Which is the whole point of his analysis. Any one attack roll is likely-but-uncertain, adding some excitement to the roll. But the outcome of the combat as a whole is rarely dependent upon any one attack roll but the sum of many, giving you a pretty fair idea about how things will go.
It isn't really "exciting" until the end approaches,
then combat becomes exciting
if the outcome is close enough to be in doubt. Until that point it is just a back-and-forth exchange until the averages play out and the most likely outcome (90+% as victory) is achieved. This is why many players complain about the slog of it.
Frankly, I don't think an attrition model (like combat) would work for social challenges.
Where D&D breaks down is where too much weight is put on a single pass/fail roll.
But it never does, does it?
When a single roll is called for, the situation doesn't have too much weight. Where multiple rolls are called for, it does. Combat has a lot of weight (ultimately possible death and the end of the adventure for a PC), as where social rolls don't carry the same weight. If one option fails, others can often be attempted because "failure" in an ability check doesn't necessarily mean the "end".
Consider the other d20 roll--the saving throw. When the stakes are higher, a series of rolls is used (often at least two, and many times three or more, even each round for a minute until the save succeeds or the effect ends).
Now, consider an exploration challenge, something like making a Strength (Athletics) check to climb a cliff face. There is a very poor structure to resolve this except (by design) that it is "up to the DM". Do you make a single check? A series of checks? After all, depending on the height of the fall, you might be killed, right? If you don't make the DC, do you fall or just fail to make process? When do you fall? And so forth...
Social challenges carry even less weight and have a strange consequence that good role-playing should override poor rolls, possibly even making rolls unnecessary. They currently (as we know) have just as poor a structure, relying almost entirely on DM fiat.
So, how to implement them, and if it is even worth it to create a system for it, when the weight is often insignificant, would be equivalent to having 15th level PCs fighting each encounter with CR 1 creatures they encounter. Most games don't have time to expend bothering with challenges without significant weight and therefore gloss over such things.
The result is social challenges are resolved in a single roll many times. You could certainly do a series of checks might be made along the way as a scenario unfolds, with each individual check leading towards success or failure, until the path is resolved and the scenario ends, but this depends vastly on the scenario and what interest the players have in rolling during it instead of role-playing during it.