Notes on filing a patent
A pencil and an eraser come up often as the stock example when people explain patents. The pencil is already known and so is the eraser, so you might assume that sticking an eraser on the end of a pencil counts as new and therefore gets granted. Patent law says otherwise. Simply combining two known technologies may be new, but it is not treated as an inventive step. A patent covers technology that is both new and inventive, and that distinction is what matters when you file.
So what makes a good patent? The most important thing is finding the problem that anyone building this technology would unavoidably have to solve. Say you have feature A and feature B. If A+B is hard to get granted, think about registering A+B+(C) instead, where C is the piece that cannot be avoided when solving the A+B problem. Finding that C is very hard, and it does not turn up overnight. It takes a long stretch of research.
Once a product exists, the best protection comes from finding and registering as many of those unavoidable problems as you can. Leave a few elements out and competitors can walk into the market through the gap. If knowing every angle up front is too hard, an easier route is to read similar patents and work out which problem drove each of them.
Patents earn bonus points on government R&D applications, so a lot of people file for that reason. A patent filed for the sake of having a patent is genuinely easy to register, and equally meaningless. Filing a patent meant to keep a third party out of the market takes at least six months, counting only the drafting time. Sometimes six months of work still does not end in a granted patent. If you are a startup and the cost is a burden, going through a government program, such as the Seoul Intellectual Property Center in Seoul, is one way to spend less. Patent R&D programs run every year, so they are worth looking up.
One common misunderstanding is that patents split into business model (BM) patents and ordinary patents. They do not. There is one patent system, and underneath it sit categories such as mechanical device patents, BM patents, pharmaceutical patents, and medical device patents. It is not only BM patents that can be filed without a product. Any of them can. What matters in a patent is the idea rather than the working implementation. Anything inventive and new can be registered.
Samsung is one example. Twenty years ago, before touch displays existed, it filed a patent on a method for moving content between two transparent display panels. An idea alone becomes a patent even when the technology does not exist yet. The legal term for that is a technical idea. When such a filing is examined, examiners search for prior art and may reject it, and they sometimes cite the content of films. That counts as a disclosed idea and blocks registration. Turn that around and you get a trick people use in reverse: publishing an idea on a website early to stop someone else from patenting it. The catch is that whoever published it cannot register it either.
Once a patent is granted, nothing can be edited, including the claims. Settled content stays settled. Because examination takes a long time, the granted right can end up meaning little by the time the market has moved on. A divisional application is one way around that. File a divisional and you can amend at any time, as long as you stay within what the specification already describes. Since nobody knows how the claims will change later, anyone entering the market has a hard time designing around it. If you want a strong patent, filing a divisional is worth doing without exception. Think of it as keeping the right unsettled so its content can still change.
It also costs less, because you do not have to register every claim to cover the unavoidable problem. Registering a single specification is enough. One application legally protects one invention, so only one goes in. But with a divisional, if someone else runs a business on similar ground, you can register that technology and secure it first, then file another divisional off the existing patent to keep the right alive. Competitors trying to enter without a patent of their own cannot tell how to respond, which blocks entry effectively.
During examination, prior art searches legally cover the whole world, though in practice examiners lean on the Big 4: Korea, Japan, the United States, and Europe. China and Russia come up in some cases, but searching worldwide for prior art is the baseline.