Really old question. But as a professor with a PhD in personality psychology, I want to answer the underlying meaning behind your question rather than your question itself.
At least secondarily, what you seemed to be asking was: wouldn't the sorting procedure engender negative consequences for students?
The answer, I believe, is yes.
Firstly, it serves as a self-fulfilling prophecy. See, for example, the famous work of Robert Rosenthal on the Pygmalion Effect. To summarize, a psychologist gives a completely fake test to a number of children, and makes up results, which he gives the teachers. A small portion, he tells them, will "bloom" intellectually that year. At the end of the year, when he returns, those kids have indeed performed better than the other children, because (the author's interpretation) the teachers have responsed to those children differently (e.g., answering their questions with more alacrity, giving them more material because it is believed they can "handle" it). In other words, having the Sorting Hat say in front of a crowd of people (essentially), "You're brave!" or "You're smart!" or "You're basically evil!" will cause a) the child themselves, b) their peers, and c) the professors/staff to treat them as if they were those things. Through a number of subtle but persistent habits, behaviors, and attitudes, these things will therefore be more likely to actually come to pass, even if the child were not necessarily inclined in that direction. See also The Stanford Prison Experiment, which demonstrates that people tend to take on the roles assigned to them by others.
This is roughly akin to the socioeconomic concerns that educational psychologists have about testing that places different ethnicities into consequential programs (e.g., "gifted," "special needs").
Also, as you mention, it does not allow for the possibility of personality change over time. While people's personalities are generally consistent over time and across situations, and some researchers have believed that it stabilize in one's 20s or 30s, research is demonstrating that people's dispositions are actually dynamic across the lifespan. Thus, the Sorting Hat makes a guess about a person based upon a very preliminary sort of understanding of their traits at 11 years old, and pigeonholes them eternally, reinforcing its (potentially flawed or at least underdeveloped) estimate because of the fulfilling prophecy described above.
Finally, I want to point out that the Sorting Hat uses as it's personality assessement system one which current personality psychologists like the least: a categorical measurement. That is, you're either Gryffindor OR you're Slytherin OR... This is akin to saying that people are either funny OR they're not funny. Today, we a) favor a continuum when measuring traits (such as the Big Five), so that you can be "a little" funny or "moderately" funny, rather than FUNNY = YES or FUNNY = NO. Also we b) like to acknowledge the interaction between person and situation, such that in certain circumstances you might be funny (or brave, or loyal, or ambitious) whereas in other circumstances the same person might be a coward, disloyal, or passive. People are a lot more complex than the Sorting Hat gives them credit for.
So, my answer to your meta-question, or secondary question is, yes: the Sorting Hat is basically an evil social experiment. =) At least, in a real-world, developmental sense. That doesn't stop me from enjoying its role in the fantasy world from whence it comes.
