I didn't vote, as there are just too many variables to consider. As protection circuits are applied and used now, it would seem that the manufacturers recommendation would probably be the best, as different manufacturers likely have different requirements, due to slight differences in the cell's makeup. What value it would be therefore, would depend on each manufacturers individual cells.
Why don't companies select a better quality more on spec IC that properly cuts off somewhere around 2.75v?
As to the QC control you've observed on many protection circuits, I agree, this is a problem. Every once in a while they don't even work at all! It doesn't seem that until the tolerances and quality of the components used in these circuits is improved, that it really makes sense to pursue this endeavor. :sigh:
If you ran your cell down to the low voltage cutoff every time, you would be doing damage to it by brining it down to 2.3-2.5v (where most protection circuits seem to kick in).
The protection circuits added to loose LiCo cells by distributors (note that Li-Ion cell manufacturers refuse to provide loose cells with protection circuits) are not a convenience item to remind the user that the cell needs to be charged, but rather a last ditch safety effort to prevent an over discharged cell from being charged and possibly "venting with flame". It is there to protect the user, not the cell.
As for discharging cells to the cutoff point of the protection circuit, you're screwing up your cells to some degree, anytime you discharge them below about 3 volts under load anyway, depending on the current value. So, discharging cells until the protection circuit cuts out, is just a really bad habit to develop.
Now, if what you are suggesting is a protection circuit that is more to protect the cell, and not the user, then I think I'd go for something in the neighborhood of 3.0 Volts. The problem is that this value would actually vary greatly, being dependent on the current load under which the cell is used. For low current uses for example, the value would be necessarily higher, and for high current drain applications it would have to be set lower.
This is where the problem with your idea is. There is really no way for the distributor to know how the cell is going to be used. This is already a problem with current protection circuits, as they are now. Short of providing protection circuits with different cutoffs for various current loads, and, as a result having to provide many versions of basically, the same cell, it can't really be done. I don't foresee that happening.
The best solution is to provide protection circuits within the device, as Laptops, power tools, and some lights do. Many of my lights have this feature, and allow the use of unprotected cells, which solves the problem entirely. :)
Dave
