> I think a lot of commenters here don't appreciate that prosecuting a large bank involves sifting through literally tens of millions of documents looking for patterns of intentionality, which as you say is very hard to prove, and where the subjects of the investigations in question know it's hard to prove
And to reiterate, this isn't something that you can 'solve' by throwing machine learning or data analysis behind it. It means sitting down and reading through tens of thousands of emails between potentially hundreds of people, sent over the course of years, to understand the inflection in a given sentence, or to understand if anything is insinuated through the use of a given word. Then matching that back to actions. Then explaining to the judge and/or jurors as to why it's correct, and defending again the opposing counsel saying 'Oh you're just reading too much into it'.
And to reiterate, this isn't something that you can 'solve' by throwing machine learning or data analysis behind it. It means sitting down and reading through tens of thousands of emails between potentially hundreds of people, sent over the course of years, to understand the inflection in a given sentence, or to understand if anything is insinuated through the use of a given word. Then matching that back to actions. Then explaining to the judge and/or jurors as to why it's correct, and defending again the opposing counsel saying 'Oh you're just reading too much into it'.