I don't think too much should be read into the '50 filter' test.
Putting the whole idea of obscene amount of filters being used, it is not a controlled experiment in the least.
To make it even remotely 'scientific' the filters should have been of all the same make and model, and even more from the same batch.
That alleviates the issue of 'cheaper' filters clouding the results straight away, as no filter would be better or worse then one behind/in front of it.
Also, the amount of filters produces a barrel/scope/tube effect - was the base image they did without any filters taken with the same tubed affect or just a straight shot from the lens?
The variables are huge and vast from the 'experiment' they've done that any reading into it, should be taken with a large pinch of salt.
The only way you will know if a filter makes a difference is to take a pic with one on, and one without, then sit and pixel by pixel examine the differences.