Well then, I guess you'll also be shocked to hear that speakers that measure well, corellate well with listener preference in properly conducted DBT's.

Again, Floyd Toole has a test rig which rotates the speaker in space - they take 72 anechoic measurements in the vertical and horizontal plane and then they know how the speaker radiates sound. They then put that into a computer model of the room and have a damn good idea of what the repsonse is like at the listening chair.

In blind tests people prefer smooth, flat frequency response.

Putting that another way, the meaure the speaker and know whether people will like it or not - in other words you could be deaf and design a speaker that people will like. Moreover, our deaf speaker designer won't end up building a hearing aid for himself. (see below).

Why should he "use his ears"? Are his ears (or yours) a representative sample of the target population, or is he (or you) merely building a prosthesis for his own hearing defficieny?

Again, coming back to the work by Toole, in his double blind testing, people with frequency related hearing loss had the most polarised opinions on what speakers sounded good and bad, and that these unfortunate folks would tend to choose speakers which mitigated their loss.

What does your audiogram look like?

 

Maybe you just aren't taking enough measurements...