Agree, it would be nice to be able to evaluate keymaps using some kind of objective measure or score.
But when it comes to measuring the overall agility of a chord I feel like just measuring the speed from a "home" chord ignores a lot of the realities of typing.
When typing a word (say 'think') what we are really doing is transitioning from one letter to the next. An easy way to think of this as a series of bigrams. So for 'think' the transition bigrams are: 'th', 'hi', 'in', 'nk', The speed of a particular chord then is the average of how long it takes to move from the previous letter times the relative frequency of that transition for all bigrams a letter is involved.
At first glance this might seem too complicated to compute all bigram combos but because certain bigrams occur much more frequently, we can effectively ignore most bigrams.
Also, sometimes a config can be fast but error prone. In TMX having to double tap a key is "fast" but often causes spurious key presses. I'm sure there are probably other issues like this in configs in general.
Finally, this only helps with measuring per chord speed. Many configs I've seen here really are focused on exploring an "idea" (like walking, or MCCs or no thumb, etc). The config's big idea is what is optimized and that idea may be more important to those users than pure WPM speed.