Hacker News Reader

> "First-to-invent / first-to-file makes no sense, as it seems that if two people invented the same thing at the same time, that should set a strong presumption of "obvious." If two people file for the same patent, unless one is very clearly first in all respects, each should act as prior art for the other. "

I think this is too much of a simplification. Imagine the case of blue LEDs. It was known that these would have a huge market, and there were probably 100 different research groups around the world spending millions of dollars and thousands of hours trying to determine how to make them commercially successfully - even with some early weak prototype and research papers. That's clearly not 'obvious'.

In our history, one team (well, really an individual) managed to perfect the process first, but it could easily have been two or more teams randomly hitting the same solution at almost the same time.

I'm not sure what the 'right' thing to do in that circumstance would be - award it to the team who happened to try 'x' first, the team who filed first, or shared somehow.

But my point is that simultaneous does not necessarily imply obvious.

In real life, the inventor of the blue led was basically scammed by his company. The company earned a lot of money and basically shifted the whole business to produce blue leds. Veritasium recently made a video about that https://www.youtube.com/watch?v=AF8d72mA41M

I think this illustrate another point: patents should protect inventors, not companies. Companies shouldn't be able to amass a "portfolio" of patents to bully their competition either.

I like the idea of making sure the inventor gets some portion of royalties.

To be honest, it'd probably help my employers too. If that was the case, I'd have dozens of patents under my belt. As is, I see little upside to applying for patents for my employer.

My general feeling is that the price of a patent application should grow in some non-linear way with the number of patents. Having a few, targeted, focused patents should be cheap. Google has around 100,000 patents (source: random web site; could be wrong; if it is, swap out for any other big company and big number).

For reference: Patent office fees for various services are currently $100-$1000. Legal fees are probably $20k-$100k. The patent examiner has very little time and capacity to do a good job, and there is a resource mismatch. In abstract, an alternative would be a smooth rise in costs along the lines of:

- First patent is $50

- Tenth patent is $500

At this point, moving from individual-scale to corporate-scale:

- Hundredth patent is $50,000 (so about equal resources to both sides)

- Thousandth patent is $200,000 (subsidizing the individual patents)

- Ten thousandth patent on is $1M to file

In practice, this hides a lot of complexity. I think the biggest problem I can't think for how to solve (suggestions welcome!) is the corporate shell game. Google sets up a thousand subsidiaries, each of which files 100 patents.

Perhaps a simpler alternative:

- Individual personally filing a patent: $50

- Individual with lawyer / LLC / S-Corp / small business: $5000

- Private C-corp: $50,000

- Public C-Corp: $100k

Would you imagine that you could you still buy/sell patents?

In which case - Google, rather than applying for patents itself, offers a patent purchase programme for employees. It will buy every patent filed for by it's employees as "Individual personally filing a patent" for 1000 USD.

Plus a $10k transfer fee and a $100k patent conversion fee to change a personal patent to a C-corp patent.

Transfer fees are not a bad idea in either case. If a company goes under, it often leaves a portfolio of patents which get auctioned off to the highest bidder. There should be some minimum bar there.

> 3) Change patent duration by domain. 20 year software patents make no sense (5 year ones seem okay). 5 year patents on e.g. aerospace seem to make no sense (20 year ones seem okay).

One other idea would be to require a working prototype to be demonstrated as part of the filing otherwise the duration of the patent would be halved.

> Patents worked fine 100 years ago, when there were fewer of them.

I don't think it's about the number of patents. In the distant past, patents were intended to allow individual inventors to monetize their invention for some time by selling licenses to companies.

Today, patents are mostly used as a weapon in corporate warfare and by patent trolls to extort money from companies, both a complete perversion of the original idea.

Not sure if it is possible to return to (and enforce) this original intent of protecting the individual inventor, but if not we should just get rid of the whole broken patent system in general.

You can: increase the price for patenting something. Google will think twice before spending 100K for hello world. Patent trolls you'll always have, but you need to reduce them.

The system has to become again "healthy" to foster real innovation and competition, not a garbage bin to block each others.

That would make it impossible for the 'archetypical' individual inventor working in their garage or basement (if those actually still exist) to get a patent on their invention though.

Maybe companies shouldn't be allowed to own a patent. If a patent is always owned by the original inventors (so it moves with them if they change jobs, and any patent license fees always go to the inventors), it would help. But I guess there's always loopholes in such simplistic ideas ;)

> if those actually still exist)

That's what I fear is not the case anymore. And I am not sure we're getting back there. With so many regulations in all industries, it has become impossible to innovate from your home. Unless you really want to patent BS like big companies do, which is ugly...

It sucks, but that's how it has become.

If you want to protect such minorities, you can offer them a discount, but make it clear indeed that the patent belongs to the individual for even longer, not to the company.

Doing nothing as it has been until now certainly doesn't help.

> You can: increase the price for patenting something. Google will think twice before spending 100K for hello world.

This is a terrible idea. Google will not think twice before spending 100k for a patent on hello world, but startups and small businesses would be priced out of the patent system entirely. All this would do is further increase and entrench the power of large corporations to the detriment of startups, entrepreneurs, SMBs, and ultimately, consumers.

My suggestion is to double the fees every year so that only patents will be kept which really generate relevant economic returns.

My suggestion is 1000 USD in year 1.

After 10 years the fee is 1m USD per year.

After 20 years the fee is 1bn USD per year.

