Patents have suffered a fate similar to something like metrics that have succumbed to Goodhart's law.

Patents could've worked somewhat originally, at least in principle, but soon people started gaming the system, rendering the patent system worthless. Except for maybe certain domains patents do not currently host important inventions in the meaningful sense. Especially in software, execution matters much more than the actual idea.

And the execution part usually is akin to a trade secret: for example, it's not so much what Google and Amazon do, it's about the mountain of work they did in building all that and also making it scale. That's the non-trivial part that you couldn't replicate even if you read and bought all their patents.

A low-hanging fruit of fixing the patent system would be to reduce it to a mere notary service of timestamping ideas. Establishing the patent de facto would require the patent to be defended in courts: this alone would cut down the amount of bogus patents.

Defending a complex, vague system would cost a lot of money, balancing the usefulness of gaining a court-verified monopoly on the idea. On the other hand, a simple but truly novel idea would be rather easy to defend, even for a private person. There's no downgrade to the current state of matters: it's not as if the current patent system would allow individual inventors a low-cost way to sue and win big companies trying to abuse their patents already.

There are other ways but in any case gains need to be balanced with cost, that's the only way to make people and companies prioritise and valuate the true worth of their ideas. No party is going to spend a lot of money trying to defend a vague patent that was written only to win the first-to-file race and game its coverage within the patent system. They already know it's bollocks.

>A low-hanging fruit of fixing the patent system would be to reduce it to a mere notary service of timestamping ideas. //

The law says you can't get a patent for an idea. Which is right, ideas are cheap.

It sounds more like you're problem is "insufficiency" of the disclosures. No need to throw the computer away because one stick of RAM doesn't work.

> A patented invention should be described in sufficient detail, and the description is made public. This is how important inventions do not get hidden form sight forever, even if they don't bring any profit.

And the problem is that the patents of today do their best to be as vague as possible (in order to maximize the "area under patent"), while either being obvious or not revealing sufficient details for one to recreate the patented invention from the patent. Not to mention that patents lengths are not in line with exponential rate of progress.

So if they already fail at the functions they were created for, how worse are they from trade secrets? Maybe what we should do is not only get rid of patents, but also reduce protections around trade secrets? People will keep trying to make money anyway, this will still lead to inventions, but that way, at least society will get to make use of those inventions.

(Drug research is an exception I'm not sure how to handle, though.)

Yeah, a patent should only be granted if the patent officer manages to build a working prototype using it as a guide. A competing patent officer should only get the title of the patent and try the same, to check for obviousness.

I partially agree. It's not feasible to expect the patent examiner to reduce a technology to practice, primarily because a lot of physical technology requires very expensive equipment and some skill to master (chemical vapor deposition chambers for thin films for instance). The ownus should be on the applicant. However, it is certainly feasible for an inventor to demonstrate a working prototype for each claim. Based on the number of rediculous awarded patents I've reviewed over the years, I imagine simply requiring demonstration will probably reduce the number of patent applications by 90%.

This requirment would have many benefits. However, a down side is that not all entities have the resources to build their inventions. If I come up with a new idea for an efficient rocket engine, I would have to license the idea to a SpaceX, etc, to get it built. But it would be difficult to approach a SpaceX and secure a license without first getting a patent.

You will need the resources to build it eventually. You can easily protect yourself from an individual entity stealing your ideas using a Non-Disclosure Agreement (NDA).

Not always, but in general, a proof of concept prototype (built in parts, at small scale, or simulated) will be an order of magnitude cheaper than a production unit.

Would a white paper for a free energy generator be accepted to a respected peer reviewed journal? Probably not, but there are tons of these truely worthless granted patents out there diluting the system. I strongly believe we need inventors to prove their claims via demonstration, one way or another.

Drug research seems more like an infrastructure thing more than a profit thing. Fund the research publicly. Academic competition is enough to ensure research is effective.

Drug manufacture however seems more like a profit thing. Open up the results for companies to make generics. Profit motive will ensure efficient delivery.

