It is hard to know exactly what fake work is looking forward, but it is easy looking backwards. Which means you must test your assumptions continually over time. As what was real work in the past can easily turn into fake work. PG's "Alarm Bells" idea seems to be a good start.