IBM Adopts Open Patent Policy

Andy Updegrove writes to mention a New York Times article about IBM's bold new move to reform patent practices. The nation's largest patent holder will adopt several new policies intended to clear up the veil of secrecy and wall of lawsuits that plague the patent process. From the article: "The policy, being announced today, includes standards like clearly identifying the corporate ownership of patents, to avoid filings that cloak authorship under the name of an individual or dummy company. It also asserts that so-called business methods alone -- broad descriptions of ideas, without technical specifics -- should not be patentable. The move by I.B.M. does carry business risks. Patents typically take three or four years after filing to be approved by the patent office. Companies often try to keep patent applications private for as long as possible, to try to hide their technical intentions from rivals."

Comments

Value proposition

[tomstdenis]

Patents on many things are kinda moot. Put a patent on your CPU design, but only a handful of companies on earth can actually make an ASIC...If foundry's are a dime-a-dozen what's your value? [hint: they're not, which is why being able to make reliable chips is a value proposition worth holding onto]

If companies just focused on things they can offer, at qualities no one else can then they'd make money. It's when they get this entitled sense of "I have a right to be making gobs of money regardless of what I do" that we get into this patent mess.

Tom

Re:Value proposition

[Brickwall]

It's when they get this entitled sense of "I have a right to be making gobs of money regardless of what I do" that we get into this patent mess.

But it's not a case of "regardless of what I do" - it's a case of "I did something unique, and I want to be rewarded for it". That's why the US Constitution specifically provides for patents of limited duration. If they truly developed something new, they are entitled to a reward (as opposed to the ridiculous "business model" patents that caused RIM and the Blackberry so much trouble). If IBM is truly leading the way to junk the business model patent, they are doing all of us a service.

Re:Value proposition

[Chacham]

Being obstructed by a THOUGHT is ludicrous. Getting patents on things that all of your competitors could easily do as well means your business is not well founded.

If your competitors can't do it, there is no reason for a patent.

The patent is made to protect the filer. The protection is that it protects his time and money.

If i create the better mouse trap, and the world starts beating a path to my door, i need to invest money to start making more. The problem is, by the time i put the money and effort in to getting my factory going, an old-style mouse trap company will use one production line to make this new one. End result, i'm out all that money because they can outsell me due to preexisting infrastructure. However, if i get a patent, i can put my money and effort into it, knowing that i am protected for a period of time.

Re:Value proposition

[networkBoy]

Can you honestly tell me with an electron scope you could figure out how the circuit was made
Maybe not, but I can with the following assets:

FIB (Forced Ion Beam)
E-Beam
SEM (Scanning Electron Microscope)
AFM (Atomic Force Microscope)
Industrial espionage
Hiring some of your key staff
Bribing your FAB &&|| Foundry (especially if in APAC)
Bribing your customers (they often have datasheets not publicly available)

(honestly I don't need all of this, various subsets can accomplish the same end results in varying ammounts of time (days, weeks, monts. Always < 1year)).

Ask me how I know (Hint, my company has both been on the receiving end and giving end of this. The giving end was to prove/disprove whether they stole our design. They did.)
-nB

Re:Value proposition

[morgan_greywolf]

I agree there is a problem if your employees steal your companies technology and take it elsewhere. That's where employee contracts come in handy. I think instead of a Patent act we should have a secrecy act. That is, you can be sued if you take advantage of someone elses secrets by poaching employees or espionage [but not via co-invention].

We do. It's called trade secret law, and it's a very old form of law rooted in British common law. Virtually every tech company out there makes use of trade secret law on a day-to-day basis. Of course, the problem with trade secret law is this: what if I can figure out how to make a 65nm circuit by examining it with electron microscope? Well, then guess what? The trade secret isn't secret anymore and it loses its protected status. That's where patents come in.

I'm not in favor of abolishing patents -- I think patents are a good thing when used correctly. It's the patent abuse that ruins the whole system. I don't think we should throw out the baby with the bath water. What we need to do is reform the patent system, and IBM has taken a good first step in leading the way. No more business process patents, no more vagueness, and for godsakes, if you haven't implemented your invention yet, you don't deserve a patent. And the USPTO needs to stop granting patents for things that are blatantly obvious to anyone skilled in the art.

