under-pitching will most certainly cause off-flavors. in the growth phase the yeast throw out all kinds of off-flavors. it's just that:
1) they will typically clean it up after fermentation is complete when there wasn't too many off-flavors thrown out there.
2) the calculators have a very high bar for what under-pitching is
3) they are also typically very skeptical when it comes to how quickly the packaged yeast cells start dying off.
go ahead and do the experiment yourself. create a 1.090 wort. pitch in one vial. report back with the results.
my take on this experiment - and I am a big fan of brulosophy and read basically all their articles I think - is that it has to do with multiple variables. There are two important lessons:
1. the phase space of parameters that can make good beer is NOT as small as many of us like to believe. - thats a great news. And we sort of all know it. Its not difficult to make a decent or even good to great beer, even if you screw up on a few items and don't quite nail it.
2. You can vary any single parameter (pitch rate, boiling time, mash time, fermentation temperature etc.) by some surprisingly substantial but also somewhat reasonable value, and, provided other parameters are controlled for, the beer will come out great. In fact it will be indistinguishable from the "optimal" parameter beer. Imagine a circle-shaped blob of "ideally tasting beer" in multi-parameter space of all brewing variables. You can tweak any single parameter, and as long as you start at the center of the blob, you can get away with it.
However, if you compound the errors, which means you start way off the center and then walk further away from the center in multiple directions, you will get off "ideal taste" real quick.
Basically, if you are under pitching, your yeast health is bad, your sanitation practices are terrible, your fermentation temperature is too high - esters (or too low - stalling), you can't seem to follow the recipe properly, you bottle too early, or too late, or your transfer technique suck and you still believe in secondary - the errors compound in rather nonlinear fashion and produce a terrible, or at the very least not-so-tasty beer.
so it's a bit like everything else in life. No single tiny event matters much, despite butterfly effect theory, but combine enough little tiny "faux-pas", none of which would matter individually, assuming everything else is perfect, but take a bunch of them together, and you have a disaster of epic proportions.
I also think if you plot distribution of errors among various brewers, they are not random, there is a great deal of correlations. This is 80/20 effect. 80% of beer is consumed by 20% of population (I think in reality it's more like 95% of beer is consumed by 5% of population). Similarly, 80% of brewing errors probably happen to 20% of us.
The brewer who doesn't control his fermentation temperature is also probably the same brewer who is not super-anal about sanitation, or making a starter, or following the recipe precisely. The brewer who is super-anal about fermentation and sanitation is also unlikely to under-pitch 1-year old vial without a starter into a russian imperial stout wort at 120F, or transfer their beer 3 times while splashing it around just for fun, and then not use enough priming sugar.
So you end up with a sort of binomial distribution. Some people get a lot of off-flavors and infections while they tweak a thing here and there and still can't get it right. So they form an opinion that brewing is super-hard.
While most brewers who pay even a little attention to sanitation and follow recipe and good practices, end up with decent to excellent beers, even if they mess a parameter here and there. Brewlosophy is an extreme example of that, showing that if you control ALL parameters and dial them in perfectly, you have a huge margin of error in any other single parameter. But it's probably not a good lesson to most brewers who may be off on a few parameters to begin with and basically need to get everything else perfect, or they are off the "good tasting beer" blob in this 20-dimensional phase space.