You're assuming (or implying) that there's some canonical scale of quality, that it's absolute, and that it can be perfectly determined.
None of these are true.
There is, frequently, a probabilistic scale of quality -- there are works or instances which will, generally, be considered "good" and some which are considered "bad", and, say in an iterated experiment model you'll generally come to at least some level of agreement as to what's what. Sometimes there are even very clear winners and losers.
But there's a hell of a lot of variance. There are things which people, often many people, like, often a lot, which others simply despise. I am, for example, not a fan of rap, hip-hop, house, and similar music. I really like Beethoven's late string quartets. I'm aware that by popular vote my preferences are likely unpopular.
And even where there is widespread agreement after the fact on something, experts in the field may not come to agreement. The Beatles were panned by several early music critics and record labels. Rob Malda's initial review of a new cellphone was less than whelmed. Oh yeah, it was called the "iPhone", you may have heard of it.
Tests for categories as seemingly refined as classical music performance -- both auditions of performers, and trials of classic vs. modern instruments, and fine wines, show that there are huge differences in non-blinded vs. blind trials (for music), and that wine experts are incredibly unreliable in how they grade, or even describe the qualities of, wines. Though this does make for an entertaining way to spend an afternoon or evening with friends.
http://www.sfgate.com/business/article/Blind-Auditions-Putti...
http://freakonomics.com/2010/12/16/freakonomics-radio-do-mor...
In another interesting study, unreleased music was introduced to several different "worlds" of test subjects -- within a world the experiences were shared, but not between them. The result was that, from the same starting selection, different "hits" were selected by each.
http://www.npr.org/2014/02/27/282939233/good-art-is-popular-...
And that's the whole point.
In any competitive selections process, especially where there are far more qualified applicants than there are positions, there will be many qualified applicants who aren't admitted. The reasons for denial will vary, but ultimately the selectors must adopt some arbitrary mechanism. Lottery would probably be the fairest, say, by assigning a rough percentile ranking to each candidate (perhaps by decile), selecting all within the top category (the clear winners), none below some cutoff, and a random selection within the grey range. More often culture, bias, hubris, or other factors will lead to other bases for decision.
But thinking you can, perfectly and without error, assess the quality of something as both nuanced and capricious as the possible value of a technology start-up concept based on a few brief exposures is really pretty arrogant. And all but certainly wrong.