The current schedule of 23 years aligns nicely with this. If a Pharma patent is really generating billions of returns then maybe it would be worth for the company to pay 8 bn USD fees in year 23.

patents should be like 10 years max on any account in my opinion. You shouldn't have to wait a generation for one to expire.

If the system worked at all, an inventor could still make money selling IP weapons to corporations.

"For every complex problem, there is an answer that is clear, simple, and wrong." ~ HL Mencken

The problem is not the number of patents. If you limit that number, what you will get is a host of patent trolls and other largely-bad actors optimizing to be the first to get each year's patent quota.

I especially agree with #5, if you don't invent/show reasonable progress on an algorithm/device/etc within X (TBD) years the patent is useless; you should have kept it a trade secret or worked on it. Also, anything that is overly broad should be laughed out of the patent office. also current patents are far far too long

You remind me of https://xkcd.com/592/

For every point, you have to take a subjective decision that is in no way trivial.

1) Additionally to what happens today now you need to compare patents (how would you increase the cap for non obvious if you can't say A is more non obvious than B)

2) Additionally to what happens today now you need to decide if something was possible 9 years ago. Our search tools are not that great at "present me the image of 9 years ago" and even then not everything is public, etc.

3) Additionally to what happens today now you will have same patent applied to multiple domains with small variations.

4) Yet another process, which now happens in court if someone thinks the patent must be defended (a granted patent does not mean it can not be later invalidated)

5) How do you decide very clearly first ? No improvement versus today.

I do agree with you though that the current system is not ideal. But a simple change like reduce to 10 years the duration might be the best next step (and while I like 5 years idea, making big jumps has more chance of push-backs from invested actors)

I thought you'd be linking to 793

https://xkcd.com/793/

> For every point, you have to take a subjective decision that is in no way trivial.

Have you ever been through a non-trivial patent filing process? Any other non-trivial administrative law case? A non-trivial court case?

This is precisely why these processes are so complicated, both under the rules I proposed and the rules currently in place. It's why we have judges and lawyers. The way policy happens is people float ideas, much as I did, and as other people have. Those get discussed at a high level. If something looks reasonable, people do deep dives and work out the details. Those are further smoothed over as laws translate into regulations, and the contours are defined in case law.

I don't think your comment reflects an understanding of the problem we're trying to solve, which is narrowly patent trolls, and broadly, that it's impossible to do anything without inadvertently stepping on a dozen patents.

WE NEED FEWER PATENTS.

> 1) Additionally to what happens today now you need to compare patents (how would you increase the cap for non obvious if you can't say A is more non obvious than B)

You don't need to compare. The patent office administrative apparatus needs to adjust the bar each year. There's already a bar set by them. That's the point of delegating to an executive branch -- they can make changes like this one quickly.

> 2) Additionally to what happens today now you need to decide if something was possible 9 years ago. Our search tools are not that great at "present me the image of 9 years ago" and even then not everything is public, etc.

Same as prior art. A patent examiner looks and has a discussion with the patent filer. It can be invalidated in court.

> 3) Additionally to what happens today now you will have same patent applied to multiple domains with small variations.

Which I would like to reduce or eliminate. When LSA was invented, there was a swarm of patents applying it to every which domain, making it impossible to do much of anything.

> 4) Yet another process, which now happens in court if someone thinks the patent must be defended (a granted patent does not mean it can not be later invalidated)

With one difference -- and this was lack of clarity in my writing -- my process kicks in at the time a patent is filed. If I want to avoid patent trolls in my domain, I can set up a search for relevant patents.

> 5) How do you decide very clearly first ? No improvement versus today.

This is simple. I am switching from OR to AND. You have to be first-to-invent AND first-to-file. If you're first-to-invent, and I'm first-to-file, neither of us gets the patent.

Indeed, if I'm first-to-file, but you can show that you invented before I filed, no patent is granted.

The improvement is fewer patents.

Do you know anybody working in a patent office? I know people in the European patent office and they are overwhelmed by numbers of application, they need very educated people to read through lot of (boring) documents and they do not get much incentive to be perfect (I mean salary is fixed).

> needs to adjust the bar each year.

You make it sound like this is moving a slider - I have strong doubts this is the case, someone need to write a description, explain it, make sure people understood how to interpret it. Doubt that changing rules each year would work well.

> A patent examiner looks and has a discussion with the patent file

You seem to agree that this is additional work.

> Which I would like to reduce or eliminate.

Making a policy must take into account side-effects. The intention can be good but the result of it can be bad

> my process kicks in at the time a patent is filed

One more process? There is a search already when you fill (or pre-fill don't remember the terminology) something. In my limited experience, results are - let's put it mildly - average.

> if I'm first-to-file, but you can show that you invented before I filed, no patent is granted.

What is the difference with today? But this is easier said than done. Patent language is complex, details are very minor and people will argue very every little thing.

Fewer "active" patents and "less incentive to be a troll" would also be obtained by reducing drastically the granted period and that is clear and simple to implement.

> WE NEED FEWER PATENTS.

If a system is implemented with this as its primary guiding principle, what you will end up with is fewer good patents, and a vastly higher proportion of junk or actively malicious patents.

We don't need fewer patents. We need to eliminate software patents, better fund the patent office so it can employ enough people to properly review all the applications it gets, and probably do something different with biological and pharmaceutical patents (too many of these are built heavily on public research, and AFAIK it's far too easy to make small changes and get the original patent extended if it's still lucrative).