A best play necessarily exists in every position. Just because I can't tell you what it is doesn't mean it doesn't exist. That's like saying "There's no such thing as a biggest object because you can't tell me which object is the biggest.">>
That's completely illogical and it doesn't follow the dictum that different players have different tastes. Obviously a computer's assessment is the assessment of its programmers.
It is completely logical based on the definition of the word "best". Also, you've proven you don't understand what these programmers are doing at all since you think they're personal judgement comes into play.
I don't know what I expected from someone who thinks there isn't a consistent outcome from two perfect agents playing chess.
I think the fallacy here is in thinking there is a single best move always. Because there's not just one possible way to win, that's not necessarily true right there. Especially in the first move. Of course, we're assuming a deep deeep deeeep engine, with unbounded computational power. There's a theoretical limit given by the fact that there are really theoretically infinite outcomes. I mean infinite, unless you bound the number of plays.