At the time, the SAT was purported to provide a score that predict ability to perform academically at higher learning institutions.
Along with other factors such as GPA, and participation in extra curricular activities, a school could reasonably determine how well a student would do.
In practice, the normal distribution for scores correlated with the distribution of college performance. It was a reasonable predictor of success, but it did penalize students from certain backgrounds.
Because the test was devised by psychologists and statisticians, uncovering the pattern to the types of questions and the expected answered allowed test prep people to devise tricks to improve scores beyond the expected deviation.
Sure, a big donation by the student's dad, but that's a known quantity. I took the same SAT you did if it matters.
Which certain backgrounds are you referring to? I'd ask you for the references to show the supposed boost that test prep gives to SAT scores, but then I'd have to find the papers that fail to reproduce it...
Finding data from 1998 online is much harder.
Which is precisely what you want in an assessment test. Any assessment test.
You only want a normal distribution if the quality under assessment is normally distributed, but you do want a test where the worst candidate does better than chance, and exactly one candidate gets a perfect score. That's an ideal which is only approximated, but it is the ideal.