| Episodes | 94 |
| Mentions | 150 |
| Cited here | 39 |
| First — last | #1 — #729 |
| Top guests | Joe Long, Philip Frieden, Jeri Ellsworth |
| Related | intellectual property · open source hardware · venture capital |
A patent is a granted legal right over an invention that operates as a bundle of negative rights: it gives the holder no right to make or sell anything, only the ability to stop someone else from doing so.[420] The underlying bargain is publication in exchange for the right to exclude—the patent document is public and anyone may read it, but using what it describes exposes the user to suit.[203][246] A patent’s term runs about twenty years, after which the disclosed design is free to build.[103] Patent rights are national: a granted patent lets the holder block importation into the country that granted it and sue there, and confers nothing in countries where no corresponding patent was obtained.[321]
Nature and scope of the right
Unlike copyright, which arises automatically the moment a work is created, and trademark, which exists in an unregistered common-law form, a patent has no instantaneous or unregistered equivalent and must be applied for.[420] An idea as such is not patentable; what is claimed is a particular method of implementing it, which is why narrow implementation differences can each carry their own patent.[602] A patentable invention need not exist as hardware: an application can be granted on a conceptual design that was never constructed, the requirement to furnish a working model belonging to an earlier era of patent practice.[247]
Because a granted patent’s disclosure is public, describing or republishing what it teaches is not the infringing act; what the patent restricts is making and selling the claimed invention in competition with the holder.[246] A patent publishes the invention as described and claimed, but not the files needed to build the product—an open-source hardware release, by contrast, gives away the schematic, the PCB files and the bill of materials in full.[6]
Utility patents and design patents are separate instruments with separate terms, and both need checking before cloning a mechanical part. On the Lego Technic reproduction project, Jason Huggins verified that the design patents on the beams had expired before proceeding, while noting that expiry had not stopped the rights holder from suing.[369] Marking a product with expired patent numbers can expose the manufacturer to a penalty assessed per unit sold, reportedly on the order of a hundred dollars an item, which is why patent markings are removed once the underlying patents lapse; Fluke’s revised 87 multimeter dropped the patent numbers that appeared on the back of earlier units.[52]
Priority and proof of invention
Under the first-to-invent system the inventor carried the burden of proving the date of conception, which made contemporaneous witnessed and dated notebook entries essential.[178] Laboratory notebook discipline at defence contractors required every page to be written in pen, dated and signed, with unused space on a part-filled page struck through so that nothing could be inserted later and back-dated.[178] The United States subsequently moved from first-to-invent to first-to-file, so priority turns on the date an application reaches the patent office rather than on proof of when the invention was conceived, and the evidentiary demand on dated notebooks fell accordingly.[178]
Drafting and prosecution
The process begins with a disclosure meeting at which the inventor explains the invention to the attorney, and the inventor should be able to state what has been invented in a single sentence before drafting starts.[420] The attorney’s core task is translation of scope: the inventor describes the invention in engineering terms and the attorney drafts claims that capture it as broadly as the art allows, deliberately avoiding the specific vocabulary an engineer would use, so that the resulting rights cover more than the one embodiment.[420] Specific numeric limits written into a claim are what let a competitor design around it—a claim reciting 3.3 to 3.5 volts is escaped by building at 3.6 volts—which is why claim drafting avoids concrete quantities wherever possible.[420] When an examiner cites art that reads on a broad claim, prosecution proceeds by amending to the least restrictive limitation that still clears the art: faced with a reference showing twelve-volt levels, the claim is narrowed to “less than 12 volts” rather than to the applicant’s actual values.[420]
In a field worked for decades, such as computation or communications, the broad claims were taken long ago and what remains is infill—the tread pattern on a tyre rather than the car.[420] Applications drafted by the inventor without an attorney, known as pro se filings, are characterised by poor claims even when the underlying invention is sound, since claim scope is the part of the document that requires specialist skill.[420] Filing an application does not mean a patent will issue: Jeri Ellsworth had every application refused and described prosecution as long, drawn out and tedious.[52] Obviousness is the routine ground for refusal during prosecution and is difficult to argue against because anything looks obvious once it has been seen; Julia Truchsess, holder of twenty-two or twenty-three granted patents, described this as the recurring friction of the process.[424]
Where the underlying technique is long known, a patent is obtained by scoping the claims to a specific application: a patent pending on transferring data to a device by flashing light was directed at configuring Wi-Fi credentials, because sending data with modulated light was not new.[202] A common filing strategy is to take a single underlying idea and file on many variant implementations of it, so that coverage extends beyond the one embodiment actually built.[283] Large semiconductor companies apply internal pressure to generate patentable work but screen disclosures before filing, with committees reviewing proposed applications rather than filing everything submitted.[270] In the Australian system as described by Paul Thompson, an initial filing made when an idea looks promising opens a twelve-month window in which the full application must follow, so the first filing buys a period of protected secrecy rather than granting rights outright.[481]
