My cynical guess is that they used “page load time” as a metric to prove their new system was “faster”.

And maybe only tested internally on a faster system, or with hot-cache repository loads? It doesn’t seem to do it on reloads, though going away for a bit and coming back seems to cause it again (and, visibly, different parts of the page load at different times).

Anyway, It definitely reeks of “enterprise” so I guess Microsoft finally got enough people into github to steer the ship towards the iceberg.