Do even papers with the same main author differ in style? What if there are "fraud people" who just write more obscure and their papers are typically frauds? I

Do even papers with the same main author differ in style? What if there are "fraud people" who just write more obscure and their papers are typically frauds? I also wonder how the same approach would apply to politician speeches (does their degree of obfuscation change over time or pre-vs-post election?) or even more directly measurable things such as the FOMC minutes. In the best case we could at least see whether they know they lie (they obfuscate unpleasant truths / lies). Then again, it could be difficult to find a dataset with enough positives to train a meaningful categorization algorithm.