Daniel sent us a prompt about patents — specifically, where Google Patents actually pulls its data from, what the best free tools are for keeping tabs on a technology like indoor active noise cancellation, and then a bigger question about how patent systems differ across countries and how they even came to exist in the first place. And I'll admit, I've clicked through to Google Patents maybe three times in my life and never once stopped to ask where the documents were coming from.
That's the thing — most people treat it like a search engine and never think about the plumbing. Google Patents is an aggregator, and it pulls from over a hundred patent offices worldwide. The big ones are the USPTO, the European Patent Office, the Japanese Patent Office, the Korean Intellectual Property Office, and China's CNIPA. But it also indexes WIPO's Patent Cooperation Treaty filings, which is where you file one international application that covers a hundred and fifty-plus member countries. So when you search Google Patents, you're not searching one database — you're searching a federated index of basically the entire world's patent literature going back in some cases to the seventeen nineties.
So somewhere in there is a patent for a better butter churn signed by someone in a powdered wig.
The first US patent was issued in seventeen ninety, signed by George Washington. But to answer the practical question — if someone wants to browse patents on indoor noise cancellation without paying anything, Google Patents is genuinely the best starting point. It's free, the full text is searchable, and the machine-translated patents from non-English offices are surprisingly readable now. The alternative is the USPTO's own Patent Public Search tool, which is also free but has what I'll generously call a learning curve. Then there's Espacenet from the European Patent Office, which has over a hundred and fifty million documents. Those are the three giants.
The assumption that you have to pay to read patents — that's just wrong?
Completely wrong, and I'm glad you flagged it because it's one of those things everyone assumes. Patents are public documents by design. The whole deal is: the government gives you a temporary monopoly — twenty years from the filing date — in exchange for you telling the world exactly how your invention works so that anyone can build on it after the patent expires. The publication is the quid pro quo. You pay to file and to maintain the patent, not to read it. The fees can be substantial — filing, search, examination, maintenance fees at three and a half, seven and a half, and eleven and a half years — but reading is free.
The competitor concern Daniel mentioned — someone buying up patents to diminish a head start — that's not about preventing people from reading the patent. That's about owning the rights so nobody can practice the invention.
And that's a real strategy, especially in tech. Companies build defensive patent portfolios specifically to deter lawsuits — mutually assured destruction with legal briefs. IBM has led the US in patents for something like twenty-nine consecutive years. They're not commercializing all of those. A lot of them are deterrents.
Mutually assured destruction with legal briefs. The lawyer equivalent of a cold war submarine.
About as expensive. But to loop back to Google Patents — one thing worth knowing is that it's not just a search tool. You can set up alerts. If you're tracking indoor active noise cancellation, you can create a saved search with the relevant classification codes and keywords, and Google will email you when new patents publish. The classification system is the secret weapon here. The Cooperative Patent Classification has a specific subclass for active noise cancellation — G10K eleven slash one seven eight — and if you search by that code rather than just keywords, you catch things you'd otherwise miss.
G10K eleven slash one seven eight. I'm writing that down. Or I would be if I weren't a sloth. That's a lot of characters.
It's the Dewey Decimal system for invention. And the CPC is harmonized across the US and European systems, so one code searches both. That's part of the answer to the country-by-country question — the classification systems have converged substantially over the past twenty years. But the patent systems themselves? Very different histories, very different philosophies.
Let's go there. Where did this whole idea come from?
The modern patent system traces back to Venice in fourteen seventy-four. The Venetian Senate passed a statute that said if you invent something new and useful, you register it with the city, and for ten years nobody else can copy it without your permission. That's the first codified patent law in the world. The word "patent" itself comes from the Latin "patere" — to lay open — because the monarch would issue "letters patent," which were open letters, not sealed, publicly declaring the grant.
So the name is literally "this is not a secret document.
