Restoring double slit interference by disconnecting the recording device I think it was my graduate quantum mechanics textbook (Gottfried and Yan), I read a description of the double slit experiment.  It first described how the act of measuring which slit the particle goes through destroys the interference pattern.  Then it went on to give a detailed exposition.  It gave a example in which the detector at the slits writes to a computer hard drive which slit the particle passed through.  Then it said that even if you leave everything the same, and have the detector interact with the particle at the slit, you can still create or destroy the interference pattern by disconnecting or connecting the hard drive at a location arbitrarily far from the experiment.  I want to read about this again but I am not sure where I read it.  I got a copy of Gottfried and Yan but it does not have the thing I want in it.  It is a different version than I used in grad school as well, and I'm no even sure this is where I read what I'm looking for.  Does anyone know where I can find or read about something like this: how even putting the detector to detect the particle does not destroy the interference pattern if it is not allowed to write the result to the HDD.
References appreciated, thanks!
 A: 
even if you leave everything the same, and have the detector interact with the particle at the slit, you can still create or destroy the interference pattern by disconnecting or connecting the hard drive at a location arbitrarily far from the experiment.

I would be surprised if any (good) textbook says that, because it's not necessarily true. If you have a detector that detects which slit the particle goes through, then the process of detection probably already qualifies as a measurement, even if we don't connect the other end of the wire to the hard drive. I say "probably" because I can't rule out some clever way of avoiding any practically irreversible processes until the moment the signal hits the hard drive, but that would be a pretty amazing technical feat. In any conventional set-up, the measurement has already occurred long before the signal hits the hard drive.
The occurrence of a measurement does not depend on whether or not we humans record the outcome in any useful way. It only depends on the fact that nature records the outcome, in the sense that different outcomes lead to different mutually orthogonal states of the world that cannot be mixed with each other by any future practically-measurable observable. When that happens, the interference pattern is destroyed, and no feasible future action can bring it back — because that kind of practical irreversibility is built into the definition of "measurement."
