Cost Estimation in Mega Infrastructure Projects

I recently read an article that i think would be relevent here. http://www.theatlanticcities.com/jobs-and-economy/2013/07/why-mega-projects-end-costing-way-more-expected/6364/ . Mega projects that stretch over many years are bound to have some budget overruns, but to the tune of $300 million to $2.5 billion? With the increase in technical know-how and IT resources, should we not be able to estimate the costs appropriately? Is the author correct in that fact that there is optimistic deluding? I am surprised at the level of gap between the estimated cost and actual cost. According to Flyvbjerg, the error is not in the estimating technique. If it was so the numbers would be randomly high or low and not consistently low. His opinion is that most projects are under budgeted to get funding, project approval and for putting a political spin. The key fact seems to be the manipulation of inflation rate or other cost factors used by consultants which is used to tweak the numbers.

http://www.psmag.com/business-economics/derailing-the-boondoggle-4334/