Jeremy Crawford changed his answer to that question from his previous RAI answer to the following RAW interpretation:

The bonus action comes after the cantrip, since using your action to cast a cantrip is what gives you the ability to make the weapon attack as a bonus action. That said, a DM would break nothing in the system by allowing an Eldritch Knight to reverse the order of the cantrip and the weapon attack.​

After that, on May 11, 2018, someone asked him on Twitter if the same principle applied to his 2015 ruling on Shield Master, which led to him changing that ruling as well.

So the Eldritch Knight’s War Magic is really what kicked this all off, and I suspect the reason Crawford decided to abandon his RAI ruling had something to do with the fact that while Shield Master, Two-Weapon Fighting, Polearm Master, etc. all grant bonus action attacks conditioned on taking the Attack action (with possible additional conditions), War Magic grants a weapon attack conditioned on casting a cantrip (presumably by taking the Cast a Spell action). And while Crawford acknowledges in his revised War Magic ruling that it breaks nothing to reverse the order established by his ruling, it does bring up the issue that if you make the weapon attack first and are then prevented from casting your cantrip, it changes your action from Cast a Spell to Attack. That isn't necessarily a problem either, but it might have been something that swayed him.

Okay, this is a pretty clear version of your reading, but it runs into a few problems. One, the Attack action does NOT grant the Shield Master shove, so whether or not it's a game feature is completely irrelevant. The conditional is built into the Shield Master feat, which is a class (or racial, for variant humans) feature. That conditional says that you only get the bonus action if you take that Attack action on your turn. So, you paraphrase is eliding very important information -- namely the conditional nature of the bonus action. You can't delete information and claim to be reading RAW, or RAI, for that matter. This entire tangent doesn't advance your case.

I agree with your three points above. This post wasn't so much meant to advance my general position, however, as it was to call into question the particular reading of this passage that I've seen come up in this thread to support the idea that "you do not have a bonus action until given one." You yourself have used this argument several times in the post to which I'm responding. The case I'm making in this respect is that Shield Master is the game-feature that lets you use a bonus action (with conditions) according to this passage, so if you have the Shield Master feat, you do have a bonus action to use, and that taking the Attack action on your turn is merely concomitant with using your bonus action.

 

[Asgorath]

Personally, I don't have any problem with narrating the shield bash after the attack. The attack sets up the foe by unbalancing them enough for a skilled warrior to take advantage with a well-timed shield bash that sends the foe staggering or knocks them down. Ta-da!

This is a brilliant point. If you want to shove a target who is in a defensive stance ready for your attack, then it takes part or all of your full Attack action because it's harder to do. Once you've made your attack(s), the target could be off-balance enough that the extra juice from the Shield Master feat allows you to slip in a well-timed shove to knock them off their feet.

 

[Maxperson]

I think you perhaps misunderstand. We all know what Crawford said, we've all read the new Sage Advice and most of us have probably watched the videos on YouTube. The point is that some of us are of the opinion that Jeremy is wrong, that his new "ruling" goes beyond simply interpreting the rules of the game and is making up new and unnecessary restrictions, and is Bad Advice.

This is essentially the same as when I look at the rule in the PHB that says that after you reduce a creature to 0, you can retroactively decide that you were knocking it out. I think it's wrong and a bunch of malarky, so I created a house rule and changed it.

You don't like what he said, so create a house rule that ignores it and move on.