How Big Things Get Done - Bent Flyvbjerg

• Gems

  • "You want the flight attendant, not the pilot, to be an optimist. What you need from your pilot, and must insist on, is hard-nosed analysis that sees reality as clearly as possible."
  • "Repetition is the genius of modularity; it enables experimentation."

• Intro

  • "The Empire State designers were inspired to imagine their process as a vertical assembly line — except that 'the assembly line did the moving,' Shreve explained, while 'the finished product stayed in place.'"
  • The Empire State building was designed fully on paper before construction began. They knew precisely how many beams and other materials they needed.

• Think slow, act fast (chap 1)

  • Big projects generate "blizzards of numbers" for each stage. When doing a post-project retrospective on what went wrong, it's hard to get an accurate picture. And there's usually spin, because people's reputations are at stake.
  • "The actual mean cost over-run of a major building project is 62 percent."
  • The cost overrun of a big project follows a fat tailed distribution, not a normal distribution. This is because big projects are complex interdependent systems.
  • Home builders and remodelers can't reliably estimate the project's cost. The fat tail overages due to complex system effects are too difficult to handle.
  • Bent suggests "closing the window" by completing the project quickly, so that it's exposed to fewer changes in dynamics or black swan events that occur during construction. In other words, risk builds over time until the project is done, so stretching out the timeline is likely to incur unforeseen costs.
  • Argues for a waterfall approach. Robust, slow planning, because iteration during planning is cheap. Costs explode once you enter building construction or film production.
  • "Projects don't go wrong, they start wrong."
  • Big transportation projects often break ground with only a vision in place, not a comprehensive plan with all blockers resolved.
  • "People run around trying to fix things. More stuff breaks. There is more running around. I call this the 'break-fix-cycle.' A project that enters it is like a mammoth stuck in a tar pit."

• The commitment fallacy (chap 2)

  • "Tattoos, marriages, big projects: In every case, we know we really should think it through carefully, so why do we so often fail to do so?"
  • Strategic misrepresentation: superficial planning which glosses over major challenges, wins the contract, and then surfaces the cost overruns once the buyer is deeply committed.
  • "You want the flight attendant, not the pilot, to be an optimist. What you need from your pilot, and must insist on, is hard-nosed analysis that sees reality as clearly as possible."
  • Type I decision making — fast, intuitive — comes naturally to us, but does not work well for big projects. The iceberg of missing information is too big.
  • Planning fallacy / Hofstadter's law: "It always takes longer than you expect, even when you take into account Hofstadter's Law."
    • People tend to use their best case scenario as the average case.
  • Action bias does not work well for big projects, because the decisions aren't often reversible, as they are in business.
  • Reframe planning work to be the highest quality action: "Planning is working on the project. Progress in planning is progress on the project, often the most cost-effective progress you can achieve."
  • Escalation of commitment
    • This strategy is to spend money quickly through a lot of action, so the sponsor becomes committed. They can't back out or they would lose their investment, and lose face. "Start digging a hole."
    • People cannot remove sunk costs from their decision making about the next phase of the project.
    • Even if politicians understand the sunk-cost fallacy, they press on, because the public will likely be swayed by sunk costs.
  • On building the Pentagon:
    • "Somervell was also steeped in a 'can-do' army engineering culture that prized System One decisiveness and getting stuff done above all else."

• Think from right to left (chap 3)

  • Meaning, start with the end in mind and work backwards.
  • A slow planning process is not necessarily helpful; only if the slowness is a byproduct of depth, rigor, and imagination in exploring alternatives and second-order effects.
  • "Questioning isn't doubting or criticizing, much less attacking or tearing down." Rather, it's being curious. When embarking on a project, assume there must be more to learn.
  • Ask "why" about the goals of the client and the goal for the project, and understand them thoroughly before shifting one's thinking to the means.

• Pixar planning (chap 4)

  • To get a detailed and reliable plan, one must gather information, run simulations and experiments to discover unknowns and derisk the project.
  • The Sydney Opera House is considered by many to be the most beautiful and important architecture work of the last century. Its construction was a fiasco. The Sydney official rushed it into production before his retirement, before it had any substantial planning, so that he could leave behind a legacy.
  • Guggenheim Bilbao was designed and tested for structural engineering feasibility on a computer, before entering construction. The simulation was the real building's "digital twin."
  • Pixar does extensive paper prototyping of films before entering production. Storyboards for the whole movie are made, recorded in video, shown to employees and the brain trust, and iterated many times. It takes years of this low-fidelity simulation to get it right.
  • "By the time you see the film, it's the ninth version of the movie that we've put up." - Peter Doctor
  • Iteration provides the freedom to try things out. With one-shot production, you must avoid risks by sticking to what's safe and most likely to work.
  • The author compares his advice to Lean Startup. Both prioritize cheaply testing and derisking the plan through iteration. Minimize the cycle time.

