I suspect this is because these are wolves in operator's clothing. If you deploy true operators that are doing labor in their area of expertise, and if they can pre-empt obvious mistakes + generate reproducible playbooks, you could net out ahead.