I would add that they are stubborn in their estimate of an objectively known quantity knowing that their estimate will be compared to the "correct" value. In a

I would add that they are stubborn in their estimate of an objectively known quantity knowing that their estimate will be compared to the "correct" value. In a situation where that threat/reassurance is absent, the incentives could be entirely different. (Using the length of a border defined by natural features [http://www.economist.com/node/13496212 http://www.economist.com/node/13496212] as an objectively knowable value might be problematic, but I believe that was the intent.)