Re:Value proposition

[ceoyoyo]

The problem with trade secret law is that your technology is a secret. The point of patents isn't really to protect inventors -- in many cases they can do that themselves with trade secrets. The point is to encourage further innovation by making knowledge public. Nobody would make their mouse trap building process public because everyone would steal it... so they are awarded a patent in exchange for making their invention and specifics on how to duplicate it, public. The patent guarantees them a monopoly on it for a set period of time in exchange for sharing that knowledge.

Here's an idea

[dsginter]

Make frivolous patents illegal and punishable by a 10-year FPMITA prison sentence. Then, offer frivolous patent holders a indemnity by turning in their frivolous patents to a patent disposal system (similar to a fire arm turn in). Maybe even give them a lemon cookie for being a good citizen.

Then, allow all patent holders to submit their votes for the most frivolous patents. Prosecute the top 100 holders every month. Rinse, repeat (until their are no more frivolous patents).

Re:Seriously?

[ZachPruckowski]

If the only people who are saying that something is screwed up are the people hurt by it, that's one thing. But it lends a lot more legitimacy to a campaign to have not only victims but beneficiaries calling foul. Not to mention that IBM has a lot more campaigning weight and power than the EFF, GNU Foundation, and everyone else combined.

It's not just IBM

[Phantom of the Opera]

I.B.M. is one of several companies that have agreed to submit some patent applications for open peer review as part of the project, beginning early next year. The others include Microsoft, General Electric, Hewlett-Packard, Oracle, Intel and Red Hat.

The above-mentioned corporations do all skilled legal staff but patent litigation is not their business. IBM and GE in particular have expertise that allow them to follow through on their patents. Any "copy-cats" would have difficulty producing products from many of the more esoteric, high tech or highly process oriented technologies these companies have to offer.

If patent finding publishing becomes widespread, it will give companies the legal footing to allow them to concentrate on creating technology rather than split hairs over buzzwords. We see an aligning of real innovators against those who simply gamble that some court will award them money like mana from heaven.

not often you see

[mackil]

It's not often you see a huge business choose to be "open" like this with the patent process. Especially considering how many business make patent trolling their main money maker. However, while it is rather commendable, I fear that it may hurt them in the long run. Most opposing companies will not be so benevolent. After all, the nice guys always lose...

Building a New IP Marketplace (IBM Research)

[giafly]

'The new policy is the result of a development process that included external as well as internal input, and is based upon a Wiki that gathered the comments and contributions of "over 50 patent and policy experts from the United States, Europe, Japan and China," offered during May and June of this year. That document can be accessed at this page at the IBM site.'

Via the ConsortiumInfo Standards Blog

good, cause patents are anti free market

[argoff]

In fact, they are not only anti free market, they are genocidal. Like how they held back safety devices for 20 years in automobiles while over million people died in auto accidents. Like how US pharmacuticals sued the African nations in the world court to stop making AIDS medication generics while over a million people died of AIDS. In the future 3d printers and nanotech are going to move manufacturing out of the factory and back into the home, and those who believe in patnets will want to extend the coercive power of government into every last persons life to collect royalities and revenue streams. Dont think so? Just look at what happened with copyrights - 30 years ago, no one would have believed that they would go this far either.

Not so fast

[Moraelin]

I'm a programmer, and TBH I'm _for_ software patents, as long as we get rid of the generic "business method" patents.

See, patents were supposed to do two things, in this order:

1. First and foremost, to make sure we're getting the exact recipe to make something, instead of ending up buried somewhere out of reach. Sure, you get your monopoly on sewing machines, Mr Singer (for example), but in return society gets the _exact_ recipe and description of how it works. After 20 years, we get that in the public domain.

2. To stimulate innovation. Go research something already. If someone else patented it, well, research something else.

The current non-patented programming fuck-up serves neither point. We have millions of monkeys who don't invent anything, and don't share anything. They just copy-and-paste (even via memory, but copy-and-paste nevertheless) someone else's algorithms, and never invent anything new. Ever. And the results of even that unoriginal copied-and-pasted work remains buried somewhere behind a wall of NDAs, on some old tape in a steel safe.

Sorry, people, that's not how technological progress works. What we have here is stagnation and waste of resources.

You're affraid that patents will put your company out of business? Well then how about said company starts investing in research already? How about inventing something new already? How many people does your company pay to research new algorithms? No, seriously. Be honest. Zero, perchance? No, that's not innovation, that's not progress, it's just copying someone else's work, over and over again.

