When Not to Patent, and What to Do Instead
A patent is a bargain: full public disclosure in exchange for a time-limited right to exclude. For a meaningful share of inventions, the applicant gives up more in that trade than the right is worth.
The default advice given to inventors is to file, and for a great many inventions that advice is correct. But the instrument has a fixed structure, and its terms are not negotiable: everything is published at around eighteen months from the earliest filing date, the right lasts roughly twenty years from filing and not a day longer, it must be renewed at intervals at rising cost, and it confers no ability to make anything — only a right to stop others, exercised at the holder's own expense.
Read those terms as a contract rather than as a prize and the cases where they fail become visible. If the invention cannot be detected in a competitor's product, the right to exclude is unenforceable in practice. If the commercial life of the product is three seasons, a term of twenty years is largely wasted while the disclosure is permanent. If the advantage lies in accumulated process knowledge rather than a describable mechanism, publication hands away the only thing that was scarce.
None of that argues for doing nothing. The useful framing is a menu rather than a binary, and the serious alternatives to patenting — trade secrecy, defensive publication, registered design protection, and plain speed to market — each protect something specific. Choosing between them requires knowing what actually keeps a competitor out of the category.
What each instrument buys
- Trade secrecy
- Indefinite protection with no filing, no publication and no term limit — but it ends permanently the moment the information escapes or is independently discovered.
- Defensive publication
- No exclusive right at all. Places the idea in the prior art so nobody else can claim it against you.
- Registered design
- Protects appearance rather than function. Cheaper, faster to obtain, and shorter in term.
- Lead time
- The months before a competitor can tool up and reach a shelf. Unregistrable, but real.
One · Detection
A Right You Cannot Police Is a Cost, Not an Asset
The first test is whether infringement would be visible. A patent is enforced by the holder, not by the granting office; nobody monitors the market on your behalf and no penalty attaches automatically to a copyist. That means the practical value of a claim depends on whether you could look at a competitor's product and know it reads on your claim.
Mechanical inventions usually pass this test easily. Buy the competing product, take it apart, measure it, compare it against the claim. Manufacturing processes frequently fail it. If a claim covers a method of treating a surface, and the finished part looks identical whichever method produced it, then detecting infringement means obtaining evidence from inside a competitor's factory. Some jurisdictions provide procedures that shift the burden or permit inspection, but the practical position for an independent inventor is that a process claim over an undetectable step is close to unenforceable — while the published specification has taught every competitor exactly how to do it.
The same reasoning applies to internal software behaviour, chemical intermediates that do not survive into the product, and settings or parameters that leave no trace. Where detection is impossible, secrecy protects better than any granted right, because the information's value rests on nobody else having it rather than on nobody else being permitted to use it. This is the calculation behind decades of unpatented industrial know-how: no publication, no term limit, no renewal schedule, and no enforcement burden — just a discipline of access control, confidentiality agreements with anyone who is told, and compartmentalisation so that no single supplier holds the whole picture.
Secrecy has one unforgiving property. It is not a right against independent discovery. A competitor who works it out honestly, or reverse-engineers a product lawfully, owes nothing, and there is no remedy. That is why the choice tracks detectability so closely: what can be read off the product should generally be filed, and what cannot should generally be kept.
Publish what a competitor could reverse-engineer anyway. Keep what they could not. The mistake is disclosing something that was already invisible.
The detectability rule
Two · Timing
When the Term Outlasts the Market
The second test is arithmetic on the calendar. Examination in most systems takes somewhere between eighteen months and four years from filing before a right is granted, and the term runs twenty years from the filing date regardless of when grant arrives. For a product whose category turns over annually — an accessory tied to a device generation, a novelty item, anything driven by a trend — protection may well arrive after the commercial window has closed.
Where the window is short, the instruments that fit are the fast ones. Registered design protection covers the appearance of a product rather than how it works, is typically granted in months rather than years, costs a fraction of a patent, and is often the more relevant right in categories where buyers choose on form. Copyright arises automatically in drawings, code and written material with no registration step in most territories. Trade mark protection compounds rather than expires, and for a product likely to be imitated on function anyway, a recognised name can outlast any twenty-year term.
