And the resources to create this system would be more significant than you might think, as they would need to code for to all the unique spell effects, attack types, class abilities, monster abilities, etc. They would need to decide how each variable translates into the equation in different circumstances, and then test the application of this new system across all the many ways that it would need to be applied. All of that energy could instead be used to create more content for the game, improve the story, add popular class options from DnD books outside the PHB, etc.
That may be partially true, but I would expect that Larian already know exactly what are average effects of any "unique" effect, since that is important for balancing reasons. Also, only effects that directly rely on the outcome of dice roll are an issue here ( like GWF "if you roll 1 or 2" ), and those are in minority. All effect that add or reduce some amount of damage but do not directly rely on specific roll result would remain exactly the same ( like GWM adding +10 damage , or weapon enchants etc ).
So I still believe that required change would be comparatively small, and that "all that energy" would be well spent in making game approachable to people who are not DnD veterans - and I believe than gamers without heavy DnD background are majority among gamers worldwide. Granted, there will be even some of those who do not care about randomness - after all, XCOM has it also. But then, there is a reason why "refresh random on load" savescrum mod on XCOM was one of most popular - gamers generally get annoyed by misses.
There is probably some philosophical reason behind it : people tend not to notice when they get lucky and hit mob with max damage, but they sure as hell notice when they miss. In other words, minor good luck is not noticed, while minor bad luck result in annoyances. I'm aware that people used to DnD are also used ( or at least resolved ) to that - but , as noted, most potential gamers are not DnD veterans.
