Physics Simulation Forum

Hi,
I just came across this: http://www.freshpatents.com/Method-of-simulating-dynamic-objects-using-position-based-dynamics-dt20070419ptan20070085851.php

I don't understand how this could have been patented since it's an extension/generalization of Jakobsen.. which they explicitly mention in the paper!

raigan

... then you haven't read much about the modern state of the U.S. Patent and Trademark Office. In a nutshell, it's completely broken. They don't have the technical prowess to oversee these types of patents, and furthermore their stated goal is to award as many patents as possible. Who cares if there's any validity to them, right? And if the patent holder calls you out, it will cost you upwards of a million dollars to fight, even if you eventually prove the patent was invalid in the first place.

By the way, if one wanted to play the "patent game", one could try patenting as many similar and related methods as possible. That way, if the original patent holder claims that you've violated their patent, then you just need to find one of your patents that they've violated. After that, some sort of settlement could be found between you and the other party, which is probably cheaper than lawyers.

I'm hoping that this was directed more towards rival physics companies (havok, etc) and less towards small, 2-person indie developers

Also, what possible purpose could it serve to publish a computer science paper detailing your patented method, other than to try and snare people who aren't aware that it's patented!?

raigan

Yes, it's rival companies like those who might play the patent game. Nobody's likely to go after small developers until they've made enough money that it's worth it. That's what lawyers, in general, do.



That did occur to me, pretty sneaky if it went down like that. I didn't quite figure out from your link who actually owns that patent. I presume it's Ageia's and connected to one of the papers that has been linked to here, for example "Position Based Dynamics" by Muller et al?

Plenty of others have laid out the groundwork why the USPTO is bulls**t especially with regards to software patents, but nothing's going to be done so it's hardly worth talking about.

After reading this recent news..

http://www.nytimes.com/2007/05/01/busin ... 80&ei=5070

I had at least a tiny shred of naive hope for the completely broken US pay to win patent system.



I believe large rival companies use them more as nuclear weapon trading cards. I won't blow you up if you won't blow me up.

As for attacking a small developer.. No problem. It's like the RIAA threat tactics. Pay $100 for a lawyer to send out a letter, get $5,000 back. ( Ah.. The American dream! ). This type of blackmail can easily happen when a company goes out of business and their patents are sold to the highest bidder.

btw. Erwin.. What's your position on posting links to patents? I personally think that they should be disallowed as from my understanding.. Willful infringement of a patent can get ya some lovely triple damages. I NEVER click on a patent link or read patents, but it's safer for everyone and the companies they work for if the links are never posted.

That article was encouraging ... thanks for that. The patent in question in that case would seem to be a pretty obvious solution even for somebody outside the auto industry.

I'm a bit confused about how things end up working in practice -- for instance, AFAIK Stam's "Stable Fluids" is patented, however it's being used all over the place: http://www.plasmapong.com/

It's also the basis for many of the real-time fluid solvers I've seen presented, GPU-based or otherwise. Is everyone just taking huge risks?

You take a huge risk for breathing in the US.

Patents are not like trademarks where you have to constantly defend them across the board. With a patent you can pick and choose who you will attack.

Does Stam personally own it or does Berkeley ( paid for by your tax dollars! )? I understand that it cost around $1M do sue/defend someone for/from patent infringement. Kinda takes it out of the playground for most individuals with the dream of going pro at being a mob boss.

I believe Alias (or, Autodesk now i guess) own the patent; of course it's based on semi-lagrangian stuff that has prior art for the past several decades..

Hypothetically, if one was using a patented algorithm, would it make any sense to try to obtain a license from the patent-holder to use the algorithm (and thereby expose yourself)?

raigan

So if this is really so expensive, I would argue nobody sues a small indy developer where you never get your 1M back and you will never sue somebody who can actually defend himself because in this case you want to be 100% sure that your patent has no prior art. Otherwise this would be like shooting into your own foot...

I think in the case of Ageia their patents are mainly to please their investors and in my very personal opinion AGEIA will be history more sooner than later. They have no product and don't sell anything. Or in other words - they give their engine away for free and nobody wants their PPU card. At one point they will run out of money...

Anyway, isn't the patent system about to change in the US. In Europe we don't have software patents luckily...

Cheers,
-Dirk

Bankruptcy/failure of Ageia almost makes things worse, as someone will end up owning the patents, and they'll be wanting to recoup as much as possible from their investment!

My main problem with this, hypothetically, is that if you _did_ develop a really kick-ass physics solver which had some things in common with the patented method, you wouldn't be able to open-source it (or develop tutorials or presentations, etc) as you'd be exposing yourself to future litigation..

Aside from the copious prior art regarding the PBD solver, I've also found some "simultaneous/parallel art" -- this recent paper presents a solver which seems _very_ similar to PBD, yet they make only a passing reference to it: http://www.cs.columbia.edu/cg/ESIC/esic.html

raigan

I hardly think censorship is the answer, nor is sticking your head in the sand. We actually have some power here, and it appears the tool for us to exercise our power is nearly within our grasp.

http://money.cnn.com/magazines/fortune/ ... /index.htm

So, keep posting the patents, or rather; keep a look out for ones that are still in the review phase and then we can shoot them down.

That's the trick. You don't actually have to sue a small developer. You just have to threaten to sue with your $100 lawyer letter.



Considering IBM's patent history I would not trust anything they have their hands in.



We can all dream.

I just remember that AGEIA holds another patent which is a description of the ODE Quickstep solver. So along with the Jacobsen position based solver they now very obviously claim two very popular algorithms - that are by now means invented by them - as their intellectual property. This alone is an unbound cheek already. I also wonder if the patent holders are blended so much by now that they are stealing other people work and claim it as their own which is just pathetic. Those kind of people should be excluded from any conference like SIGGRAPH, Eurographics, etc since they very obviously don't respect other people work...

Havok owns patents too, some people even patented springs, EA criterion has all the ex mathengine patents. I guess when you are in competition and your competitors own patents you are somehow forced to do the same. For example Havok at the moment doesn't offer a deformable solution and this new patent could clearly be an obstacle for them. I dont agree with these patents but i think we must recognise that those are the rules of the game at the moment. This patent if seen from the Ageia perspective can make a lot of sense.


i agree on this. If all the work we use would have been patented we could not write a single line of code and the same is valid for the same people who are patenting things at the moment. It's slightly arrogant, however that's how business works...

cheers,
Antonio