Every R&D result faces a junction with no way back: publish it in a patent application and the secret dies; keep it secret and you cannot stop an independent discoverer; publish defensively and you close both doors yourself in exchange for freedom to operate. This article builds the decision matrix on scoreable criteria — reverse-engineering exposure, life cycle, cost, detectability of infringement — instead of legal-department habit.
Legal basis
The three routes rest on three different foundations in the IP Law 50/2005/QH11, as amended by Law 131/2025/QH15. Secrecy rests on Article 84: no registration, no term, conditioned on necessary secrecy measures. Patenting rests on the grant system: disclosure exchanged for a twenty-year monopoly under Clause 1 Article 93, with annual maintenance fees. Defensive publication creates no right for the publisher; it creates prior art that stops rivals patenting the same solution, because a disclosed solution no longer has the novelty Article 60 requires.
An overlooked consequence of Article 60: publication also destroys novelty for the publisher itself if it later changes its mind, save for the law’s narrow grace cases such as disclosure by an unauthorised person or disclosure in statutory circumstances within six months. A defensive publication is therefore a one-way decision, to be approved like a waiver of rights, not like a communications exercise.
Applied analysis
The working matrix scores five criteria. One, can rivals derive the solution from the marketed product: if yes, secrecy dies at launch, so patent. Two, is the commercial life longer or shorter than twenty years: longer favours secrecy structurally. Three, is infringement detectable: process patents are hard to police, so many registrable processes are still kept secret. Four, budget: multi-country filing and maintenance costs grow geometrically, while secrecy costs sit in control operations. Five, market signalling: certificates serve fundraising, valuation and licensing; secrets are invisible to investors.
On relative cost: secrecy is cheapest in fees — close to zero — but costs continuous organisational effort. A domestic patent costs State fees under the industrial-property fee schedule, agent fees if used, and annual maintenance across twenty years; extending abroad through the PCT multiplies the number several times over. Defensive publication is nearly free in cash, but its price is the permanent surrender of both rights — the largest opportunity cost of the three where the solution turns out to have high monopoly value.
Recommended actions
Combination is standard practice, not the exception: patent the solution layer that products reveal, keep secret the operating parameters and data products do not show, and defensively publish the side variants you will not exploit but do not want rivals to monopolise. The three layers must be decided in one review meeting before the first filing, because the first patent application’s disclosure decides the fate of every connected secret.
The concrete case of a production formula — secrecy versus patenting — is analysed in the published FDI-LC-4542 article on flaw.vn; this article supplies the general matrix for all R&D results, and the term of secrecy protection is deepened in EXP-2008 in this same batch.
Frequently Asked Questions
When to keep it secret, when to register?
Keep it secret where the solution does not leak through the product, the life cycle exceeds twenty years, and the company controls people and data flows. Patent where the solution reveals itself once the product ships, or where a monopoly is needed for fundraising, licensing or blocking rivals — including independent discoverers. Publish defensively where you do not intend to exploit a monopoly but want durable freedom to operate and to stop rivals taking a patent on the same solution. These three questions resolve most real cases.
Decision criteria?
Five criteria: reverse-engineering exposure from the product; commercial life against the twenty-year patent term in Clause 1 Article 93 of the IP Law; detectability of infringement — process patents are hard to detect and therefore often disadvantaged; whole-life cost of each route including the operating cost of secrecy; and the need to signal to markets and investors. Score each R&D result on the five and the winning route is usually clear; where scores tie, protect the option whose loss is irreversible and triggered earliest — the filing decision must close before any disclosure.
Cost of each option?
In ascending cash order: defensive publication is cheapest, costing little more than preparing the disclosure document, but pays by surrendering rights. Secrecy carries no State fees; its cost sits in control systems, undertakings and training — a continuous organisational cost. Patenting is the most expensive: State fees under the current industrial-property fee schedule, agent fees where used, annual maintenance across the twenty-year term, and translation and foreign filing costs for multi-country protection. Compare whole-life cost, not the first invoice.
Can they be combined?
Yes — that is standard practice. The most common combination: patent the visible solution core, keep parameters, data and operating know-how secret because products do not reveal them, and defensively publish side variants to block rivals. The condition is that layers are delineated before the first patent filing, because the disclosure in an application destroys the secrecy of everything connected put into it. The company also needs an internal clearance gate so no department can publish on its own — through a paper, a fair or a bid file — and collapse the others’ strategy.