That's why I find it rather interesting that most of Marshall's results don't achieve significance at the .05 level.
With how hard it is to make equal batches sans the purposeful change in one variable, one would assume that all the slight differences would cumulatively result in easily being able to pick the different beer in a triangle test.
Unless they cancel each other out. But I tend toward how you're interpreting it, such differences would seem to increase the likelihood of detection of....something.
Marshall has repeatedly shown that many of those minor changes (pH, temp, fermentation temp, water chemistry, etc.) just don't matter that much.
If one choses not to believe his results, great.....but unless they are brining additional data to corroborate their belief, it's just that, a belief.
Pastors believe, scientists think.
I'm in a theoretical/hypothetical mode right now that suggests brewing great beer doesn't come from a single change, but from the cumulative effect doing a number of best practices.
I have had three reactions to Marshall's findings and I'm looking at how to resolve them.
One is to take them at face value, the results say it doesn't matter much if at all what we do. That's possible. Maybe beer brewing is an extremely forgiving and robust process, and certain details don't matter enough to be perceived.
The second is that the panels of tasters may be doing something to limit their ability to perceive such differences, i.e., drinking or eating things beforehand that limit their abilities. This is much of what I've contributed to this thread.
The third is potentially the most interesting. What if Marshall's (and cohorts') results ARE true, that the panel tastings are accurate and legit, and that the lack of differences usually detected is how it is.
Does that mean nothing matters? Not necessarily....there's another possibility at least. That is, that there's a cumulative effect of best practices on final product. Imagine for the sake of illustration that 1.0 is the threshold of perceptibility. Imagine any one change one might make produces a .8 result (below perceptibility, in other words). A single change in an experiment might never rise to perceptibility but the cumulative effect of many small changes might. For instance, four different changes, examined in isolation, produce no discernable difference. But taken together, .8 + .7 + .5 + .8 add up to 3.0, which would cause us to notice a difference.
I know the short-and-shoddy experiment they've tried to do attempted to assess this in part, and IIRC they do tend to show an effect though not a great one. Not sure again if the panel's abilities to taste are compromised, but there's a bit of a suggestion there that maybe there is a cumulative effect. But it's a single time, and only that--which to my mind means worth doing more research on.
The approach I've bee using as a homebrewer for a while now is simply the idea that best practices are exactly those, and the cumulative effect of my brewing with best practices will result in better beer. So even though some results suggest single changes don't have much effect, a cumulative effect might be of greater importance.
My beers have gotten better and better as I've done that, and I don't think that's a coincidence.
So--which of the three above is correct? Or are all three correct in their own way? And is there a fourth I haven't listed?
<sigh> Science is hard. But it's fun.