I think that’s the best advice yet, i.e., not to worry about it. I have a link below to another thread with conflicting info between yeast calculators, and the yeast lab itself over cell viability. At best, it’s mainly guesswork determining the initial viable cell count, which makes the starter size question, also guesswork. I seem to recall Denny not worrying too much about this either.
It seems odd that a straight forward hefe recipe would produce substandard results for you. What do you attribute that to? Just for the record, I note that Gordon took an NHC gold for his hefe, which was a straight forward recipe. Of course, he employed a traditional decoction. I’m not saying that decoction makes a better hefe, just that you can’t blame Jamil’s recipe, if that’s what you imply. 
I do cell counts for every beer and I can tell you that the calculators are right within a range, but sometimes you’ll end up with 100E9 cells total in a 2 liter starter.
You can use visual queues to tell you when you are getting starters with a low count—they’re the 1 starter out of 100 that leave you questioning whether the yeast is just dead.
If you want to pitch at the rate given by the calculator then mrmalty.com is usually right there give or take 30E9 cells. I actually use a 1.020 starter wort with a shaker at 120RPM, so any variability I get is probably from that difference.
Are you talking specifically about weizens here? I get the impression from your book that you mostly do 1L starters and repitch.
How are you determining the viability of the cells?
I don’t. I’d like to order methylene violet one of these days…
Excuse: Kai’s reply in this thread about why he doesn’t stain http://www.homebrewersassociation.org/forum/index.php?topic=5484.msg66305#msg66305
Yes, when I make starters I normally do 1L. I repitch most yeast, but I haven’t gotten good results from the 3068 so I normally use fresh. I didn’t know why until I saw Jess’ response. That’s part of the black box aspect of brewing; you don’t have to understand why something happens as long as you can control the inputs to get the outputs you want. But it’s always nice when the reason is exposed so you are less likely to draw wrong conclusions if you try to go beyond what you’ve learned is true. So I’m less likely to try to get fresher and fresher 3068 yeast, thinking that was the cause and will be more likely to control the amount of yeast repitched in a weizen. Something to explore in future batches.
Kai makes a very good point. If you want cell counts from fresh cultures, then viability is going to be so high that you can just assume 90-100%. Unless something goes horribly wrong, of course, but don’t boil your yeast and you probably only need to worry about viability on stored yeast.
Does anyone actually see growth from a smack pack in a 1 liter starter? I wouldn’t expect much growth at that pitching rate. Maybe half a doubling of the cell number, if you use a stir plate or starter, but probably little-to-none with a static starter. So does that mean intentional underpitching?
I think Neva Parker said a 1L starter will give you something like 20% growth and a 2L starter might give you up to 100%. And, yes, I use a stir plate. Why is a large growth rate an objective? It’s not like I’m building up yeast from a small culture. White Labs and Wyeast advertise that they supply sufficient yeast in their packages for a normal batch of average strength beer. I’m just interested in getting the yeast ready to go so as to minimize lag time. If I’m making a bigger beer, I’ll use more yeast or repitch from a previous batch. I guess I’d turn the question around. If you’re making a larger starter, does that mean intentional overpitching?
Not that I really fret over a few 10s of billions of cells here or there, but I usually try to follow George Fix’s recommendation of 0.75e6/mL/degree Plato. Depending on freshness and fill level, a pack would usually end up being about half Fix’s recommendation for what most brewers would consider a typical beer. For a 12-13 plato beer, a 2 liter starter should be just about right, assuming 100e9 cells in a pack, about 90% viability in a reasonably fresh pack and a single doubling of the yeast population on a stir plate.
For an older pack, a 1 liter starter might bring you back up to the original cell number in the pack, or, if you get lucky, you might hit Fix’s recommendation if you got a really fresh pack filled on the high side, starting with 130e9 cells, have good air transfer into the flask (loose cover, large headspace), and got 1.5-fold increase in cell number.
What I meant to be asking, though, is if you intentionally underpitched specifically for Weizens.
No, I don’t. A normal pitch has always worked fine for me. But this thread has taught me to be cautious of overpitching, whether by repitching from a previous batch or by getting yeast from a brewpub (both things I do often).
I did not mean to give the impression that the beer was substandard. Rather, it, in my opinion, lacked some of the complexity that I have gotten from other recipes. Jamil has some great recipes. I make them all the time with wonderful results. But, for my pallet, that one wasn’t one I wanted to make again. Of course, that was several years ago. So maybe I should give it another go. With a simple 50/50 hefeweizen recipe, I wouldn’t be surprised if decoction added a nice flavor component, though I have found that I can’t tell much of a difference between decocted beers and single infusion mashes (a bopils I made notwithstanding).
On that recipe though, it is more of a matter of personal taste. The beer turned out fine. I just didn’t love it like I have others I have made. I’m also willing to concede that Gordon and Jamil are undoubtedly better brewers than I am and can surely accomplish more with a 50/50 hefeweizen recipe than I am able to. That aside, I have brewed a lot of them in the last ten or so years and for the styles I brew, probably about 80% of them are German. But your point is well taken. Results depend more on the brewer than the recipe.