Comments on Software Patents

wdb Allow me some political notes. It does not matter whom we give our sympathy and whom not. But. As the German magazine c't mentioned, the case mattered of software patents. Software patents are contra-productive. Software patents should be removed from the world. Software patents have no legitimation of existence.

Larry Smith That's "legitimate existence". They have no "legitimate existence." And I can vouch for that. The only thing I can conceive of that is dumber than software patents is effectively immortal copyrights.

In a fast-moving world there is less and less need for such long periods of "protection". Rather than furthering theory and fact in science and the arts they instead become ways of forbidding such progress - always for good business reasons that have nothing to do with serving customers. Copyright and patent should be good for ten years and not renewable. If you haven't made money off your patented idea or copyrighted work in 10 years then you probably aren't and the material should go into the public domain. If you have made money with it, congratulations, your material should now pass into the public domain anyway.

And most of all, violations of patents or copyrights should be a tort, not a crime. The United States in particular has the dubious distinction of having a larger percentage of its population in prison than even the most repressive regimes in history. We have problems enough with putting minor criminals into toxic prison conditions and having them come out worse and more violent than when they went in. Non-violent people can only survive prison by becoming violent. Prisons are now manufacturing centers for the most violent and evil people one can imagine. We do NOT need to start tossing non-violent offending p2p downloaders in and having them come back with much more lucrative and violent ambitions learned at the knees of the masters we introduced to them.

Thank you for your patience.

FPX has written about the subject of software patents at the LZW page.

Duoas I've been playing around with images again, and as I will often do when coding in Tcl I will look on the internet to see if someone has coded something more efficiently than I came up with. To my complete shock, Googling "bit replication" took me to a page informing me that two dorks at DEC actually patented it on 14 March 2000 [L1 ]. I can hardly believe that the US Patent office would issue this, but beyond even that...

I've been using bit-replication when dealing with image data since the late 1980's. (It was a common, well-known technique even then.) I've also been tying my shoes by myself my whole life. Does someone have a patent on that? What does it mean to me?

Does it mean that I have to go and remove all instances of bit-replication from my code, past and present? Or am I expected to pay royalties to use code a baboon could devise in half an hour? Will I get sued for publishing code that uses it (free or otherwise)? If I have to use multiplication instead I might as well just trash my whole project, as the image processing would become intolerably slow.

I'm freaking out here. Is this absurdity actually possible? What can I do?

CMcC this is an ethical issue as well as a legal issue. AFAIK, the legality in the US pretty much requires that you get the advice of an accredited legal professional in the field of patents, otherwise if you infringe even inadvertently you can suffer punitive judgement. Smacks of collusion between patent attorneys, corporate monopolies and legislators who've been paid off, doesn't it. Welcome to, or is it

Ethically, it may be that your purpose in life is to serve as a beacon to others. It may be that civil disobedience is the appropriate choice. Thoreau: "Under a government which imprisons unjustly, the true place for a just man is also a prison." [L2 ].

You are not alone in this, see [L3 ] for some of the support which may be available to help you establish prior art.

Lars H: This particular patent doesn't look like a software pattern, though; it's more a hardware patent, describing a bit-replicating apparatus made up of shift registers. It also seems a tad strange that they describe their method mainly as being simpler than multiplying by the scaling factor, when in fact bit-replication is (up to rounding method) this multiplication, only accomplished through shifting and adding. Is shift-and-add even much of an advantage on today's CPUs?

Duoas Yes, I've since looked a little more closely, and it is (IMHO) a valid patent on hardware implementation of the algorithm, and considering the technique is so widely known and used I don't think that I will have any trouble.

As for performance, the reason I am so internet happy when programming Tcl is that performance in Tcl (and other scripted languages, whether byte-code compiled or not) is often at odds with the usual wisdom. It so happens that a bit-replication algorithm appears to be wasted in Tcl (unless someone knows something about expr that I don't --and I don't know much), but in machine code the difference is significant, from 2.5 to 4 times faster, depending on implementation (my figures).

I haven't really finished looking this through, but thanks for the responses.

tb - A computer is "per se" a machine for copying bits and essentially it does nothing else. :)

DKF: Arguably, it's also a machine for applying NAND to bits…