• Are you experienced? (chap 5)

  • Bent argues that experience is paramount for big projects, but it's routinely deprioritized when assembling the team. People want the money and jobs from the big project, and will lobby to be the ones to execute the project.
    • E.g. hiring domestic contractors rather than more experienced foreign ones.
  • Government project planners don't visit other countries to learn how they accomplished a similar project. The knowledge is siloed.
  • "Drawing on data from five hundred brands in fifty product categories, they found that almost half of pioneers failed, compared to 8 percent of settlers."
    • The first-mover advantage in big projects isn't clear. Being a pioneer may be glorious, but it puts the project at risk because you have no one else's experience to learn from.
  • It's not just people that become experienced, but equipment and techniques. Using experienced, proven tech reduces risk.
  • "Technology is 'frozen experience.'"
    • In tech we assume new is better, but the new option is inherently inexperienced.
  • Experienced people have tacit knowledge that is easy to undervalue, because it's tacit.
    • "We can know more than we can tell."

• So you think your project is unique? (chap 6)

  • Estimates usually proceed from some anchor value. But then no one asks if the anchor is reasonable for the project. Sometimes the team is even unaware that their estimate is anchored to another value. It's probably best to dismiss the initial anchor entirely, and proceed strictly from first principles.
  • Constructing the Sydney Opera House was a unique project. But it's ultimately an opera house and there is much to learn from other opera house projects.
  • Reference class forecasting
    • Consider a project as part of a class; another "one of those." E.g. the class of all opera houses.
    • Produce a top-down estimate by comparing it to its historical predecessors.
    • This is an "outside view" which helpfully factors in the impact of historical unknown unknowns.
  • This outside view is meant to combat "uniqueness bias." Uniqueness bias is an attempt to adjust a project's estimate up or down from the category's mean based on vague assertions of how this project is different from the rest.
  • Old project data is hard to come by. Planning teams have no reason to save it.
    • "Project planners and managers have a mindset that is focused on the future, not the past."
    • To get it, call people who have done similar projects and get their estimates.
  • His approach to managing black swans: review all very delayed projects of the past, enumerate the reasons for the delays, and come up with mitigations for each. E.g. put hundreds of archaeologists on retainer so a big tunnel dig doesn't get blocked if you encounter artifacts that need to be excavated.

• Can ignorance be your friend? (chap 7)

  • Bent makes the case about how one can only reason about the value of planning from data — a large number of case studies — rather than from anecdotes, especially successful ones, because they suffer from survivorship bias.
  • About The Pixar Way: "When stakes and stress are low, we are freer to wonder, try, and experiment. Planning is creativity's natural habitat."

• What's your lego? (chap 9)

  • Modularity: build huge things by building many small things. You get many reps and can master the process of building the Lego. E.g. building many individual solar panels eventually yields low cost solar panel arrays.
  • Iteration is possible, whereas it's not in one-shot projects.
  • If you build one huge thing, like a nuclear or hydro plant, you only get one rep, and it's one of a kind, so the learnings don't transfer easily to future reps.
  • "Repetition is the genius of modularity; it enables experimentation."
  • Modularity cuts tail risk.
  • "Manufacturing in a factory and assembling on-site is far more efficient than traditional construction because a factory is a controlled environment designed to be as efficient, linear, and predictable as possible." There is no weather or site-specific difficulties.
  • Examples
    • Rather than damming a whole river, divert some of it and build a small hydro-electric unit; repeat.
    • The Giga Factory in Nevada is built from 21 smaller factories, and each one of those is able to produce stuff once it is built.
    • Shipping containers turned cargo transport into a Lego.
    • Small modular reactors (SMR) for nuclear power plants.
  • "Modularization does more than [reduce costs]; it radically reduces risk."
  • (Great table plotting big project types — nuclear, solar, tunneling, mining, rail, olympics — and their fat vs. thin tail risk.)
  • (A case study about how quickly and successfully we've learned to scale up wind farms.)