Here is my “best practice saying”:
The label “best practice” is most typically applied to emergent, questionable or even bad practices so that people holding minority opinions can win arguments.

If you cannot win an argument for your approach based on the merits of the approach – simply label it as a “best practice”. The logic used is, “Who could argue against a best practice”.
Sadly this approach works because of the tendancies of human nature best reflected in the Stanley Milgram experiments:

http://en.wikipedia.org/wiki/Milgram_experiment.

People using this “best practice” tactic use verbal prods very similar to the Milgram experiment such as making the statement, “The best practice requires that you comply.” or “You must comply.” in a monotone voice while wearing a lab coat and horn-rimmed glasses.

P.S. This is not just about the “deprecating Sakai 2.x static covers” argument (which I fully expect to lose) – it is equally a reference to IETF’s BCP-47. While BCP-47 may be a fine idea – the 100+ page document approved in September 2009 looks more like wishes and dreams than actual “best practice” – it looks like “what some group of 10 idealists someday hopes to be adopted in the far-off future as best practice’. I have no problem with the ideas in BCP-47 – in many ways the ideas are quite good and very forward-looking – I just find it galling that a significant change in direction for representing locale is arrogantly labelled “best practice” so early in its lifecycle with so little adoption and support around it.