Most projects do not fail because they are technically hard. They fail because they are not worth building.
The failure before difficulty
There is a failure mode that happens before any code is written, before any experiment is run, before any grant is submitted. It is the failure of problem selection — choosing something that is tractable but trivial, novel but useless, or important but misframed.
This failure is invisible because it produces no error messages. It produces only a slow draining of energy.
Taste as a constraint
Taste is not aesthetic preference. It is a constraint on what problems are worth your finite attention. Good taste in project selection means:
- Choosing problems where your specific skills create asymmetric advantage
- Avoiding problems that are already well-solved by people with more resources
- Preferring problems where the answer, if found, would actually change something
Institutions encode taste
When you build within an institution — a lab, a company, a competition — you inherit its taste, whether you agree with it or not. The institution's incentives become your constraints.
The best builders I know are deliberate about which institutions they build within, and honest about which compromises those institutions require.
What serious looks like
Serious projects share a few properties:
- The question is well-shaped — specific enough to be answerable, general enough to matter
- The methods match the question — not over-engineered, not under-powered
- The builder can explain why this, why now, why them
- Failure would still produce useful knowledge
Everything else is hobby, performance, or procrastination dressed as productivity.