I'll try to keep this short, but there are so many points to address...
I guess I'm obsessed with getting things right.
With certain standard caveats, and within reason, of course I believe races times can be predicted with mathematical formulae. I can't even think of a another way to predict race times, except to make them up.
It's true, I "simply don't understand the point that the VDOT tables overstate the importance of VO2max and understate the importance of economy". I don't think it really is a point, and in case you ever establish the point, I don't agree that it's true. I've already said that VDOT tables are economy neutral. You get a high VDOT score based on a performance, which could be due to a high VO2max, or a high economy, or both -- VDOT tables don't make a distinction.
For athletes who use the VDOT tables, there is no connection between VDOT and their own VO2max and economy, except that the product (quotient?) of VDOT and generic economy is the same as the product of their real VO2max and real economy. Presumably you refer to the test subjects. What went into the generic curve was the VO2, velocity, and economy measurements of tested subjects, presumably with both VO2max and economy developed to some level. Why wouldn't the generic curve represent the VO2max and economy of the tested subjects adequately? Why would one be overstated or understated?
Did you read my comparison of Purdy to VDOT? Surely you don't accuse Purdy of any VO2max or economy bias? Purdy is based on World Record velocities for all running events. Yet I found that for distances between 5K and the marathon, a maximum difference of about 1%.
The high VDOT values are not an extrapolation of generic curve. The generic economy curve can compute a VO2 value for any velocity, up to the speed of light (and faster). The generic curve does not need to be extrapolated to accommodate faster performances related to higher VDOTs.
An average economy does not represent an average level of fitness. Elite runners can have the highest level of fitness, yet have a worse economy than the generic one.
Let's try and look at a different way. Suppose my econo-Bekele with a VO2max of 70 runs a 12:37 5K. I can compute his real economy, and his real VO2max. And I can look up his VDOT, and calculate the corresponding generic economy. The very same factors that cause variance in VO2max, cause the same variance in economy. If I compare the ratio of my real economy to the generic economy, it will be the same ratio as my real VO2max to the VDOT index.
That's really the whole point. VO2max is a bad predictor of performance because of the wide variation of VO2max, and equally wide variance of economy across the population. The very same factors are behind both. If we eliminate the wide variance of economies, by just picking one, then we eliminate the variance of the "VO2max calculation". It's no longer real, but that doesn't matter. If you want you real values, you need to do a VO2max test. It's really that simple. We could have picked any economy. If you want, we could define Clayton's economy as the generic one, and produce the exact same table. The only difference would be that VDOT indices are replaced with different CDOT indices, which would be a trivial linear transformation (much like converting from Celsius to Fahrenheit). The remaining columns of times for distances would be roughly the same.
I have no doubt the VDOT tables give accurate comparisons of race performances at different distances, for equally apt athletes properly trained for that distance. As I said before, they correlate will with Purdy. VDOT isn't the best performance predictor (Purdy is!), but for distance events it corresponds very well to Purdy. You could say it's bad for sprints (which tap into other energy systems), but this is well known and to be expected from a "Performance Table for Distance Running".
You keep saying conceit, which seems strange to me. This is what scientists do -- come up with equations to model what they observe. I presume Daniel's choice to map velocities to VO2's was largely due to his background of physiological testing. He probably re-used or simplified existing equations and models he was familiar with. Of course there are many other ways to predict race performances, and indeed there are no shortage of race predictors built on all of these ways.