Objective scores are useful. But what would also be useful would be like a metacritic for speakers. Simple probabilistic reasoning dictates that if a large percentage of people in a sample like something, then you are likely to like it too. Whereas it's not clear what exactly the Harmon score is predicting or why.
Here's a small example of the Harmon scoring problem.
Revel F328Be scores 6.5 - with sub 7.7
Revel M105 scores 5.9 - with sub 8.2
I have both of these speakers. I foolishly bought the M105 based on it's high score realizing it was smaller and would be less dynamic but the score led me to believe it would be fantastic. I A/B'd the M105 with the F328Be and it's a joke. The F328Be beats the M105 by a MILE. Using a sub with the M105 cannot make it sound anywhere near as good as the F328. The M105 tweeter is weak compared to the F328Be. It's a totally different animal. While the F328Be is solid down to 29Hz and can fill a 24'x28' room while barely on - the M105 easily begins to distort with bass under 70Hz. No contest between these two models. Yet, the score can easily be read to show a different story.
Since that experience I place very little weight on the Harmon score when the speaker is not the same size with similar drivers. The composition of the drivers can make a big difference in the sound too. Yet, I don't see any part of the score that considers it. For all practical purposes, I can't rely on the Harmon score when making a purchase. I might look at the data - but I know comparing the scores can be a HUGE mistake even among speakers from the same brand.
I listen to Dr. Sean Olive explanation of how the Harmon score should be used and it didn't bring out all the issues I experienced. Dr. Sean Olive indicates speakers with a difference in score of less than 1 point can be similar. I did not find that to be true. The score is broken for my use when there is no component for displacement, size of the speaker or driver materials. It's way too limited for me to use as a major factor in speaker purchases.
Last edited: