[DarkmSparks]

Im still very undecided on software patents and have been since they were first awarded

software patents in general are a bad idea imho, go trade secret and never let the ideas enter the public domain.

otoh

society benefits a lot from good ideas entering the public domain, and the patent system is fairly effective at stopping everyone keeping their secret sauce a secret.

Since the US has them, and this decision seems fairly clear cut (or Amazon and google would have got them thrown out as invalid already) about all I can say is congratulations to Kove.

[tobyjsullivan]

I’d be curious to hear an anecdote about a single software patent that benefited society at large.

For example, I agree it would be very interesting and arguably valuable to have a public document describing how Amazon built S3. Unfortunately, these patents don’t describe what AWS did. They describe how some random “inventor” thought a system like that could be built (more realistically, of course, that rando likely never dreamed of any valuable application like S3).

[radicalbyte]

I've yet to see a software patent which wouldn't have been reinvented by someone else looking at the same problem. The thing with software is that the cost to experiment is extremely low compared to other fields so we can, and do, "just try it".

At this point in my career for example everything I do is basically tweaking or adapting patterns I've seen in other systems to the ones I'm working on.

[somewho]

Cost to experiment being low doesn't make non-obvious ideas obvious. As a trained researcher with patents I'm no longer surprised by how often simple ideas have remained undiscovered for decades.

[denton-scratch]

> go trade secret and never let the ideas enter the public domain.

If you rely on trade secrets, then you can't sell your software without protecting it using a dongle or whatever; your "customer" might in fact be your competitor. It's not that hard to figure out how a piece of software works, and to write your own code that works the same way.

It's not hard to figure out how some novel machine works either; but how to fabricate it is often a tougher challenge. It's easy enough to explain how a transistor works; fabricating reliable transistors at scale is another question.

[DarkmSparks]

thats mostly why most software is SaaS now. No need for a dongle, the end user never gets near the actual software.

[denton-scratch]

Hmm. A dongle was used to protect trade-secret software in the days before software patents. Now SaaS is used instead; therefore software patents are still unnecessary?

I used to work for a firm that rented all its premises. My boss explained to me that the company's assets were its software, and nothing else. The patent portfolio is the physical manifestation of those assets - you can borrow against a patent portfolio, but not against a vault full of trade secrets.

You'd think he'd have mentioned staff; but staff can leave, and you can't secure loans and investment against employees.

Anyway, it looks as if that firm at least was using patents not as a way of preventing unfair competition and the stealing of inventions, but as a financial instrument. That's not what they're supposed to be for. And the fact they are used that way makes it horribly difficult to reform the patent system; if patents are like (e.g.) bonds, then reforming the patent system carries the risk of crashing the financial system.

[DarkmSparks]

Software patents include a detailed explanation of the way the technology works, in return for telling society how the technology works (so anyone can use it at the expiry of the patent) the government grants exclusive use of the technology - patents are a trade.

If an inventor keeps (can keep) their technology secret, it is better for them to not patent it - and they can keep exclusive use of it as long as they can keep it secret.

Trade secrets laws are far more brutal than patents iirc.

This is why I am undecided, there are no easy answers.

[newsclues]

If software patents were to become a good thing the duration of the patent should be very short.

[DarkmSparks]

but then there would be no incentive to publish the internal workings of the innovation, and you are back to trade secrets. 20/25 years is already fairly short tbh.