Look at the data used to assess the players in the article I linked for example: http://www.canishoopus.com/2013/6/3/...r-starter-bust (and I know this is entirely arbitrary). By analysing data it attempts to asses the % chances of each player becoming a Bust, Bench, Starter & Star. For example Bennett: BUST=27%, BNCH=38%, STRT=32% & STAR=3%; Zeller: BUST=16%, BNCH=35%, STRT=41% & STAR=9%.
Now I am not suggesting for one minute that these figures are definitive, but for the sake of discussion these figures indicate that Bennett has a far higher chance of being a Bust (I agree), while Zeller has significantly higher chance of being a Starter (I agree). Now I would suggest that in reality Bennett has a higher chance of being a star than Zeller but it does serve the purpose of asking what is BPA? Is it simply the player who has the highest chance of being a Star, regardless of the fact they also might have the highest chance of being a Bust? Or is it the player that is most likely to be a Starter? You then need to take injuries (Greg Oden), character (Demarcus Cousins) into account.
I think that's where tiers come into it. In order to have a discussion between players they need to be graded out at a similar level. You then have a discussion based on injury etc.. Where there's a clear grading difference you have your BPA, but when it's very close other factors can and should be brought into the conversation.