And that's the philosophical core of the whole system. Then England picks it up with the Statute of Monopolies in sixteen twenty-four, which is where we get the idea that patents should be limited in time and only granted for new inventions — before that, the Crown was handing out monopolies on salt and starch and playing cards, basically as patronage. The Statute of Monopolies said no, only for actual inventions, and only for a limited term.
The Crown handing out the exclusive right to sell salt. That's the kind of governance that gets you a civil war.
So then the US Constitution, Article One, Section Eight, Clause Eight — "to promote the progress of science and useful arts, by securing for limited times to authors and inventors the exclusive right to their respective writings and discoveries." That's both copyright and patent in one sentence. The first Patent Act was seventeen ninety, and the first patent examiner system came in eighteen thirty-six. Before that, patents were just registered — nobody checked if they were actually new.
For forty-six years, the US was just taking people's word for it?
And the result was chaos — conflicting patents, obvious inventions getting approved, endless litigation. The eighteen thirty-six reform created the Patent Office as we know it, with examiners who actually search prior art and determine novelty. That's the system we still use. And it burned down in a fire that same year, destroying ten thousand patents.
Of course it did. The first real patent office burns down immediately. That's practically allegorical.
And by the way, the fire is why US patents are numbered starting from one after July eighteenth thirty-six — everything before that is an "X patent," reconstructed from copies held by the inventors themselves.
The X patents. That sounds like a rejected Marvel franchise.
There are about nine thousand nine hundred of them, and about two thousand eight hundred are still missing. So if you find an original X patent in your attic, the patent office would very much like to hear from you.
Alright, so that's the origin story. What about the country differences Daniel was asking about? If someone in Singapore and someone in the US are both working on indoor noise cancellation, does the patent process look different?
Yes and no. The format of a patent document is remarkably standardized worldwide — abstract, background, summary, detailed description, claims. The claims are the legally operative part; everything else is supporting material. That structure is basically universal now, largely because of international treaties harmonizing the requirements. But the actual examination process and what counts as patentable differs significantly.
Give me the fault lines.
The biggest one is software. The US has gone back and forth on software patents — the Supreme Court's Alice decision in twenty fourteen made it much harder to patent abstract ideas implemented on a computer, but people still file them and some get through. Europe is stricter: software "as such" is not patentable, but software that produces a "technical effect" might be. For indoor noise cancellation, that distinction matters because a lot of the recent innovation is in the signal processing algorithms, not the hardware.
If you invent a clever new algorithm for canceling street noise, you might get a patent in the US but not in Europe?
You might get it in both, but you'd have to frame it differently. In the US, you'd claim the method. In Europe, you'd claim the apparatus that implements the method. Same invention, different legal framing. Then there's China, which has become the world's largest patent filer — they passed the US in twenty eleven — but the system operates differently. Chinese patent law has two tiers: invention patents, which get substantive examination, and utility model patents, which are basically registered without examination. A utility model gives you a ten-year right, but it's much weaker because nobody checked if it was actually new.
China's numbers are inflated by these unexamined utility models?
In twenty twenty-three, China's CNIPA received about one point six million patent applications. The USPTO received about six hundred thousand. But the Chinese number includes a huge number of utility models. That doesn't mean there's no innovation — there's plenty — but the raw count isn't comparable.
The patent equivalent of counting clicks instead of purchases.
And then there's the enforcement question. A patent is only as good as your ability to enforce it. In the US, patent litigation is expensive and takes years. In Germany, patent cases are heard in specialized courts and can be resolved in twelve to eighteen months. That's why a lot of companies sue in Germany even when the defendant is based elsewhere — if you can block sales in the German market, you have leverage.
If I'm developing indoor ANC technology and I want to protect it globally, I'm not filing one patent. I'm filing a family of patents across multiple jurisdictions, each tailored to local rules.
You're probably starting with a PCT application — the Patent Cooperation Treaty. That gives you a single international filing that buys you thirty months to decide which countries you actually want to pursue. The PCT doesn't grant a patent — there's no such thing as an international patent — but it streamlines the initial process. After thirty months, you enter the "national phase" in each country you've selected, and each office does its own examination.
