A month by month record of scientists time spend on different tasks is on its face absurd. The proposed methodology, automatic textual analysis of scientists written records, giving you a year worth of a near constant time split pre AI is totally unbelievable.

The data quality for that would need to be unimaginably high.