Invention-promotion firms that advertise to market an inventor’s idea make their money by signing the inventor up for expensive patent filings and taking a large share of any proceeds, rather than by selling the invention.[1]
Prior art and defensive publication
Publishing a design is put forward as an alternative to filing on it: once the work is public it stands as prior art, making it harder for anyone else to obtain a patent over the same ground.[143] Defensive-publication repositories exist specifically so that an idea can be placed on the public record as prior art, which is more reliable than relying on a personal blog post to establish the date and the disclosure.[143] Defensive publication secures no rights for the publisher—it cannot be converted into a patent later—and its only function is to keep the ground clear so that others cannot obtain one.[143][420] Prior art on the public record does not prevent a suit from being filed; the defence still has to be argued.[143] Conversely, deliberately not searching for patents does not protect a designer: not knowing that a patent covers a standard connector arrangement is not a defence to infringement.[178]
Costs and enforcement
The commonly quoted figures are about ten thousand dollars to obtain a patent and about a hundred thousand to defend it the first time, with the patent only demonstrating its value once it has survived that defence.[232] A ten-thousand-dollar retainer quoted up front is used by patent firms as a filter on individual inventors, and it is small against what prosecuting an application actually costs.[420] Experienced patent attorneys prefer corporate clients that file in volume and avoid independent inventors, who demand attention and are the source of malpractice claims.[420] Pursuing a patent as an individual inventor can consume hundreds of thousands of dollars without producing a return: an Australian inventor of a mains outlet with a built-in USB socket is described as having spent that much, and the idea was subsequently built by everyone anyway.[602]
Assertion of a patent works largely as a threat: the holder sends a lawyer’s letter and hopes the recipient backs down, because a recipient with money to fight can refuse and the holder may not be able to afford the case.[242] Holding a granted patent is not by itself decisive; the outcome turns heavily on how much each side can afford to spend on counsel, and anyone with funds and motive can bring a suit however weak, ruining the defendant with the cost of defending it before a court reaches the merits.[178][203] A patent holder asserting a stereolithography patent sued not only Formlabs, the company building the printer, but also Kickstarter, the crowdfunding platform selling it, on the theory that the sales channel was aiding the infringement.[123]
The cost of enforcement puts patents out of practical reach well above hobbyist scale: Dave Jones judged that even running a ten-million-dollar company he could not afford to sue a copier, and the case in which a patent pays off for a small engineering business is a startup being acquired by a larger company rather than protection of a product in the market.[347] For a one-person business the economics do not close at all: Ryan O’Hara abandoned patenting in favour of publishing designs openly and staying ahead by continuing to release new products.[153]
Venue and jurisdiction
United States patent suits at one time concentrated in a single federal district in Texas because its local procedural rules favoured plaintiffs, and entities asserting patents established a presence there.[260] A subsequent Supreme Court ruling reaffirmed a 1957 decision holding that the Patent Act requires a patent suit to be brought where the defendant company is incorporated, overriding the broader general venue rules and requiring a genuine corporate presence in a district before a suit can be raised there.[347]
On the account of Alex Lidow, whose gallium nitride device company asserted its patents against a state-funded Chinese competitor in disputes running simultaneously in several countries, the United States is the hardest jurisdiction in which to enforce a patent, with China, Germany and Japan all easier—an ordering described as having reversed over the previous fifteen to twenty years.[719] The cost of United States enforcement comes from repeated challenge in parallel forums: a win at the International Trade Commission can be followed by proceedings at the Patent Trial and Appeal Board, in the customs office and in federal court, each with its own appeals, with no bar on bringing the same challenge again.[719]
Portfolios and corporate strategy
Patent portfolios are themselves a principal asset in large acquisitions: Larry Page stated that Google’s purchase of Motorola was substantially about acquiring its patents, a holding of roughly sixteen thousand.[56] Large-company patenting is driven by aggregate portfolio weight rather than by the value of any individual invention; the relevant comparison is how many patents each side holds.[144] A portfolio functions simultaneously as defence and offence: holdings are trading stock in a dispute, and once a portfolio is large enough the holder can initiate assertions rather than only answer them.[129] Brett Fox’s stated reason for a semiconductor company holding patents at all is that it should expect to be sued for infringement eventually regardless of whether it set out to infringe anything.[129] Patent fights between comparably resourced competitors tend not to end in exclusion: Xilinx and Altera litigated repeatedly and settled by cross-licensing their portfolios once the legal spend outweighed the benefit.[103] Philip Freidin, who held roughly two dozen patents assigned to AMD and Xilinx, received a commemorative plaque for each one; recognition rather than payment is the usual reward attached to inventions assigned to a large employer.[103]
One route out of federally funded university research into a company is to file first and license second: the founders of Cree wrote ten patents covering the silicon carbide technology they had invented, then asked North Carolina State University for an exclusive licence in exchange for stock in the new company, which the university cleared with the funding agency.[71] On John Edmond’s account, patents were only part of what kept silicon carbide LED manufacturing to a single company; the larger part was vertical integration, starting the company around both silicon carbide LEDs and silicon carbide electronics so that no one else could produce the material.[71]
