Twitter's algorithm is actually relatively simple; the basic structure is a prediction engine that tries to assign probabilities to certain events coming out from you seeing a post (do you reply?, click it?, etc.). It then applies a score to each of those events, multiplied by that predicted probability, and decides based on that whether this is a good post to show you. (An obvious example being trying to predict if you'll report a post, which provides a strong negative contribution to the final score, so posts you're likely to report are less likely to be shown to you)
All you need to hack it is just to add constant terms not multiplied by anything, like "did Elon musk post this or reply to it?" and you can skew the result easily.