http://www.ted.com/talks/malcolm_gladwell_on_spaghetti_sauce...
tl;dr when one asks a question involving people, and you want the most fitted data, then you need to consider grouping/segmentation of the population into clusters of preference.
What is the best spaghetti sauce?
It's a flawed question, as it contains an invalid assumption.
Thus what Curtis seems to be describing is a (great/awesome/very good) etc set of knives, but not 'the best'.
Very good = a maximization of universal requirements
Best = maximization of universal && local requirements (population segmentation preferences, spacial and temporal context etc)
Example: Those forks may be best for Curtis at his dinner. They are certainly not best for me, on my camping trip. Or best for a tribe in Africa with different shaped mouths and habits etc. Or best for someone eating Chinese takeout. etc