An even bigger problem here is that guys that might have a small differential are exactly the type of guys not to be running 400s/800s. Guys like Sage, Yuki, Tadese, or any typical endurance oriented non-pro rarely go and run 800s. There is no doubt we lack data on the running side.
Most of what you wrote above I'm in strong agreement with. When we talk about the women running 52s...the prerequiste for 52 is some pretty serious FRC (aka anaerobic capacity). You don't run 52 without a decent FRC, and those with a decent FRC are the type we would expect to see have a significant drop off. The power drop between a 1' and 2' effort can easily be on the order of 25-50% for an athlete like this (doesn't translate perfectly into running speed due to non-linearity of wind resistance and other factors). The main point is that when it comes to women running 52 or 53, I agree with you 100%. You might see something marginally less with a woman who can run 56, and possibly noticeably better if you find a women that struggles to run low 60s, but can still manage say a 69' half marathon.
I promise you, from quite an extensive time looking and working with power data and power duration curves, with cycling and now with running data backing up that running power operates similarly to cycling power, that this is the relationship. Athletes have an anaerobic capacity that can be measured in kJ and that converted to watts. If you take a runner with an FRC of 24kJ, then over say, 2' that athlete has 200w he can produce anaerobically. If his aerobic threshold is say, 300w, then he can do 500w for 2'. Over a 5' effort that athlete will always get his full aerobic contribution of 300w, plus his anaerobic contribution, which in this case would now be 24kJ spread out over 5' or 80w. Thus his maximum power over 5' is 380w. For a 1' effort you might see something like 650w being possible (at duration less than about 2' full aerobic utilization isn't possible, which does complicate things). So what you end up with is an athlete that can do
1000w for 30"
670w for 1'
500w for 2'
433w for 3'
380w for 5'
326w for 15'
313w for 30'
306w for 1hr
Someone who is a true endurance monster might have an FRC that is more like 12kJ, yielding with the same aerobic ability (worth nothing that this usually will be higher in an endurance oriented runner than for the FRC guy, this being a major reason guys like Symmonds don't run 27:30 on the track for 10,000)
610w for 30"
470w for 1'
400w for 2'
366w for 3'
340w for 5'
312w for 15'
306w for 30'
303w for 1hr
The main thing to note is the significantly lessened decrease between events. Even bumping that athlete up to 33% better aerobically still only gives
680w for 30"
560w for 1'
500w for 2'
466w for 3'
440w for 5'
412w for 15'
406w for 30'
403w for 1 hr
This shows why a guy like Symmonds, who would likely struggle with breaking 30 on the track, can be in the neighborhood of Bekele at 1500, blow his doors off at 800, and then at 400 it's just not even close. 16.6%
Someone like a Sage though, who with a 57s 400 really doesn't have a huge anaerobic capacity, might be closer to 10-20% in power drop from a 1' to a 2' effort. Interestingly enough, that 12kJ FRC suggests something on the order of a 10-20% drop depending on whether you're aerobic ability to "good" (something like 75' for half being around 20%), and as low as 10% for someone who could run perhaps a 59' half. If you take Sage and assume around 12kJ, you find that his 1' to 2' drop would likely be around 12-15%. What's the expected power difference between that needed to run 57s 400 (approx 15.8mph) and 2:00 800 (approx 15mph)....16%, which is very near the ballpark of the times Sage actually ran.
Equally interesting with Sage, is that if you take his best bike climbs of Magnolia (approx 30' climb) and NCAR approx 5' climb; you get that he likely did NCAR at around 5.04 w/kg and Mags at 4.29 w/kg. The difference between these power outputs? Approximately 17%. Well for a 300w threshold guy on the bike, which is where Sage likely is given those times, and using that 12kJ estimate for a more endurance oriented guy...what difference do you get between 30' and 5'? About 12%. Seems reasonable given that Sage said he was tired when he hit Mags.
Yea, I'm being quite technical in the explanation, but I've got time to kill and it's a good discussion. The bottom lines here are:
-This relationship holds true beyond all shadow of a doubt for cycling
-It seems to hold equally true so far on running, but admittedly there is less power data out there. Given the strength of the trend, I expect it's as true for running as it is for cycling, especially given how easily you can curve fit for runners using the exact same principle of FRC/AWC
-Running IS more complicated to bring these concepts to, as power to speed translation isn't set like it is on the bike. A cyclist that starts pedalling 20% harder up a steep hill will go 20% faster. A runner who puts out 20% more power, even at lower speeds where wind isn't a factor, might not go 20% faster if efficiency decreases.