Reading 06 · Strategy
Intellectual Property — To Patent or Not?
A patent is a bargain: protection in exchange for disclosure. But when LLMs can legally ingest that disclosure and extrapolate around it, the bargain tilts — and silence can be the stronger moat.
The thesis
For startups in the AI era, traditional patents can be counterproductive: public disclosure feeds the very systems your competitors ride. Build market share and defensibility first; disclose from strength.
The argument, in five moves
Rethinking IP when disclosure feeds the machine
AI amplifies disclosure risk
Patent filings require full disclosure that LLMs can legally ingest under fair use, letting competitors reverse-engineer adjacent solutions without ever infringing the patent.
Disclosure accelerates the search
Publishing an invention defines the search space on the frontier. With simulation and testing costs falling, rivals can iterate across it faster than legal protection can prevent.
Barriers matter more than rights
Real advantage lies in knowing where the domain barriers are — not in holding a patent — especially as foundational models standardize whole domains.
Speed over legal protection
In fast-moving innovation cycles, the key advantage is the speed of innovation and market-building, rather than IP protection as such.
A founder's playbook
Play it safe: protect knowledge until you have established enough market position and resources to actually defend the moat.
Knowledge disclosure of any kind is risky; legal protection for ideas seems to be on thin ice.
The Nyāya lens · गुप्त विद्या
The tradition of guptā vidyā — guarded knowledge — held that some knowledge is transmitted only when the recipient is ready and the moment is right. Strategic secrecy is not paranoia; it is the dharma of timing, deciding not just what is true but when it is wise to disclose.
Distilled from the original essay on engkraft.com.