I understand Empirical’s idea of hacking the long intervals for maximum stroke volume adaptation. It’s great to see someone thinking hard about this stuff.
My point was that it’s still pretty simple. If you just do the work (no matter which form) you’ll grab most of the benefits. If you want to optimize with hard start high cadence, then great. If you just love 40/20s or 30/15s and they provide some specificity then great.
It sounds like most experts agree that accumulating time at a certain intensity is key. So one can play with (1) various interval lengths, (2) work/rest ratios (3) intensity gradients (e.g. hard start vs no hard start) in order to maximize time at that intensity and our own (relative) well-being.
What seems less clear is how we should define the threshold for that minimum intensity to optimize for. For example, we could optimize for accumulated time over 85%+ of MaxHR, or over 90%+ of MaxHR, or over a certain power threshold (e.g., max aerobic power multiplied by x or FTP multiplied by y). Alternatively, we could use HRR (or a different formula) to approximate x%+ of VO2 max (depending on the formula, the 90% of VO max for example might imply 92%+ or even 95%+ of MaxHR).
Optimizing for time above 85% MaxHR vs. 90%+ MaxHR vs. 92%+ would result in very different protocols and RPE.
For example, 90%+ would probably require starting hard and doing medium-length intervals, but optimizing for 85% could be achieved more easily without starting hard and by doing really long intervals. I recently compared 3x6 with a hard start and 3-minute rest vs. 3x8 with a very mild hard start and 2-minute rest for myself personally. For me, the former seems better for time at 90%+, and the latter is better for 85%+. Both were… well, invigorating! 
(I’m setting specific racing goals and mental training aside here, approaching this strictly from the perspective of “maximum gain in VO2 max with minimal required suffering.”)
Why, and you cant say seiler.
For myself, 91% MaxHR is riding at threshold. One reason to toss out % HR and just go do max repeatable work.
Go out and pace vo2 intervals at the edge of blowing up. Like a 20-min TT but shorter.
It’s pretty simple. You can’t determine ‘optimum’ so ride at the edge. And total time for someone like myself, say 12-16 minutes total (4x4-min or 4x5-min) and progress out to 20+ minutes. Take that out to 25-30 minutes total if you are very well trained and need more stimulus. Round numbers for those targets.
And for myself, I can’t achieve 90% of HRmax for much length of time at all even while doing 5x5 intervals. My legs always give out first.
Max repeatable as @WindWarrior suggested is good. I’d say that max repeatable with one more interval left in the tank is probably better.
To make it even easier, the studies suggest accumulating minutes in the severe intensity zone - above FTP. Get up there and accumulate minutes. Don’t try to kill yourself.
Thank you for the cliff notes!
Did they elaborate on what the “severe domain” is? Edit: so it’s just above FTP? That seems…odd. One can obviously do a lot more 105% than 115%.
Severe is part of the 3 zone model used in many ex. phys. studies. Severe is simply above FTP.
I don’t know if there are ‘more is better’ studies like if 20 minutes at 115% is better than 30 minutes at 105%.
This.
I’m not suggesting one use ERG for intervals, but if it’s how one likes to perform them, consider trying a workout that hits both higher cadence and higher torque. Dial in your power, then do the first minute @ 110 RPM, dropping by 5 RPM each minute, so 110, 105, 100, 95, 90. 90 is not “low” cadence, but it is certainly “lower”, and will give you a clear increase in torque and wear on your legs more than the higher cadence. After a few weeks try 10 RPM drops, so 110, 100, 90, 80, 70. 80 and 70 will definitely give you a nice surprise when it comes to crushing torque. Don’t worry, your heart rate will be quite high.