Yes, software patents do carry the stigma of having been abused and mis-used by patent-trolls. There were a lot of bullshit and obvious patents snuck through just because the patent office got disoriented by anything that mentioned "in software" or "on a computer". Ooh, it's the same old volume knob, only now "on a computer"... that sounds soo high-tech, let's patent it. Duly noted, and I too wish we'd be rid of _those_ already.

But there are lots of things which aren't trivial at all. And blimey, I'd love to see more of those researched and documented.

E.g., to give the old (and now expired) whine about the LZW patent, how about you invent a compression algorithm from scratch, if you think compression is trivial. Yes, LZW (and LZSS and arithmetic compression and everything else) seems trivial when you just copy it (even via memory) from someone else's book. Sure, copying is easy. Now you try _inventing_ a new one, then tell me how trivial that was. If you're not damn good at maths, I doubt that you'll even know where to start. No offense. I tried and didn't know either.

Re:Not so fast

[TheRaven64]

E.g., to give the old (and now expired) whine about the LZW patent, how about you invent a compression algorithm from scratch, if you think compression is trivial.

I was not familiar with exactly how LZW worked, but I am familiar with basic information theory. To test this, I sat down for a minute (actually, I was already sitting) and considered how I would go about creating a universal stream compressor. I then checked the Wikipedia page for LZW. Reading around that, it seems that LZW is very similar to the algorithm that I came up with in a short time (different symbol lengths, slightly different encoding, but comparable. I'd have to implement them both to see which was better, but a quick information theoretical examination shows that they should be relatively close). Of course, these days there there are much better algorithms available; I would claim to be able to come up with something better than, or even equivalent to, the algorithms used in bzip2, for example.

About the only thing LZW had going for it was its ubiquity; the fact that the UNIX compress tool and the GIF file format used it. Designing an algorithm better than LZW is not hard, as is shown by the plethora of better compression algorithms that exist. I would expect that anyone with a degree in Computer Science or Mathematics could come up with one at least close to LZW (and if they couldn't, I'd like to know what institution awarded their degree). The difficult thing is generating a better decompression algorithm that still lets you view your GIF images.

Another good example of a software patent is the Marching Cubes algorithm. This is the standard way of generating an isosurface of a volume. Or, rather, it would be. Since it was patented, the community responded by developing the Marching Tetrahedra algorithm. Anyone with any experience working with volume graphics will tell you that both are trivial to derive from first principles (i.e. obvious), being little more than extensions into three dimensions of flood-fill algorithms (a slight oversimplification, but not a huge one). Similarly, LZW is obvious to anyone with a background in information theory.

By the way, both of these patents were filed within a year of each other. This means that both of them are expiring about now. The only positive benefit that either has had is to force implementers to come up with better algorithms (almost the opposite of the stated aim of patents) as a work-around. This is fine (although smacks of re-inventing the wheel in a lot of cases) for the most part, but in cases like LZW it is a huge impediment to the industry since it requires the abandonment and replacement of an established file format.

IBM is trapped by its own invention

[pieterh]

IBM's patent lawyers invented software patents in the first place. You can see what they were thinking... "we patent all our hardware, now more and more of those designs are implemented in software, so we should patent software too".

The trouble is, there is no dividing line between a patent for microcode, and a patent for swinging a pizza. The moment you allow the definition of a software model to be patented, you open the gates to patents on every idea. It just takes time - 10 years - before the patent industry assiduously hacks every single definition, but it happens.

IBM is now very unhappy with the patent situation. They have invested hundreds of millions (billions, probably) in their patent portfolio but it mostly covers older technology where there is less and less licensing opportunity. Meanwhile the patent business is creating record turnover, which deflates their patent portfolio.

Yes, IBM is against business process patents. Big deal. Any business process can be reworked as a software patent. Any border that tries to separate the 'good' software patents from the 'bad' ones can be hacked until it's gone.

The only reason large firms like IBM, SAP, and Microsoft still support the software patent model is because their patent policy is dictated by patent attornies. If the CFO or CTO was in charge, it'd be different.

As for the suggestion that patents were "closed" before... bizarre. The whole justification for granting a patent monopoly is to reward the inventor for publishing his work.