Slashdot Mirror
IBM Breaks Open Source Patent Pledge
Jay Maynard writes "IBM has broken the pledge it made in 2005 not to assert 500 patents against open source software. In a letter sent to Roger Bowler, president of TurboHercules SA, IBM's Mark Anzani, head of their mainframe business, claimed that the Hercules open-source emulator (disclaimer: I manage the open source project) infringes on at least 106 issued patents and 67 more applied for. Included in that list are two that it pledged not to assert in 2005. In a blog entry, the NoSoftwarePatents campaign's Florian Mueller said that 'IBM is using patent warfare in order to protect its highly lucrative mainframe monopoly against Free and Open Source Software.' I have to agree: from where I sit, IBM likes Open Source only as long as they don't have to compete with it."
So, out of 173 possible patent infringements, 2 of them were supposedly pledged to not be enforced.
I can see why you feel hard done by.
Is 1563649 a prime number?
Open source has made IBM a lot of money. Now they want to have their cake and eat it too.
...who wouldn't like to invest time and money to create something, then have to turn around and compete against someone who basically just copies it and gives it away?
When you say "copies it", do you mean "ctrl-c, ctrl-v" or "re-engineer from scratch"? If you mean the former, that's a serious accusation, and you need to back it up. What part of their work was electronically copied? If you mean the latter, then so what? Do you really mean to imply that people have the right to distribute ideas and yet still own them? Do you think the descendants of some cave-man should be getting royalties for every combustion engine that internally uses fire? Yes, we should like it when people improve on our ideas. And yes, doing something in open source is a significant improvement (perhaps sometimes even if it's not quite as good).
That surprises me, who wouldn't like to invest time and money to create something, then have to turn around and compete against someone who basically just copies it and gives it away?
No time and money was invested in creating at least one of the patents. For example, look at one of the "infringed" patents, US Patent 7254698. The claims are merely a "shopping list" or "marketing glossy" of the peculiar feature of an ALU, which happens to be installed in a particular mainframe, that the Hercules guys would like to emulate:
Does multiply/add and multiply/subtract
Five deep pipeline, one result per cycle
binary or hex floating point format
Works on two different architecture formats.
It would take about a minute to make a spreadsheet in Excel that theoretically infringes on that patent, and probably an hour or so to make a perfect replica. Really all you need to do is implement mX+b=y with a five deep stack/array, given some peculiar input and output formats.
Now IBM will sell you a circuit board circa 2001-ish that will do this. They spent all their effort making an expensive machine that implements these simple math ideas in silicon. No one is stealing their physical hardware, or blueprints, or VHDL/Verilog, etc etc.
The emulator merely does the same calculations in C, and its free.
It boils down to IBM saying "no emulating our exact instruction set"
One ethical problem with patents like 7254698, aside from obvious ones like trying to patent basic linear algebra equations, is the supporting docs are all from 1999 to 2001 ish era. But its doing the submarine thing in that it was not issued until August 7 2007, "around a decade" after they were shipping silicon, more or less, sort of. And it won't expire until around 2023 which in the computer field is an absolute eternity.
I will give IBM credit, that unlike a patent troll, they actually built silicon to do something, not just patented an idea. But not much credit.
I have not looked into all hundred+ patents but they're probably all very similar to this one, but for other parts of the CPU instruction set.
"Science flies us to the moon. Religion flies us into buildings." - Victor Stenger
It's not that this particular case has 173 patent infringements, and 2 were ones IBM promised not to use so the project is pooched anyways. That's not it at all. Sure - the end result is the same for this Hercules emulator but that's not the point.
The point is IBM said they wouldn't use these patents against open source projects, and just did. Therefore the 500 or so patents that they claim are off limits to open source obviously aren't. Their promise is useless because now we know that as soon as it is expedient they will use these patents against open source.
In other words this Hercules emulator is merely the litmus test for IBM's open source patent promise, with lousy (but sadly typical) results.
Weaselmancer
rediculous.
Why would anyone at IBM still remember what they said in 2005? That's ancient history.
Five years is ancient history? What grade are you in, son? IBM started business in 1885. THAT'S ancient history. Thinking like yous is what's wrong with business and politics today -- nobody thinks long-term.
Free Martian Whores!
That's a really important point. I'm rather surprised the patent pledge didn't include an exception for companies that sue IBM. Either way, when this company sued IBM, as far as I'm concerned, they became fair game. This isn't IBM suing an open source project. It's IBM counter-suing a company that sued them first.
Even if IBM can't use those two patents (and it's not clear if they can't, given that TurboHercules is not an open source project, but rather a company that appears to be leaching off of an open source project), it seems completely reasonable for them to use the other patents defensively in this way.
Check out my sci-fi/humor trilogy at PatriotsBooks.