My official stance on the movie is that it was diverting but judging Inception within Christopher Nolan's oeuvre, I'd rank this around The Prestige or Insomnia. In stories that create their own worlds from scratch, if they create their physics and rules arbitrarily, then viewers will arbitrarily find rules to question as inauthentic or inorganic, as I did with the Limbo idea which seems like an inelegant exception to and cheap way of creating tension out of the "You just wake up if you die in a dream" rule.
That said, some of the film's detractors complain about how its overly crisp depiction of dreamworlds doesn't resemble real-life dreams, a complaint that I find fairly weak because Inception is only ostensibly and superficially about dreams -- at its heart, it's about watching film.
The line I'm drawing from dream to film spectatorship isn't that radical of a leap in the context of psychoanalytic film theory, which is founded on Lacan's post-structural take on Freud. Dreams are obviously within the realm of psychoanalysis, but psychoanalytic film theory puts its critical stamp by arguing that watching film is particularly akin to dreaming. This argument begins on a basic level -- film and dreams typically occur in a darkened space that suppresses most of the subject's senses as it envelopes them in audio/video stimulation -- before it heads off into the esoteric stuff (see the mirror stage, suturing), which I'll leave alone for the real critics (and because as far as Lacan goes, I'm petty bourgeois).
However, the main point I want to bring to emphasize is that in psychoanalytic theory (in both its literary (read: Freudian) and film strains), the entire text is the unconscious dream. If the text is the unconscious, then the individual characters within the text aren't the subjects of the dream, they're simply projections, if you will, of the text's unconscious desire -- which is echoed in the "people" that populate the dreamscapes in Inception and are called, you guessed it, projections. (Apologies if I'm mangling Lacan, my excuse is that it's hard, I didn't do the reading, and slept through the lectures.)
Keeping all of this in mind, "Is the whole movie a dream?" isn't the question to ask -- rather, we ought to ask, "Is the whole movie a movie?" in which case we know the answer (if we want to get cute, maybe we could say it's a dream with a movie -- within a movie). As a result, complaining about the staircases (that it's "pointless," or because they only work "if you're not actually on the staircase" is beside the point. It's not for the sake of the characters that the staircase is there; the it's for the sake of the film viewer to laying bare the dream-like, involuted Möbius loop of film-watching and indeed all of narrative itself.
To leave theory aside for the moment, another reason why I find the "it's not like real dreams" complaint to be trifling criticism is because real dreams are boring and incoherent. They might seem engrossing while or immediately after dreaming, but when you scrutinize them dramaturgically, they lose their engrossing qualities because they probably were never there. It's the Stoner's Paradox -- what someone finds captivating when they're not sober or conscious often turns out be unremarkable when they're sober or conscious. (Also, the blogger's complaint that Inception is almost entirely composed of montages? That seemed to serve the Soviet montagists fine.)
But that particular blog is based in the UK, and I recall that Lacan isn't terribly en vogue there (or anywhere besides America -- USA! USA!).
As to the question we shouldn't ask -- "Is the whole thing a dream?" -- I would say yes, since all the shots we see of Cobb's kids are all shown with in the same "memory effect" style, and that includes when Cobb is reunited with them.