I’ve bolded above uses of the word before with which I disagree. Conditional statements (if X, then Y) are not statements of causality, so no specific temporal order between X and Y is required or even implied. Jeremy Crawford admitted as much when he tweeted, “My rulings, and the logic they rely on, are entirely within the context of D&D’s rules.” Needless to say, I don’t find this sort of circular reasoning to be very convincing. It isn’t necessary for “you take the Attack action” to be true before “you can use a bonus action to shove a creature...” It just needs to be true on your turn.
You're right about "if X, Y" not implying causality, but wrong about timing. There's a very strong timing involved. But, first, we have to show why the statement isn't just "if X, Y" in the case of Shield Master, but "If and only if X, Y."
This is simple. You do not have a bonus action until some ability gives you one. In this case, when discussing the bonus action associated with Shield Master, there is only one way to get it, take the Attack action on your turn. If you do not do this, you do not get the bonus action to shove. Therefore, the correct structure, formally, is "if and only if", or IFF. So, the structure is, "IFF you take the attack action on your turn, you..." get the shove.
IFF statement require that X be true for Y to be true. So, timing-wise, if X is not true at a given moment, then Y is not true. Timing does exist. X may or may not cause Y, so no implied causality, but Y only exists when X does.
To move a bit forward, you've already acknowledged that there are points within your turn where it can be considered "before your action" and "after your action." You did this when you offered the Monk Flurry of Blow as an example of a bonus action with timing. When means, you acknowledge there is a point in your turn where you may not have taken your action. Since actions are discrete things (there is no declaration of actions step, there's no smearing of actions in the rules, the rules treat actions discretely, etc.), then there must be possible a state on your turn where "you take the Attack action on your turn" is not true. Therefore, you must first take the Attack action before you get the bonus shove.
And, yes, I understand you have a preference to treat actions as smeared and fungible in time, but that doesn't jive with your own offering of flurry of blows. If declaring an attack action is suitable for Shield Master because declaring is the same as taking, then so it must be for Flurry of Blows, only you're left having to explain where "after" your Attack action really occurs. Further to this, the rules clearly state that you may split your move up how you wish before your action and after your action, again showing that there is a before and after state available. Finally, "If you take the Attack action on your turn..." may initially appear English imprecise enough to read as "If you take the Attack action
at any time during your turn..." this is counter-indicated by the complete lack of acknowledgement of this very imprecise interpretation. That you can choose to interpret it this way is you adding into the rules the "at any time" reading. You're welcome to it (I just remove the timing from Shield Master rather than redefine the way the rules are intended to work), but it's not RAW. The "at any time" conflicts with other timings in the game. Frex, if a creature has a Readied action to attack back if you attack, when does this resolve? Accordingly, if you declare the Attack action such that it's sufficient for the Shield Master feat bonus action trigger, then is that sufficient for the Readied action trigger? You've declared you will attack on your turn, you've earned the benefits of your feat's conditional bonus action, do you suffer the consequences of the Readied action?
