> They dont get into how they got to the 10^38 figure, but to me that seems like the most fishy part If the test was multiple choice with 5 options for each qu

> They dont get into how they got to the 10^38 figure, but to me that seems like the most fishy part If the test was multiple choice with 5 options for each question, and you assume that if they got the question wrong, they randomly guessed the answer, and their answers to all incorrect questions were independent of each other, and forget to account for the fact that one of those 5 options is correct (or there were 6 options, which is less common for exams in my experience), then you get approximately 1 in 10^38 for the probability their wrong answers were all the same. But those are bad assumptions. People don't randomly guess the answer every time they answer incorrectly, and people who have shared experiences (like say taking the same classes) are more likely to answer the same way. And answers to different questions are probably correlated, especially in such a long test, where there are likely multiple questions covering similar material. Now, it is possible that computed the probability a different way, and it is just a coincidence they came to the same number. After all, the above method doesn't account for how likely it is that they answered correctly on the 242 other questions they answered the same, which would probably require some analysis of other students who took the same exam and maybe how similar the twins were on previous exams. But it does seem suspicious that it is easy to get the same number with a very naive estimation.