Sounds to me like they're working on the margins. IOW: they have a punchlist of a bunch of tiny little things that either slow the process by 0.01% or increase resource costs by 0.01% and they're trying to fix them all in hopes of having a decent combined positive effect.
The problem, of course, is that each of these changes must be considered on its own to determine the 2nd order impacts like "does this change worsen the user experience". We all know that SLR doesn't do that very well--its easier to make a snap judgment then look for some number to justify the decision.
And now we get to maybe the crux of the matter: the tech bro mantra of "move fast and break things". This lack of caution and the tendency to gamble that the end result will be worth pissing off some customers. IMHO this kind of mentality has so permeated tech that they use it even in inappropriate situations like this -- this is the unwritten second part of the mantra "move fast and break things; then go back and fix them later".
