Having read the full paper, it's about two things: exploring how DH can be mathematically attacked (but still computationally infeasible, just much easier), an

Having read the full paper, it's about two things: exploring how DH can be mathematically attacked (but still computationally infeasible, just much easier), and, ways that a library implementing DH could be backdoored/weakened to make one of those attacks computationally feasible to a well-equipped adversary. EDIT: if you like code more than math, the paper links to this repo for some examples: https://github.com/mimoo/Diffie-Hellman_Backdoor https://github.com/mimoo/Diffie-Hellman_Backdoor