Which sounds expensive.
Filing a PCT application costs a few thousand dollars. Then translating it into each country's language, hiring local patent attorneys, paying examination fees — you can easily spend fifty to a hundred thousand dollars for a modest international portfolio. For a major pharmaceutical patent covering dozens of countries, the lifetime cost can run into the millions.
Which is why the system has a built-in bias toward large companies with deep legal departments.
That's the critique, yes. And it's a legitimate one. The Patent Pro Bono Program in the US tries to address this by matching solo inventors and small businesses with volunteer patent attorneys, and there are similar programs elsewhere. But the structural cost barrier is real.
Let's bring this back to the specific technology Daniel mentioned. Indoor active noise cancellation — there's been a flurry of patents in the last couple of years. What's actually in them?
The recent patent activity breaks into a few clusters. One big area is room-scale ANC — using multiple speakers and microphones distributed around a room to create quiet zones. Silentium, an Israeli company, has been filing in this space. They have a system that samples noise at a point and generates anti-noise through speakers placed around a room, creating a bubble of quiet around a bed or a desk. Another cluster is in automotive — canceling road noise inside the cabin. Bose has been big on this, and so has Harman. A third cluster is in windows — transparent ANC systems that cancel noise passing through glass. There's a really interesting patent from a group at Nanyang Technological University in Singapore from twenty twenty-two that places microphones on a window frame and actuators on the glass itself, turning the entire window into a noise-canceling surface.
The window becomes the speaker. That's elegant.
And it gets at why this problem has been hard since the nineteen thirties. Paul Lueg, a German engineer, filed the first ANC patent in nineteen thirty-four. His idea was to cancel sound in a duct using a microphone upstream and a speaker downstream, generating an inverted wave. It worked in theory but was practically impossible with nineteen-thirties analog electronics. The latency requirements are brutal.
Nineteen thirty-four. So people have been trying to cancel noise since before the jet engine was invented. They were trying to cancel noise they hadn't even heard yet.
The jet engine went into service in nineteen forty-four. Lueg was ten years ahead of the problem. But the core challenge is physics. Sound travels at about three hundred forty-three meters per second. If you're canceling a five hundred hertz tone, the wavelength is about sixty-nine centimeters. Your anti-noise has to arrive within a fraction of a wavelength — which means your system has to sense, process, and respond in under a millisecond. For lower frequencies, the wavelengths are longer, which makes the timing slightly more forgiving, but you're still dealing with sub-millisecond latency. In a room, it's even harder because sound reflects off walls, creating a complex three-dimensional field. Canceling at one point might amplify at another.
The nineteen-thirties idea was sound in principle and impossible in practice.
It stayed mostly theoretical until the nineteen-eighties, when digital signal processing chips got fast enough to do the math in real time. The first commercial ANC was in aviation headsets — David Clark and Bose — and then in the nineties you started seeing it in consumer headphones. But headphones are the easy case. Your ear is at a known, fixed location relative to the speaker. A room is a whole different problem.
Which is why the recent patent activity is interesting. It means the enabling technology has crossed a threshold.
Three things have changed. One, processors are fast enough to handle multi-channel ANC with adaptive algorithms that track changing noise fields. Two, MEMS microphones are cheap, tiny, and precise — you can sprinkle them around a room without spending a fortune. Three, machine learning lets systems predict noise patterns and pre-generate anti-noise rather than just reacting. That last one is where a lot of the recent patents are focused — using neural networks to model the acoustic transfer function of a room in real time.
Acoustic transfer function. That's the mathematical description of how sound bounces around a given space.
Every room is a filter. Sound goes in, reflects off walls, furniture, people, and what arrives at your ear is the original sound convolved with the room's impulse response. If you can model that response quickly enough, you can cancel the sound anywhere in the room, not just at a fixed point. That's the holy grail.
The patent filings suggest people think they're getting close.
