Excellent points, Paul. I hadn't really thought about defining separate Rx values for shorter and longer distances, but it's a great idea.
As you say, race time prediction is tough to dial in when trying to deal with the population in general. A siingle scheme that adequately covers all distances and all abilities simply doesn't exist. Add other variations such as terrain, course accuracy, and weather (as I was so rudely reminded this morning), and things get really hairy.
On the plus side, though, is the fact that we're not involved designing rockets, here. We just wanna know about how fast we might expect to go out in the next race, or see roughly how past races of different distances compare to one another. As programmers, a deviation of 30 seconds in the projection of a 10k may keep you and I awake at night. But it's not really that big a deal in the overall scheme of things. Most runners could plan a race strategy from either end of that 30 second spread and come out in good shape. It's blowing the *first mile* by 30 seconds that sinks the ship before it leaves the harbor.
So, those runners who are obessed with dialing in the numbers can study the models, choose the one best suited to themselves, learn what adjustments need to be made, and fine tune their stratgies to the n'th degree (I'm in that boat). Those who are not quite so obsessed, but still want to better understand their potential and improve their times can use any of the prediction models with good results. Smiles all around!