In your experiments, you know, at least subconsciously, that the possibility exists that you will be served Pepsi. In proper experiments the subjects don't know what they are tasting. You merely ask them for a preference. You also make it clear that both glasses may contain the same liquid (and preferably the experiment leader doesn't know which is which). In such cases, the result holds: M is preferred.
Secondly, it's a statistical result. It doesn't mean there aren't people that can't taste the difference.
As for the wine:
Many experiments are flawed. I don't know what experiments you saw, but if you ask experts to distinguish between a glass of red and a glass of white wine, of course they will succeed. The expectations have been set properly. One white, one red. In all cases where people are asked to distinguish between two different things and they are also presented two different things, they will succeed.
That does not mean they won't fail to note something is white wine when you give them a glass of seemingly red wine and ask them to rate this red wine.
The point is not that there isn't any difference between wines. There is and oenologists will be able to tell you the difference... as long as you told them upfront that there is a difference (and there is indeed that difference).
The problem with talking about the detail - that it was a statistical result - is that this doesn't match up with the headline abstract (that experts can't tell the difference).
As for the the statement about my Coke v Pepsi - first time it happened, I had as much idea that it might be Pepsi as the people in this experiment had that they might have been served white wine masquerading as red. I wasn't particularly thinking about it. I'd ordered a Coke, the drink came in a Coke glass and I was fully expecting to drink Coke. My first thought wasn't "Oh, this is Pepsi". It was "There's something funny about this Coke".