I was looking through some of the recent ones — there's a Samsung patent from twenty twenty-three for a system that maps a room's geometry using a phone's camera and then optimizes ANC speaker placement based on the 3D model. There's an Apple patent for ANC embedded in a HomePod-like device that adapts its cancellation field based on where people are sitting, detected via ultra-wideband chips. The patents are getting very specific, very applied — that's usually a sign that products are coming.
The Apple one — using UWB to track where people are and steering the quiet zone toward them. That's the kind of thing that sounds like magic until you realize it's just a lot of math happening very fast.
That's the story of ANC as a field. The math was worked out in the thirties. The implementation has taken ninety years.
Let's circle back to the practical question. Daniel wanted to know how to keep an eye on this space using free tools. You mentioned Google Patents alerts and the CPC code. Is that the full answer, or are there other aggregators worth knowing about?
The Lens is an open-access patent and scholarly literature database that's really well designed — it's a nonprofit out of Australia, and it has nice analytics tools. PatentScope is WIPO's own search tool, good for tracking PCT applications specifically. And if you're really serious about a technology area, you'd want to use the USPTO's Patent Public Search for US filings and Espacenet for European ones, because they sometimes have documents that Google Patents hasn't indexed yet. Google is fast but not always perfectly up to date — there can be a lag of a few weeks.
The workflow would be: set up a Google Patents alert for the CPC code, maybe a second one on The Lens, and then check PatentScope periodically if you want to see PCT filings before they go national.
That's exactly what I'd recommend. And I'd add that if you're tracking a specific company, most patent offices let you search by assignee. So you can set up an alert for "Silentium" or "Bose" or "Harman" and see everything they're filing in the noise cancellation space. Companies can't hide patents — they're public by definition.
That's the part that still surprises people. The patent system is fundamentally a disclosure system. You tell the world your secret, and in exchange the government gives you a temporary right to exclude others from using it. The secrecy is the opposite of the point.
It's one of the oldest information-sharing institutions in the world. And it's weirdly democratic in that sense — anyone can read anyone else's patents. A teenager in Jakarta can pull up the latest Apple ANC patent and understand exactly how it works. Whether they can build it without infringing is a different question, but the knowledge is public.
The knowledge is public. That's worth sitting with for a second. In an era where everything is subscription-walled and paywalled and gated, the entire global library of human invention is free to read.
It really is. And the patent system as we know it — examination for novelty, limited term, public disclosure — is only about two hundred years old. Before that, inventors protected their inventions through secrecy. Guilds had trade secrets passed down through apprenticeships. The Venetians changed that by saying: tell us what you made, and we'll give you a legal monopoly for a limited time. It was a trade. And it worked so well that every industrializing country adopted some version of it.
The guild system. So before patents, the way you protected your invention was to never tell anyone outside the guild, and to make sure your apprentices were sworn to secrecy.
That worked for some things — glassmaking on Murano, for example. The Venetian glassmakers were forbidden to leave the island on pain of death. But it didn't scale. You can't have an industrial revolution based on trade secrets. You need publication, you need people building on each other's work, you need the patent literature as a cumulative record of technical knowledge.
The patent literature as a cumulative record. That's a nice way to think about it. Every patent is a little brick in a wall that's been building for five hundred years.
It's a wall that's now growing by something like three and a half million patents a year globally. The total stock is over a hundred and fifty million documents. It's the largest structured repository of technical knowledge in human history.
Which raises a question. If there are a hundred and fifty million patents, most of them are expired. The twenty-year term means everything filed before two thousand five or so is now in the public domain. That's an enormous library of free-to-use technology.
And that's an underappreciated resource. There are consultancies that specialize in mining expired patents for useful technology. In the pharmaceutical space, it's the entire basis of the generic drug industry — once the patent expires, anyone can manufacture the molecule. But in engineering fields, people often don't think to check expired patents. They reinvent things that were already solved in nineteen eighty-seven.
The wheel, repeatedly reinvented, patent number six million and something.
There have been multiple patents on new kinds of wheels filed in the last twenty years. The wheel is not a solved problem, apparently.
