He's both right and wrong--or right only to a point, and very wrong after. Every table does have its own context, and the game needs to account for that. Ironclad, absolute rules with no variation would not function. But he is wrong to make so much of the resulting so-called "variety."
Thing is...we're all humans, we share a heck of a lot of common features. This forum is for English-speakers, and for most D&D fans, English is their mother tongue. (There are exceptions, even on this forum, I know, but D&D is "natively" English-language.) The vast majority of people who play D&D share the overall Western cultural zeitgeist. Even with the growing cultural exposure of D&D, the vast majority of players are "nerds," usually at least a little bit math-inclined.
So, we all start from the same text--the rules sitting before us--and we (well, most of us) have both overall human thinking processes AND a shared bedrock of cultural and social experience. That's gonna mean we often draw the same conclusions, for the same reasons. We will often recognize that a particular design idea is just generally clever, because good design can transcend tables. And we recognize that there are a ton of mechanical elements that will be nearly or truly universal, e.g. you aren't really playing "D&D 5e" anymore if you aren't rolling d20s for most of your "did it work"/"did you avoid it" rolls.
This might seem trivial, but it's utterly vital. It tells us where actually effective design must lie: extensible frameworks, not individual rules.
Because IF D&D were just a collection of ironclad individual rules, Colville would be 100% correct. But it isn't--or, at least, it shouldn't be. And, contra his claims, it IS possible to some extent to account for a wider degree of variation on things. Yes, if you go for the absolute extremes (e.g. only a single PC vs 8+ PCs) you're going to run into issues...because extreme cases are ALWAYS issues when statistics are involved. But 4e's XP Budget rules actually work extremely well to gauge the difficulty of encounters and challenges even if you don't have all four "roles" present. More importantly, having those roles in the first place helps you identify what might go wrong so it becomes easier to adjust if you need to: in other words, you can make design that is better for supporting DM tweaking.
Sooooo....yeah. We can still talk about it. But we have to let go of both the fiction that (as he says) D&D can be "solved," AND the fiction that every table is a unique and special snowflake completely unlike any other table ever. Doubly so when SO. DAMN. MANY. DMs adamantly insist on EXCLUSIVELY playing "Tolkien races only, no firearms but 15th century plate armor and 16th century swords" etc. etc. There's a hell of a lot more in common than different between most tables.
Edit:
Oh God, then he starts talking about modifying monster stats on the fly during combat. No, nope, sorry, uh-uh. I already knew that he actively engages in cheating his players (to the point that he will even stage rolling dice while fixing the result, so people will think he actually rolled something he didn't), but modifying encounters on the fly to fit your preconceived notions of what they "should" be? Noooooooope. That's a flag so red we need to invent new color words to describe it.