Not sure what that rant had to do with the current problem...
Estimating ANY big project (public or not) is an art that few get right. The fact that REALLY big projects are mostly public tends to concentrate the failures (and their visibility) in that realm, but there have been many corporate overruns too. Any time you still have a lot of unknowns when getting close to construction you need either an obscene contingency (which isn't acceptable to the public - who see it as a slush fund - or corporate beancounters) or accept the significant probability of an overrun. We humans just aren't very good at dealing with uncertainty.
San Diego's starter light rail line was done by people with sharp pencils, and doing it with local and state funds (yes, it used both) avoided a lot of uncertainties associated with federal procedures. A key reason for not using federal $$ was avoiding the Buy American requirements for the cars. The starter line got off-the-shelf equipment from Germany, before Siemens set up shop in Sacramento, at a time when compliant alternatives were poor quality and very expensive. All later expansions there, though, have gone the usual federal route, and there are several options now for equipment that meet Buy American without being crazy about it.
BART actually came close to its original estimate, at least in percentage terms - roughly a 25% overrun, which is pretty good for large projects that take 5-10 years to build. Bonds covered just under $800 million; cost for the original system was about $1 billion (in 1960s dollars!). By the time they really needed the extra money, the feds had started supporting transit.
San Luis Reservoir is nearly full. The aqueducts have been pumping full blast, which is the way that lake fills - very little actual watershed. [
cdec.water.ca.gov] shows, as of yesterday, 92% of capacity and 113% of historical average. IOW, operating as intended when there's a lot of surplus flow in the Delta. Other major reservoirs in N & Central California have also filled dramatically over the last few months. SoCal lakes fed from the aqueducts also are near-full, and Shasta has been pushing releases way up to (they hope) avoid having the spillway operate (because it's a waste of water and cancels the flood control benefit). IOW, the drought is taking a vacation this year even if the Water Board hasn't accepted that yet.