I'm generally a big fan of incremental improvements over major upheavals. Incrementalism seems to be the only way to affect change in complex systems. With big changes, you have no idea what the side effects will be. With smaller changes, you get to see how the system responds and adapt your strategy based on the feedback.

But when does incrementalism not work well?

• When you really need to start over. Sometimes you need to start from scratch. If you're a carpenter and you want to be a millionaire, incremental improvements to your woodworking skills might not get you there. You need to do something very different from what you're currently doing. You might need to get a loan and start a bigger business, go back to university for an advanced degree, or take some other big bet to make that kind of money.
• When there are winner-take-all effects. In war, one side wins and the other loses. The winner gets everything, not just a little more. In winner-take-all situations, if one side is incrementally committing more resources and the other goes all in, the incrementalist will lose.