It depends what you're going to focus on.
If you're going to give the characters 15 different ways each to chop up bad guys, you need a grid. Those 15 different ways each need an environment in which to live, where there are enough *things* to be interacted with in enough distinct ways that you can count up a large number of different options.
If you're going to make combat take up a significant amount of time, that time has to be spent *doing* things. And to do things, you need stuff to do those things with. A grid provides that. Additionally, a grid keeps those various situational combat tricks your characters may possess from turning into a "mother may I" scenario. For example, if your spell shoots lightning in a cone, a grid lets you know whether you can hit 2 goblins or 5. Without a grid, you have to ask a DM who will give an answer biased by whether he wants you to incinerate 2 goblins or 5. And as these things multiply over the course of a fight, a grid saves you the trouble of memorizing everything that happened before, so that DM judgment calls early in the fight are consistent with his judgment calls later in the fight.
On the other hand, if combat is going to be short and not involve a significant amount of variety, a grid is counter productive. You end up spending time creating things you'll never use, and you extend a combat session longer than it has a right to be for the amount of fun you're likely to get from it.
I kind of suspect that 5e is going to be the former type of game rather than the latter. Which means it will objectively require a grid for quality of game-play reasons.
But I could be wrong.
Let's say "incomplete".

I feel like you need to add the words "mechanically defined" throughout your theory:
If you're going to give the characters 15 different mechanically defined ways each to chop up bad guys, you need a grid. Those 15 different mechanically defined ways each need a mechanically defined environment in which to live, where there are enough *things* to be interacted with in enough distinct mechanically defined ways that you can count up a large number of different options.
Trust me you can have tons of variety without a grid, but its not
mechanically defined variety. The game I'm currently playing in doesn't really use mechanically grid-defined powers, and we have plenty of cinematic stuff. Just a few days ago my barbarian ran across a banquet table to jump into combat amidst the evil clerics at the other side. Without the grid, you have an infinite number of ways to chop up the bad guys.
I feel the opposite way about the way 5e is headed. I think it needs to "de-grid" to get back to more basic feel. Especially since combat takes so darn long in 3+e. I mean, I like combat to be a big part of D&D, just not
the same combat for the entire session.
Also, you can do plenty of that stuff in your middle paragraph without a grid. My 2e group did it for years with no grid, just using open maps and measurement in inches. I've recently played a few games that use a "zone" method. That gives you a lot of the mechanical advantages/options of a grid without the speed cost.
Using Zones "How many targets in that zone can I hit with my Lightning Bolt?" can be answered with a d4 (probably defined either in the Lightning Bolt spell description or as a general note about
Bolt spells. That's a lot quicker (believe it or not) than waiting for the Wizard's player to take a 5'step and carefully plot his spell squares to maximize the number of critters affected. (Honestly, most DMs I know, including myself, don't have their combat narrative so closely defined that hitting 2 or 5 goblins will matter. Its the pondering of geometry that seems to take so long.)
P.S. Y'know, as much as people seem to fear letting the DM decide things... I'm starting to think the DMG needs a big section on "How not to be a twit."