Can you make the reverse case? If you can somewhat accurately perceive a speaker's character, what would be the advantage to accept speakers known to be less accurate?
Then it's down to what you like better. If I am listening to two different speakers and one has a very slight smile response, that I'm accurately-enough-perceiving, if I like that sound better, the advantage goes to that speaker. Of course there's a separate issue of "well buy the neutral speaker and eq it like the other one" but that's not exactly the point at hand. We can talk about a speaker that is not quite accurate due to design decisions, which are not directly replicable merely with EQ as well. (E.g. Omnis, etc).
Or if you cannot what is the advantage to accept speakers known to be less accurate?
If you can't accurately perceive the difference between the accurate and less accurate speaker, then one could decide it just comes down to the aesthetics, or any number of non-sonic factors (including that you may appreciate the engineering involved in making one speaker over another). And/or you just go on whatever produces the most pleasing sonic impression in the sighted conditions.
This is of course strictly looking at the options you gave me.
My position is that, in a practical sense, in many cases, it's justifiable to go with the idea one is really able to detect some relevant sonic differences between some loudspeakers, in sighted conditions. This does not ignore the issue of sighted bias; it takes it as a possible variable, and so it's a tentative conclusion open to being wrong. We employ such practical conclusions all day long, in the face of not having scientific levels of confidence for everything we do.
I mean, one really does have to answer the question: Was Harman Kardon engaged in irrational activity in all their blind testing and measuring? If the measurements weren't going to predict, or track with impressions in sighted conditions to any relevant degree, what possible motive could they have had in depending on all that blind testing to design speakers meant for sighted listening conditions?
I cannot see a plus in either case. I suppose it is something like DACs. Some cheap ones have distortion and noise at -120 db where it cannot matter. They probably sound no different than one where noise is -100 db and distortion -80 db. But if you can build one for cheap why not?
Agreed. One reason I have a Benchmark preamp when I could have bought something cheaper that would be equally transparent to my ear, is not just the features I needed, but I enjoy the idea of how well engineered it is, how well it measures. I think others on this site can get similar pleasure out of gear they know measures well, even if it is "more than they need."
However, as I've pointed out: the discussion of loudspeakers does not exactly mirror those concerns. It's very clear that both in terms of the speaker reviews, and the discussions of speaker measurements, including those criteria arrived at from blinded studies, that people take those to be somewhat predictive of what they will actually perceive when they purchase the speaker for use at home. Pointing out some awful resonance or peak isn't just greeted with some academic detachment, but it's presumed this will often lead to "bad sound" in the actual use case, sighted listening. As I said, this site presumes that the measurements and blind listening studies are relevant to recommendations for speakers in sighted listening. That really should be explicable, and I don't see how it's explicable if not for the assumption that sighted listening, while clearly susceptible to bias effects, will be reliable enough to so that we'll perceive the character of a speaker, as described in the measurements. Just like our sight is buffeted by bias effects, but nonetheless manages to be useful.
Had you given me the scenario about mitigating sighted bias, without the option of blinding, my first thought would be to have two sets of speakers one could switch between quickly. The first fly in the ointment would be they are in different positions. Harman's switcher takes care of that. Yet in that case bias from sight and reputation is quite strong. That was surprising to me when I first read about it. I think that is why like Kal and others who have done a session at Harman describe how enlightening/informative the experience is.
My next thought would be use pink noise to compare. That makes FR differences stand out more starkly than music. One might find it hard to choose which is better with pink noise, but it is easier to hear when one is different. Then moderate speed sweeps can find resonances which we know is a negative. I'd feel pretty good about that working, but I would have felt good about a quick switching comparison in the above paragraph. Results seem to indicate that confidence was misplaced. Next I might use that chink-chink-chink shaker sound that pans back and forth to elicit directional issues. If it doesn't pan smoothly you can hear it more easily than with music. I think some special signals like this are needed for sighted evaluation. In light of what is known one would be foolish to put confidence in them if they weren't corroborated by blind testing. Suppose we could come up with a handful of artificial signals known to work with decent accuracy sighted. Would magazine reviewers use that? It won't read nearly as well as exhilarated wonders described by current reviewers. Where veils are lifted, new levels of detail unraveled and wonders of space unfurled.
All interesting ideas for trying to raise some confidence levels in sighted listening.
I would disagree with the idea that is it foolish, or at least always foolish, to conclude anything about the sound of loudspeakers outside of rigorous scientific test protocols.
It would be foolish if you are looking for a scientific level of confidence not to control for sighted bias. But that doesn't mean it would be foolish in informal, everyday, practical settings to come to some conclusions. I've had my sighted impressions confirmed by measurements often enough (e.g. hearing a speaker, forming an impression of it's frequency response or other characteristics, and later seeing measurements at Stereophile or Soundstage etc, which match those) to have some level of confidence. I'm far from infallible of course, but I think a practical application of sighted listening can be justified until one has more rigorous data. (If you watch Erin's Audio corner, you'll see that he is very good at identifying speaker characteristics, frequency response issues etc, which show up in the measurements after he listens, which suggests that sighted listening is not across the board unreliable).