You are incorrect. The MCAT is not graded on a bell curve. The overall score distributions aren't exactly a bell curve if you look up the histogram data. It's close to a bell curve because with a huge sample size of ~50k examinees a year, a bell curve will result; basically, the law of statistics says that a large sample size would tend to give a normal distribution, and the aggregate MCAT data is proof that the law holds in reality. Grading is NOT based on percentile, nor on "how you did compared to others."
AAMC (
https://www.aamc.org/students/applying/mcat/preparing/85436/preparing_understandingscores.html)
"Examinees often ask if earning a high score or higher percentile is easier or harder at different times of the testing year. They ask whether they have a better chance of earning a higher score in April or in August, for example.
The question is based on an assumption that the exam is scored on a curve, and that a final score is dependent on how an individual performed in comparison to other examinees from the same test day or same time of year.
While there may be small differences in the MCAT exam you took compared to another examinee, the scoring process accounts for these differences so that an 8 earned on physical sciences on one exam means the same thing as an 8 earned on any other exam.
The percentile provided on your score report simply indicates what percentage of examinees from the previous testing year scored the same as you did on the MCAT exam."
Theoretically, the difficulty of the questions on any officially administered MCAT exam
that counts towards the final score (i.e., excluding the trial questions every exam includes unbeknownst to the examinee) has
already been predetermined by the AAMC through methods that remain unknown; this is how the AAMC decides whether the scale is 50-52 = 15 or 51-52 = 15 for any given section it administers officially. One could speculate that the
trial questions given to previous examinees serve the purpose of feeding raw data to the AAMC
for the purposes of determining question difficulty, thus lending some validity to the argument that one's score is benchmarked against others' performance, but this theory has neither been proven nor disproven to date.
Now, obviously AAMC can do whatever the hell it pleases and not have to follow what it says, but I highly doubt that it would go so far as to lying completely through its teeth on its official pages to examinees about how they score the test.