As a sysadmin, if I were to be furloughed indefinitely I would probably spin down any nontrivial servers. A server that goes wrong and can't be accessed is a really, really, really, really terrible-horrible-no-good-very-bad thing. And things go wrong on a regular basis in normal times; when the government is shut down and a million things that get done everyday suddenly stop being done, something somewhere is going to break. Some 12-year-old legacy cron job sitting in an obscure corner of an obscure server written by a long-departed contractor is going to notice that the foobar queue is empty , which turns out to be an undefined behavior because the foobar queue has always had stuff going through it before, so it executes an else branch it's never had occasion to execute, which sends raw debugging information to a production server because the contractor was bad at things, and also included passwords in their debugging because they were really bad at things...