I just watched the video and to be honest, didn't quite understand why he said you couldn't just correct the room+speaker response at the listening position.
If it is the reference I am thinking about, the issue was that of a speaker that has different on-axis (direct) and off-axis response. The measurement mic "sees" the combination of the two. If you attempt to correct, i.e. EQ, that response, you would equally impacting both direct and indirect sounds since it is upstream of the speaker. In doing so you damage the usually best response of the speaker which is the flattest and smoothest.
You might say well, what we hear the sum so who cares if we screwed up the direct sound. The problem with that is that what we "hear" is not what a single microphone picks up. There are a lot of variations that a microphone shows that we don't hear due to resolution of the ear, and the differing sound with respect to what each ear hears. See this article I wrote a while back for the drill down:
http://www.madronadigital.com/Library/RoomReflections.html
If a speaker does have similar on and off-axis response, then you can indeed EQ it.
I should note however that empirical data somehow disputes this. In the test of automated EQ system, the directivity issues of the B&W 802N speaker were improved. Ultimately the real arbiter is the ear. Having an EQ that lets one turn this type of correction on and off is critical in determining if one should attempt to EQ that portion or not.
It its because you'll only get a good sound in a small sweet spot and possibly mess it up elsewhere? Or have I missed the point?
That is an issue below a few hundred hertz where the room dominates and by definition then, moving even a few inches can substantially change the frequency response. One can correct for that with room design, treatment, multiple subwoofers and optimization (see
http://www.madronadigital.com/Library/Computer Optimization of Acoustics.html). But if not done, then yes, you do have that problem.