12. First principles over best practices

§Part IV is about habits of mind. Methods can be taught; these have to be practiced, and they are what remains when a founder is in a situation no method anticipated. The first is a preference: for what is true over what is done.

§ 12.1What a best practice is#

§A best practice is a conclusion with the reasoning removed. Somewhere, a company faced a situation, thought it through, and decided. The decision worked. Other companies copied the decision without the situation, and after enough copies the decision became a rule. "Charge per seat." "Onboard with a checklist." "Never show an empty state." Each was right, once, for someone, for reasons.

§The problem is not that best practices are wrong. Most are right in most cases, which is how they became best practices. The problem is that a startup is, by definition, not most cases. It is trying to do something that the companies who wrote the rules did not do, and the rules encode their situation, not its own.

§ 12.2What it looks like in discovery#

§The original guide gave the four moves, and they are the right four.

§Question the industry's best practices. Not reflexively; ask what situation each one was a conclusion from. If the situation still holds, keep the practice. If it does not, the practice is a superstition.

§Understand the fundamental need. Under the feature request, under the job, what is the person actually trying to have happen in their life? Jobs to Be Done gets partway. First principles keeps asking until the answer is something a human would recognize as mattering.

§Identify the core technical constraints. Not the ones the industry assumes, the ones that are actually physical or mathematical. Many "impossible" things in software were impossible for a reason that stopped being true, and the industry did not notice.

§Find the novel approach. Which is not the goal; it is the consequence. A team that reasons from what is true about its users and its technology will arrive somewhere the best practices did not, because the best practices were reasoning from somewhere else.

§ 12.3The example everyone uses, briefly#

§The version of this idea most founders know comes from Elon Musk describing how SpaceX priced a rocket: not by what rockets cost, but by what the materials cost, and then asking why the gap was so large. It is a good story and it is over-told, so one sentence of it is enough. The move was to refuse the industry's number and ask what the number was made of.

§ 12.4A product version#

§ 12.5The discipline#

§First-principles thinking is expensive, which is why nobody does it for every decision. The skill is knowing which decisions are load-bearing enough to deserve it: the pricing model, the core interaction, the first market. For those, decompress the best practice, find the situation it was a conclusion from, and check whether you are in it. Usually you are not. That is the point of a startup.