Why Most Invention Ideas Never Reach an Application
Ideas seldom fail at the moment of invention. They fail at four checkpoints that quietly disqualify a project months before anyone involved admits it has stopped.
For every invention that becomes a filed application, a large number of comparable ideas stop somewhere earlier — and almost none of them stop because someone decided they were bad. They stop because a step was taken out of order, a deadline passed without being noticed, or a question that should have been answered in week two was left until month ten, by which point the answer no longer mattered.
That is a more useful way to think about the problem than the folklore version, in which good ideas win and weak ones fall away. Attrition in invention is procedural. It happens at four recognisable checkpoints: the point where a concept has to become a specification, the point where the idea is first described to someone outside the project, the point where the existing body of technical literature is finally examined, and the point where costs step up rather than rise gently. A project can survive being unremarkable. It rarely survives arriving at one of those four checkpoints unprepared.
None of this requires legal expertise to understand. It requires knowing what the system is actually built to reward, which is not novelty of imagination but precision of description. Before any of the checkpoints make sense, it is worth being clear about what a patent is and what it is not: a time-limited right to exclude others from doing something specific, granted in exchange for teaching the public how to do it.
Three terms used throughout
- Prior art
- Everything already available to the public before your filing date — granted patents, abandoned applications, products on sale, academic papers, expired rights and forum posts alike.
- Enablement
- The requirement that the description be detailed enough for a competent person in the field to build the invention without inventing anything further themselves.
- Claim
- The numbered sentence at the end of an application that defines the legal boundary. Everything before it is explanation; only this is the property.
Attrition One · The Specification
A Concept Is Not Yet an Invention
The first checkpoint arrives the moment someone asks how the thing works. An idea stated as an outcome — a bottle that never spills, a fastening that cannot come loose, a sensor that predicts a fault before it happens — is a description of a wish. It cannot be owned by anyone, and no drafting skill can rescue it, because there is nothing there to enable.
What survives this checkpoint is a mechanism: named parts, a stated relationship between them, and a causal account of why the arrangement produces the result. If two engineers reading the description independently would build broadly the same object, the specification exists. If they would build two different objects, it does not, and any application drafted from it will either be rejected as insufficiently disclosed or granted so narrowly that it protects a single embodiment nobody wants to copy.
This is where most projects quietly end, and it rarely feels like ending. It feels like waiting for time to think. The productive move is the opposite of brainstorming: stop generating variants and force the current one into dimensions, materials, tolerances and failure modes. Inventors often find the exercise clarifying in an unexpected direction — the mechanism they can actually specify turns out to be a narrower, more interesting invention than the one they set out with. Imagination is not the constraint here; several accounts of how speculative fiction feeds the imagination of working inventors make the point that ideas are abundant. Specification is the scarce activity.
The system does not reward the best idea in the room. It rewards the idea that was written down precisely enough to be built from the page.
Checkpoint one, in a sentence
Attrition Two · The Disclosure
The Conversation That Closed the Door
The second checkpoint is the least forgiving, because it is passed without noticing. Patent rights everywhere depend on the invention being new at the filing date, and the test is measured against what the public could access — not against what competitors actually knew. A demonstration at a trade event, a crowdfunding page, a detailed post in a hobbyist forum, a product sold to a single customer: each can place the invention into the prior art against its own inventor.
Jurisdictions differ in how much mercy they extend. Some allow a grace period of roughly twelve months after the inventor's own disclosure, during which an application can still be filed. Many operate an absolute novelty standard with no grace period at all, so rights in those territories are lost on the day of publication. An inventor who discloses first and files second has therefore not lost everything — but has usually lost the territories that would have mattered to a licensee, and licensees notice.
The defence is procedural and cheap. Keep dated records of the design as it develops rather than overwriting them. Put a written confidentiality agreement in place before technical detail passes to a manufacturer, a prototyper or a potential partner, and note that an agreement signed after the meeting protects nothing said during it. Where a disclosure genuinely cannot wait, file first — even a modest early filing establishes a date that the later, fuller application can rely on.
