How Big Things Get Done: The Surprising Factors That Determine the Fate of Every Project, from Home Renovations to Space Exploration and Everything in Between

Big things are most likely to be completed successfully when they are made up of a collection of easily understood small things. Solar energy installations tend to be done on time and under budget. Installations consist primarily of repeating solar panels that are assembled into larger arrays. A half-complete solar installation is also perfectly functional. Many people have experience with solar installations. There has been a feedback of learnings and expertise.

Nuclear power plants are prone to large cost and schedule overruns. Few people have had experience building nuclear plants. They must be built correctly. Mistakes can be catastrophic so must be avoided. A nuclear plant is also useless until it is totally complete.

Big projects often involve politics. In order to get a project greenlighted, the cost is downplayed and benefits exaggerated. These rosy projections are often needed to get a project started. There are additional political issues in development. Rather than get experts, politicians prefer their constituents. (Thus, rather than hire European high speed rail experts, California hired Americans with little high speed rail experience.) Politicians also favor uniqueness rather than tried and true. Seattle opted for a giant highway tunnel, rather than two smaller ones. This did result in one of the largest single-bore tunnels, but at great cost and time expense. (Meanwhile, just a few miles away, a light rail tunnel using more standard technology was on time and budget.)

With large scale projects, it is often best to compare to a reference class of projects than try to add up all the components. If renovating a house, you are more likely to get a good estimate by looking at the experience of other houses in the neighborhood than by adding up all the potential charges individually. (Alas, it can be hard to find the data.) This approach does seem to mirror what happens in artificial intelligence. In an attempt to build a computer to beat grandmasters at Chess, Go, etc., the first attempts were based on programming rules. Computers would be programmed to understand the various possible moves and best responses. Grandmasters could still beat the machines. Then the approach switched to just studying games and learning from past behaviors. This improved the success rate of the computers. The same can improve are success rate for completing big projects.  

10/17/23 how big things get done write - reference class forecasting . Political intentional wrong. Hard to find old data.