Bootstrapping in the strict, Engelbart sense
Not the idiom. A recursive engineering discipline: A-activity does the work, B-activity improves how A is done, C-activity improves how B is done. The leverage is at C, and almost nobody runs it.
Bootstrapping in the strict, Engelbart sense
The popular meaning of “bootstrapping” is plucky self-reliance: starting a company without funding, learning a skill without a teacher. That is the idiom.
Engelbart used the word in a much more specific sense, and the difference matters.
The strict sense
A C-level discipline. The Augmentation Research Center used NLS to design NLS. Every report, every code change, every meeting note was produced in the system they were building. The methodology was the product; the product improved the methodology. Recursive by construction.
This is paired with his ABC activity model:
- A-activity: the primary work. What the organisation exists to do.
- B-activity: improving how A is done. Better tools, methods, processes, training.
- C-activity: improving how B is done. Making the improvement process itself faster, smarter, more compounding.
Most organisations only run A. Some run A + B. Almost none consciously run C. Engelbart’s claim is that C is where collective capability compounds, and the absence of C-activity is why the augmentation thread keeps stalling.
Bootstrapping = C-activity made deliberate
You bootstrap when you systematically improve the improvement process. The product you ship is downstream of how you ship; the way you ship is downstream of how you learn to ship better. Engelbart designed for the C layer first.
Why it is hard
C-activity has no immediate business case. The payoff is delayed and compounding, not quarterly. Most organisations underinvest in B and disinvest in C. Engelbart was openly bitter about this in his later years: the personal-computing industry built artifacts, dropped methodology, ignored training, and shipped C-activity to no one. He saw the augmentation potential collapse to individual productivity.
Why it matters now
LLMs make the C-activity gap visible again. Every team using LLMs is doing B-activity (improving how their A is done) ad-hoc, through scattered prompt sharing and individual experimentation. Almost no team is doing C-activity (systematically improving how they improve their LLM use). The teams that adopt C-activity discipline early will compound away from those that don’t.
For the Guildford talk: bootstrapping is the missing methodological frame for the LLM moment. Not the idiom. The discipline.