Attrition Three · The Prior Art
The Search Nobody Ran Until It Was Too Late
The third checkpoint is the one inventors most often postpone, because it is the only step whose most likely outcome is bad news. Searching means reading granted rights, published applications that never granted, expired rights whose teaching is now free for anyone to use, product literature and technical papers — and finding, more often than not, that some part of the idea is already described somewhere.
Two misunderstandings make that discovery more damaging than it needs to be. The first is treating a hit as a verdict. Almost nothing is wholly unprecedented; what matters is whether this specific combination, applied to this problem, in this way, is already taught. The second is ignoring expired rights. A patent that lapsed decades ago is as fatal to a claim as a live competitor's filing, and it is the category first-time searchers skip most reliably, because an expired document does not feel like a threat.
Run properly and early, the search does its real work by redirecting rather than stopping. A recurring pattern: the headline idea is anticipated, but a detail the inventor considered incidental — the retention geometry, the sequence of steps, the way two known components are coupled — turns out to be unclaimed. Redirection at this stage costs a search fee and a fortnight. The same redirection after an examiner's rejection costs a year of prosecution and the drafting already paid for. This is the point at which most people stop working alone, and where practical help with patenting invention ideas changes the outcome most, because the decisions taken here — what to claim, what to abandon, whether to file at all — set the ceiling on everything that follows. The tests an examiner will eventually apply are worth understanding first-hand, and the standards of novelty and inventive step decide more applications than any other factor.
Attrition Four · The Cost Curve
Costs Arrive in Steps, Not a Slope
The fourth checkpoint is financial, and it defeats projects that have already cleared the technical hurdles. Patent expenditure is not a gentle upward line; it is a staircase, and the steps fall in places that are easy to forget once enthusiasm has moved on to prototyping.
The shape is consistent across territories even where the amounts are not. A first filing is comparatively modest. Twelve months later, if an early placeholder filing was used, the substantive application falls due — a step change, since it must be drafted properly and its claims will define the right. At around eighteen months from the earliest date, most offices publish the application, and the invention becomes public whether or not it is ever granted. Somewhere between eighteen and thirty months, examination begins, and the first response is usually a rejection; the exchange of arguments and amendments that follows is itself billable work. If protection is wanted beyond the first territory, that decision typically has to be made and funded around the one-year mark, or deferred by a further eighteen months through a single international filing — after which translation and national fees arrive together, several at once.
Then the staircase continues past grant. Renewal or maintenance fees fall due at intervals across a term that runs roughly twenty years from the filing date, and they rise as the patent ages. A granted patent kept alive on a product nobody sells is a recurring cost with no offsetting revenue, which is why the commercial question belongs beside the legal one rather than after it. Deciding whether an invention is worth filing at all is not defeatism; it is the step that keeps the budget available for the ideas that will repay it.
File first and talk second. Nearly every irrecoverable mistake in independent invention is those two steps in the wrong order.
The rule worth memorising
What Survives
What the Projects That Get Through Have in Common
Set the survivors side by side and the shared traits are unglamorous. They wrote the mechanism down before they showed it to anyone. They searched before they drafted, and treated the results as direction rather than judgment. They knew which step of the cost staircase was next and when it fell due. And they decided early what the patent was for — a licence negotiation, a manufacturing barrier, an asset on a balance sheet — because a right with no intended use is difficult to fund past its second renewal.
Just as telling is what they did not have. They were not, on the whole, better ideas. Long-running accounts of independent inventing, including this profile of the company's president, keep returning to the same observation: the people who reach a shelf are usually the ones who were methodical during the dull months, not the ones who were brilliant on the first afternoon. The idea is the cheap part of the process, and treating it as the valuable part is the error underneath all four checkpoints.
Finally, the survivors are honest about the cases where the answer is no. Some inventions are better protected by keeping the method unpublished; some are better served by publishing defensively so that nobody else can claim them; some have a commercial life so short that a right arriving three years from now is irrelevant. Knowing when a patent is the wrong instrument is part of the same discipline as knowing how to file one, and it is the difference between a portfolio and a collection of expenses.
Attrition in invention is not a filter for quality. It is a filter for sequence — and sequence, unlike inspiration, is entirely within the inventor's control.
End of report