Mmmm.. I wonder if maybe some stray bits at power-on cause the init sequence to fail once in a while?

When that happens, Hijack outputs "Fan control error; disabling" to the serial console. I suppose it really should do a retry or two before giving up completely, though. Mmmm...