Measurement has happened when the data has been written somewhere?

A somewhat whimsical definition of measurement that I posted on Physics Forums yesterday, which I like enough for a moment to post here as well:

I prefer interpretations that take a measurement to have occurred only when a number has been written in a computer memory. Then a paper can be submitted to Physics Review Letters that says, "the raw measurement data (100MB) is available on a CD on request," and goes on to describe the statistical computations that were done using that data to show how well the data matches up with a proposed quantum mechanical model for the experiment. That sets the standard for measurement as "a PRL paper", in contrast to setting the standard for measurement as something like "it's in my head", or "it's in the head of someone who has a Ph.D" (which is a John Bell joke). I might be OK with a PRD paper as an arbiter of whether a measurement happened, for example, but perhaps not with a JMathPhys paper. Endless fun can be had deciding which journals' imprimatur is OK.

As well as being somewhat facetious, this is a rather instrumental definition, but it captures what the experimental data is moderately well. If one doesn't have this data, it rests on the experimenter's word that they did an experiment at all, let alone that the results are what they say they are in a 4 page PRL. The data might be written in a lab book, but given the millions or billions of data points typically generated by experiments that record individual events, it seems unlikely that the data will not be written by computer.

It's to be hoped that the raw measurement data will include all the data about the geometrical configuration of the experimental apparatus, and details of what materials were used in each part of the apparatus, and how they were prepared, that are sufficient to enable the experiment to be repeated. Every bit of calibration data for parts of the apparatus should be in there, too. An experimentalist ultimately only has to provide enough to persuade another experimentalist that the experiment was done well and could be reproduced, and every single number and aborted try that didn't work might not be recorded, but the more the merrier. A record of the meaning of whatever numbers are recorded is of course also important.

A detailed record of when and where events occurred in the thermodynamically nontrivial parts of the experimental apparatus allows us to discover what correlations occurred in the data after the event, always acknowledging that if we look at the data in enough different ways we will certainly find accidental correlations that look significant. Weihs et al., Phys. Rev. Lett. 81, 5039 (1998), "Violation of Bell’s Inequality under Strict Einstein Locality Conditions", is a pretty good example of what can be done in the way of data reporting. A significant part of the data can be obtained from him, and Gregor Weihs' PhD thesis gives lots of details (in German) about the experimental apparatus.