We're still iterating on the circle. That's either inspiring or deeply concerning.
I choose inspiring.
Of course you do. Alright, let's talk about what makes a patent a patent. The format question Daniel raised — he mentioned the painstaking detail, the technical diagrams. Is that universal?
A patent has to meet a legal standard called "enablement" — it has to describe the invention in enough detail that someone skilled in the field could make and use it. If you leave out a critical step, the patent can be invalidated. That's why patents are so detailed. They're not marketing documents; they're legal documents that have to stand up in court. The diagrams are part of that — they're not illustrations, they're disclosures. Every element in the drawing has a reference number, and every reference number is explained in the specification.
The patent drawing is a legal document, not an aesthetic one.
And patent drafters have developed a very specific visual language over the centuries. There are rules about line weights, shading, cross-hatching. The USPTO has a whole manual on patent drawing standards. It's why patents have that distinctive look — the numbered arrows, the exploded views, the cross-sections. It's not a style choice; it's a legal requirement.
The aesthetic of legal necessity. Which is its own kind of aesthetic.
It really is. And some patent drawings are beautiful. The original Wright brothers patent for the flying machine — the drawings are exquisite. Da Vinci's sketches for flying machines are basically patent drawings before patents existed. There's a visual lineage there.
Da Vinci as the pre-patent patent illustrator. He would have cleaned up in the Venetian system.
He would have. Although his habit of writing everything backward in mirror script might have caused issues with the enablement requirement.
The patent office does not accept submissions in mirrored Italian. Another barrier to entry for the Renaissance inventor.
The Renaissance was a tough regulatory environment.
Now: Hilbert's daily fun fact.
Hilbert: In the nineteen fifties, mathematics teachers in Mali used abacuses with beads made from translucent calabash seeds, which when held to the light revealed their internal fiber structure — a property teachers used to demonstrate optical density while teaching arithmetic.
I don't know what to do with that information.
Optical density and arithmetic, taught simultaneously through seed beads.
To pull this together — the patent system is a five-hundred-year-old information-sharing institution that most people don't realize is free to access, the recent ANC patents suggest we're on the cusp of room-scale noise cancellation becoming a consumer reality, and if you want to track a technology area, the tools are Google Patents, The Lens, Espacenet, and the USPTO's public search, with CPC code G10K eleven slash one seven eight for active noise cancellation specifically.
That's a solid summary. The one thing I'd add is that the patent system is under real strain right now. The volume of applications is growing faster than patent offices can examine them. The USPTO has a backlog of something like seven hundred thousand unexamined applications. Examination takes about two years on average, sometimes longer. There's a serious question about whether the system as designed in the nineteenth century can handle twenty-first-century volume.
AI is going to accelerate that. If everyone has a tool that can generate patent applications, the volume explodes further.
The USPTO has already had to issue guidance on AI-assisted inventions. The current rule is that an AI can't be listed as an inventor — only a natural person can. But AI can be used as a tool in the invention process. The line between "AI-assisted" and "AI-invented" is going to get blurry fast.
The patent system, designed for individual human inventors and guild secrets, colliding with machine-generated invention at industrial scale. That's a tension that's only going to intensify.
It connects back to what we were saying about indoor ANC. A lot of the recent patents use machine learning to optimize the noise cancellation algorithms. Is the AI the inventor of those specific filter configurations? The legal answer right now is no, but the philosophical answer is less clear.
The twenty-year monopoly on a machine's idea. That's a question for a whole other episode.
But for someone just wanting to browse patents and understand a technology space, none of that matters yet. The system works. The documents are there. The knowledge is public. And for a technology like indoor ANC that's been in development since nineteen thirty-four and might finally be about to break through, the patent record is the best way to see what's actually happening beneath the press releases.
The patent record as the ground truth of technological progress. I like that. Thanks to Hilbert Flumingtop for producing, as always. This has been My Weird Prompts. You can find us at myweirdprompts dot com. We'll be back next week.