Drugs are rarely (if ever?) developed by academics, and cost a huge amount. Low estimates for bringing a new drug to market are around $500M+:

They found that companies took an average of 7.3 years to win FDA approval, at a median cost of $648 million. Only two drugs had research costs over $1 billion. Adding in the cost of capital at 7 percent increased the median research and development cost to $757 million—less than a third of the Tufts estimate.[1]

[1] https://www.theatlantic.com/health/archive/2019/03/drug-pric...

I agree that the pharmaceutical industry spends a lot researching drugs. And so does the government eg [1].

In the situation of an imaginary future where patents are scrapped, pharmaceutical companies will be unlikely to continue to invest in the current fashion. I took GP to suggest that this would be an issue. I was trying to make the point that tax payers through something like NIH grants would be to take up the slack. Making the research available patent free so that generics can be manufactured.

Obviously this wouldn't be a simple transition and incredibly unlikely. But its an interesting intellectual exercise to think about.

[1](https://www.statnews.com/2018/02/12/nih-funding-drug-develop...)

You also need to add the costs of the losers.

Indeed, the next paragraph in that article discusses that.

Except that idea is simply historically untrue. Patent were started primarily as a way to modernize the guild monopolies that were common place at the time, but were unsuited given the greater levels of globalization and international trade.

The idea that they were meant to serve as a way to reduce trade secrets falls flat in the face of the fact that most places with strong patent and copyright law, also have strong laws protecting trade secrets

You are correct when you are talking about the origin of British patent law. There a king could grant a patent for a wide range of monopolies, including practicing a particular craft only by guild members, etc.

The US patent law, which I referred to, is not as old, and it definitely talked only about inventions, and explicitly required drawings, explanations, etc to be made available: [1]

[1]: https://en.wikisource.org/wiki/Page%3AUnited_States_Statutes...

Given that american patent law was created originally as a copy of british patent law, through the statute of Anne, that distinction is irrelevant or even misleading.

The idea of exponentially increasing patent costs has a major issue: not all businesses are alike. If I make a new model of fork and patent it, I can probably enter the market quickly. However, if my patent is for some new type of power plant, I will not be able to enter market within 20 years.

Refutal: raise the cost differently depending on time to enter the market?

Then some statistician will have to be paid to estimate this, meaning patent applications are much more expensive.

I hope you think my comment was constructive.

But the people preparing to build a power plant, drawing up plans say, or doing tooling, would be doing "contributory infringement". The bank that funded it would be co-liable, which is great for the patent holder if they want a payday.

isnt this already an issue with the current system too? Patents all have the same minimum cost and same minimum effective legal cost, even though some opportunities are large and some are small. I understand your criticism but I think, if properly fleshed out and part of a larger plan, some sort of exponential cost feature could be useful.

That's why SpaceX does not file any patents. It would be like handing its recipe book to its competitors.

One of the most sensible and workable solutions I have heard to the cluster-fuck otherwise known as the patent system.. and this is why nothing will ever become of it.

> Please remember that patents were invented as a way to fight trade secrets.

Does anything in the early history of patents backs the assertion that they were invented for that reason?

"Patents might be useful as a way to fight trade secrets", is not the same thing, and could even be true enough; I think "patents were invented because..." might be a back-port of more recent thinking on to a history that does not support it.

>Please remember that patents were invented as a way to fight trade secrets.

Its becoming less and less and issue these days. Top secret product frequently leaks. Reverse engineering becoming better and better. Good luck keeping thing secrets, the cost will be higher and higher.

But we still have draconian protection for trade secrets. We still have products on the shelves of the grocery store with secret formulas. If patents are a substitute, then why not abolish the legal framework for protecting trade secrets?

It's hard to limit trade secrets without severely limiting the freedom of contract.

You can still get an FDA approval for secret-formula products; a good example is Coca-Cola.

With gas chromatography and mass spectrometry, is it possible to have a secret formula these days?

Yes, those tell you what the end product is, not the actual recipe to get that. And they tell you limited information, since they tend to decompose complex compounds that are unstable when vaporized.

What does criminal law have to do with "freedom of contract"? This is why "libertarian" has become a dirty word for a lot of people who used to be sympathetic.

You can be sent to prison for 15 years for trade secret infringement. It is criminal under both federal and most states' laws.

https://www.nolo.com/legal-encyclopedia/trade-secret-basics-...