Lead time is the underrated instrument in this group. A competitor deciding to copy still has to source components, commission tooling, run a production batch and secure distribution — commonly six to eighteen months for a physical product. An inventor who uses that interval to establish supply, retail relationships and a name has built something a patent could not have delivered inside the same period. Accounts of products reaching the market, such as this record of one consumer device and the inventors behind it, consistently show execution occupying more of the timeline than any legal step.
Three · Defensive Publication
Giving It Away on Purpose
The least intuitive option is to publish deliberately, seeking no right at all. Because the prior art comprises everything available to the public before a filing date, a dated public description of your invention prevents anyone else from obtaining a patent over it. You gain no power to exclude; you guarantee that nobody excludes you.
This is the rational choice in a narrow but real set of circumstances. It suits an improvement that is valuable to use but not worth twenty years of renewal fees. It suits a component an inventor wants to remain freely usable because their revenue sits elsewhere in the system. And it suits the case where a well-resourced competitor is visibly working the same ground and the realistic risk is not being copied but being blocked. Publication is cheap — a dated technical description in any indexed venue does the work — and it is irreversible, which is exactly why it should be decided rather than drifted into.
The unintentional version is the trap. An inventor who demonstrates a prototype at an event, posts a detailed build description, or launches a crowdfunding page has published defensively whether or not they meant to. Some jurisdictions extend a grace period of roughly twelve months after the inventor's own disclosure; many apply an absolute novelty standard under which rights in that territory are gone on the day of publication. Deliberate defensive publication and accidental self-disclosure produce the same legal effect; only one of them was chosen. Practical walkthroughs of the route from concept to product, including a step-by-step account of turning an idea into something manufacturable, put the filing decision before the promotion for exactly this reason.
Four · The Mixed Position
Most Real Answers Are Combinations
Framing this as a single choice is itself the error. Products are rarely protected by one instrument. The common arrangement files on the mechanism a buyer can measure, keeps the process parameters unpublished, registers the appearance where form drives purchase, and builds a name that survives the expiry of everything else. Each element covers a different exposure, and the portfolio costs less than a defensive thicket of applications while protecting more of what matters.
Regulated categories tilt the balance further. Where a product requires testing and certification before sale, the approval itself functions as a barrier: a competitor must repeat the process, at cost, on their own timeline. Devices making health-related claims are the clearest instance — the route described in this account of a wearable pain-relief product reaching users shows compliance work and clinical positioning shaping the timeline as much as any filing does. In those categories the sensible pattern is a narrow patent on the specific mechanism plus heavy reliance on the regulatory moat.
Where filing is the answer, the form of the filing is a further decision rather than a formality — early placeholder filings, divisionals, continuations and international routes carry different costs and deadlines, and the differences between the available kinds of application and proceeding determine how much optionality survives past the first year. The choice is not simply whether to file but which vehicle keeps decisions open for the longest at the lowest cost.
Two questions settle most cases. Could a competent competitor work it out from the product on sale? If yes, secrecy is an illusion and filing is the honest option. Would the exclusive right, if granted in three years, still be commercially meaningful? If no, spend the money on tooling, distribution and a name instead. Applying both tests early also protects the budget: the four checkpoints described in the main report on why invention projects stall all become cheaper to clear once the instrument has been chosen deliberately.
Choosing not to file is a strategy when it is a decision, and a loss when it is a delay.
On deliberateness
The Verdict
A Decision, Not a Default
The distinction that matters is between a considered choice and an omission. An inventor who evaluates the alternatives to patenting and concludes that secrecy plus lead time protects the position better has made a defensible strategic decision, and can act on it immediately — controlling access, papering the confidentiality obligations, moving on manufacture. An inventor who simply never files, then discloses at an event, has arrived at the same legal position by accident and lost the option of changing course.
Sequence protects the option. Keep dated records as the design develops. Put confidentiality in place before technical detail reaches a manufacturer or partner, noting that an agreement signed after a meeting protects nothing said during it. Decide the instrument before any public disclosure, since that single event closes doors in many territories permanently. And revisit the decision when facts change: an invention that started as a short-lived accessory can turn out to have an industrial application where twenty years of exclusivity is worth a great deal, and that reassessment is only available to someone who has not already published.
The question was never whether patents are worth having. It is whether this invention, in this market, on this timetable, is better served by exclusion or by speed — and that is answerable long before anything is filed.
End of report