Understanding the limitations of a calculated based on a dataset is one thing, and saying your intuitive guidelines are more accurate is another. I think having a rule for pitching rates to get your desired flavor profiles, and using a yeast calculator would probably be the best course of action for accuracy and consistency.
I know that the calculations done by braukaiser have been verified by several homebrewers and professional yeast labs that I trust. Very few have found data sets that correspond to the growth rates that chris white used when making mr. Malty, and that he's never publicly published...
According to braukaiser, pitching into a 2L 1.040 starter will give 333-470 depending on the inoculation rate, hey look the average is 400
. If pitching from a vial, then it depends on the initial cell count and how old the vial is. Braukaisers numbers had a typical variance of 15%, and your guideline falls right into the middle of that. However without taking note of the age of your yeast, and therefore your inoculation rate, then it's going to have even greater margin of error. (I wrote enough error analysis for school, not going to carry it out through the calculation).
According to yeast, american sour beers, and several yeast lab comparies I've spoken to on the subject, typical sacc yeasts have a theoretical maximum cell density between 200 and 250 million cells per mL or 200-250 Billion per Liter. However that's with ideal propogation media. In a starter with dme and some yeast nutrient, we're more likely to get in the area of 160-200 Mil/mL depending on nutrients provided, plato of the starter, innoculation rate, and aeration.
250mL of bacteria has a much higher cell count than 250mL of yeast. According to several credible sources I've contacted, brett and lacto have a maximum cell density roughly 2.5x that of the brewers yeast. So while a 250mL starter of yeast might contain ~50 B cells, lacto/brett might contain 125B which is enough to sour/ferment 5G of wort.
As an aside, this is pretty much exactly what I do with my yeast bank. I propagate in 250mL starters, harvest a dense 50 mL slurry after decanting the supernatant, leaving me with 50mL of slurry containing ~45B cells. When needed, this gets propagated back up into a new starter 250mL if I'm just replenishing my yeast bank, or larger if I'm preparing for a batch of beer. Cultures get plated and re-propagated occasionally as needed.
I'm not aware of any academic studies performed on yeast growth rates using standard homebrewing process', the best I'm aware of is braukaisers data sets, and the verification of those data sets by other homebrewers.
Sometime soon, hopefully within the year, Omega Yeast Labs will be publishing their dataset which will provide strain specific growth rates, so you'll be able to get way more accurate models as the maximum cell density will depend on the strain. So you'll say 80B cells of say wlp 002 on a stir plate will produce ~xxx cells.
S. Cerevisiae was a strong proponent of pitching starters at krausen, as peak cell density has been achieved and it is at a time when the yeast have a large reserve of nutrients and are active. It's better to pitch them at that point, then to wait for another potentially ~30B cells and lose your advantage of nutrient reserves. This is a similar method to the 'vitality' method of starters, pitching them 6-8 hours later which usually coincides with krausen. He was also against the use of stir plates, that they caused shear stress to the yeast and provided worse aeration than a vigorous shake. My method is currently to aerate well by hand, then start the stir plate about 3 hours later. This gives a lot of aeration, and the stir plate continues to degas the media of dissolved c02, which is a leading factor for the difference in growth rates between non agitation and stir plates. Again, several yeast labs have commented on this as well, that stir plates purpose is not to keep yeast in suspension, but to remove c02 from solution.