As a scientist, it's best practice to keep a daily notebook called a "lab diary" where you document your work, e.g. details of experiments conducted, day by day.

It is sometimes important to prove when you had a particular idea (e.g. for patenting purposes or copyright lawsuits or documenting the history of a field), and it is important to keep track of the details of experiments in order to be able to reproduce them. For that latter aspect, I recommend self-documenting data nowadays, i.e. data that comes with meta-data to explain how it was derived, such as parameters used to create data in an experiment, which I often encode in POSIX filenames (e.g. <method>-n=<n>-iter=<iter>-k=<k>.eval => "randombaseline-n=20-iter=500-k=3.eval").

Of course you can maintain such a diary in the form of a plain text file, and I often do that as additional companion, but files are editable, you may lose them, and they have not much worth in terms of serving as a proof; a paper diary with daily entries, in contrast, is telling a story that is less likely to be fake.

It also makes you accountable (in case management asks you what you did on March 3 last year, you'd open your lab diary and could say "that was when we had the meeting where we decided to cancel the XYZ project"), and you can use it to extract achievements for your annual or quarterly performance reviews from it.

I use a similar text prepend now with digital todo lists. It still works, but not quite as much. Perhaps because it's not new anymore.