1) A permanent, chronological record of experiments. For example, Charles Darwin’s notebooks are available online http://darwin-online.org.uk/. These notebooks have lasted over 175 years!
2) A record of ideas and approaches taken during the course of a project.
3) Legal documentation of the timing and outcome of experiments.
Is it because we don’t “do experiments”? That seems wrong if only because every model we estimate is a kind of experiment (maybe not a good one, but still a kind of experiment).
Is it because we’re lazy? Academic researchers tend to work harder than the (very sparse) financial incentives associated with publication would indicate.
I suspect it’s because we’re not good at documentation generally. Witness the plethora of datasets without fully documented codebooks, measurement strategies, or sampling frames.
We’re all busy people so it’s hard to switch gears after so many years of not using lab notebooks. Let’s start by forcing our PhD students to do so. It can’t hurt.