I personally think the responses have been a bit harsh...It can be easy to dismiss something when you have the benefit of hindsight on your side and are looking at things in retrospect.
I don't doubt the group who conducted the experiments are scientists with years of experience in experimental design, but they might not be coffee-extraction scientist and possibly had no hands-on experience in obsessing/making coffee at home. Given what they know then (before the study), that's probably their best stab at the matter.
Prior to embarking the experiment, the scientists most likely didn't know the bed height/shape can affect the extraction time/yield. I can't imagine the scientists being given the brewers a month in advance to play with before proposing and designing the experiment. The experiment could possibly be their first and last chance. In an ideal world (unlimited resource, no deadline), you could conduct unlimited preliminary experiments and refine the experimental design before the 'official' experiment. This may not always be the case especially in a corporate research environment.
If you're not a coffee veteran/expert who obsesses about coffee extraction, the logical variable to control will likely be the grind size. 'Conventional' wisdom says that the grind size affects the extraction and that's what they controlled. But who would've expected you have the bed height/TDS % curve-ball thrown at you, which only became obvious after the study?
About the comments on using two different coffees for different roast levels, seeing that they are sourcing the coffee from Starbucks/Peets, perhaps there are some feasibility/political restrictions that we are not aware of? Any experienced scientist will have attempted to minimise any confounding factors (I have no doubt the scientists had considered this given their credential - but sometimes some compromise has to be made based on feasibility). Perhaps they didn't know how or did not have the resource to get the same coffee roasted two ways, in the required quantity and timeline?
From their perspective, they might be seeing themselves as selecting a 'representative' light roast and a 'representative' dark roast for the study, which can be a totally valid approach for scientific research. For casual/mass coffee drinkers, light roasts surely taste different enough from dark roasts, but between light roast or dark roast itself, the difference may have been much smaller/insignificant. Since the objective of the study is to compare flat vs conical brewer (not dark vs light roast), this may be an acceptable compromise although non-ideal. If I read correctly, they are comparing conical vs flat brewers given a roast level - not "light roast+flat brewer" vs "dark roast+ conical brewer".
The study may have not answered the flat vs conical
question conclusively, but surely there're still some learnings that can be be gained from the study.
Just my 2 cents...