The damage is a bit weird. A big spider falling 10 feet took 56 damage but a Duergar flying 100 feet far and 60 feet down took like 14 and got up right away.
This is the best summary. Matriarch takes so much. I am wondering if they are using the "object size" rule for damage and they damage themselves more for being large or something
They seem to be, and I approve, by the by. As long as larger creatures are harder to push (magic cheats and should be fine) then the system works as intended and makes sense. Elephants and mice and all that jazz, but in general the base game does not handle big creatures falling very well at all.
On the other hand, please don't let us deal damage by abusing this to orbital drop dragon corpses or something-or make it so that you can at least save to avoid falling creatures/objects. I don't want to kill the final boss by dropping a statue on him. It would be too lame.
Having shove replace an attack in part of an attack action would add another layer of menus to go through because once you select attack, you'd have to decide whether or not it was a shove or a normal attack for each attack in the action. This seems to be more trouble than it's worth for accuracy to the tabletop. And just having shove an Action doesn't feel right because that's not really how it functions on tabletop. Having it a bonus action is a decent compromise.
As to the distance, the BG3 shove does not truly knock people prone (there is an animation where they fall down, but the target immediately stands right back up, no prone condition applied.) If the tabletop rules were followed we're now looking at three layers of menus: Choose Attack - Shove or Attack? - Prone or 5ft? Again, doable, but irritating to do every time. Losing the functionality of the knock prone in exchange for what appears to be either a strength-based distance or else just a flat 10 ft distance....again fair compromise. (I have to go check this later...I'm suddenly not actually sure the shove distance is even 10 ft.)
We're also losing Grapple as an option....which again, the 5e variation of Grapple works pretty well, especially in combination with Shove (Shove Prone + Grapple is basically an armbar, they're now prone and have no speed with which to stand up), but in the computer game, I have a feeling it would be a bit irritating to deal with regularly.
So, losing the ability to shove multiple times in a round (one for each Attack), the ability to knock prone, and the ability to grapple, in exchange for shove 10ft as a bonus action feels fine. And it's not like this is going to make Pushing Attack, Repelling Blast, or Open Hand Monk less powerful....those powers allow you to do damage in addition to adding the shove...and Repelling Blast lets you do it at range.
So, I'm good with their alterations on shove. It's not quite the way my shove-happy table-top group does things, but we already have proven ourselves able to find cliffs, stairwells, bridges, and lava flows to shove people off of or into just fine. The compromise here is close to that.
Disagree on every point in the general and the specific.
First, menus are fine. I don't understand why people dislike them. If Larian isn't willing to go through the work of implementing them, they shoulden't have taken the project. DnD 5e has those options on purpose.
Second, Bonus action shove is the least troubling of the changes, but the distance is too long. I could actually accept the compromise if it was shorter, but don't see any need for it fundamentally.
Third, you have to choose between shove prone and shove back in 5e, which should be easy to implement. Heck, certain weapons can already knock prone.
Fourth, shoving by enemies is always worse than the player because next to no enemies have skills, so it's never been a truly effective tactic. It can work, but in situations where it is the optimal action the player is usually a wizard or something, so it makes sense to try it.
Fifth, we should be able to grapple.
So in summary, the compromises here shoulden't exist, they are implemented poorly, and there is no reason a stricter implementation is impossible.