A semiconductor patent has value only if somebody will take a licence to it, which frequently means going after a company and entering litigation to get there; applications are sometimes filed with no development done and no demonstration that the concept works.[297] In a downturn or on the way out of business, a company may hand its patent portfolio to a venture capital firm, which then asserts those patents against companies that are making products to recover its money, sometimes on legitimate grounds and sometimes not.[297]
Filing rather than keeping a technique secret is a real tradeoff in silicon design: a patent publishes how the result was achieved, letting competitors design around it, so one school of thought inside semiconductor companies holds that a circuit trick is better hidden in the die and simply shipped.[270]
Patent thickets shape component markets. Off-the-shelf conductive inks formulated for inkjet deposition are heavily patented, which restricts how far a company using them can develop its own formulation, and they are priced at the cost of bulk silver plus processing.[260] MEMS ultrasound transducers are held under patents by a small number of companies and laboratories and cannot be bought in ones and twos by an engineer wanting to build a kit around them.[448] A patent filed on the self-balancing hoverboard was weak against prior art because the Shenzhen supply ecosystem was already mass-producing every part needed, including self-balancing unicycles and electric scooters; the combination, not any new development, was the only novelty claimed.[336]
Patents, startups and investors
Venture investors assess a hardware startup partly on what barrier surrounds its technology, and a patent portfolio contributes to that assessment alongside content kept as trade secret inside an ASIC or FPGA.[147] On Jeri Ellsworth’s account, a funded startup may file principally to satisfy investors rather than to protect anything: the count of patents held is among the first questions asked in a venture round, so applications are filed to make the company attractive.[173] Among small companies, the practical purpose of filing is value signalling to investors—a portfolio demonstrates that the company has invented specific things, and is filed to support fundraising or an acquisition rather than to be enforced.[420] For a company raising twenty to eighty million dollars against a steep burn rate, spending on the order of a hundred thousand dollars on a patent portfolio is proportionate, because the outcome is binary: if the company succeeds it needs the patents, and if it fails it had nothing to lose anyway.[420] Patents brought into a company are assigned to it as a condition of taking investment; an inventor who received patents from a former employer and then raised capital no longer owned them, holding shares in the company that did.[351]
Open-source hardware makes some venture investors uneasy, and the question that follows is what patent protection and what portfolio the company has.[327] Not every investor treats a patent position as a selling point: Avidan Ross pushes back on founders who lead with their patents, on the ground that the money is meant to fund building a product rather than litigation.[327] Dave Vandenbout, on the account of a founder who ran several failed startups, likewise advises that a startup should not place its emphasis on intellectual property or assume a strong patent position will keep competitors out; value is the idea multiplied by the execution, and a good idea executed poorly is worth little.[232] Greg Charvat’s advice to a startup with a successful product is to expect competitors to weigh buying it against copying it, and that a patent alone is only a document without the legal team to enforce it; the durable barrier in his company’s case was building the sensor in a silicon process at fine pitch and volume.[729]
For an inventor intending to license rather than manufacture, the first question a prospective licensee asks is whether there is a patent, so one is generally needed before negotiation starts.[424] The licensee’s reasoning is its own development spend: it will not commit something like half a million dollars to bringing an easily copied product to market with no protection against immediate knock-offs.[424] A patent strengthens the inventor’s negotiating position in a licensing deal and, once the product is on the market, can deter competitors and knock-off manufacturers from entering, though the protection delivered in practice is limited because there are frequently ways to design around the claims.[424]
Bunnie Huang’s account of selling an open-source laptop is that a product can be monetised for long enough to fund the next generation without strong patent or copyright protection, at the cost of having to stay competitive rather than sitting on a granted monopoly.[336]
Patent literature as a technical resource
The disclosure obligation is what the public receives in return for the grant, and an engineer working in a narrow technical field should be reading the patents in that field—with the caveat that some are written strangely, some do not mean what they appear to say, and some are simply wrong.[420] Patent literature works as a process-engineering reference: Ben Krasnow reproduced coffee decaffeination by reading the relevant patents through Google Patents, where the process is set out in detail, although patent prose is generally hard to read.[75] Older Texas Instruments patents disclose far more than the claimed invention, including complete schematics for a calculator chip and the code it ran, making the back catalogue a usable trove for reverse-engineering old silicon.[361] Press and trade-journal articles reporting that a company has patented something are usually wrong in some subtlety, since patent documents are confusing to non-specialists, and headlines about them should not be taken at face value.[420]
Patents and open hardware
For a hardware design published openly, a registered trademark on the product name is the only protection that remains available: it stops a competitor using the name, but it does not stop them copying the printed circuit board or the rest of the design.[6] Because the patent bargain is publication in exchange for exclusion, a company that announces it is releasing its patents is committing not to sue rather than disclosing anything that was not already published.[203]