More patent lessons- first on submarine patents (basics!) and then on how patent pools are licensed. I don’t really want to continue this series, but the past few days have been a good reminder that there is a lot of misinformation out there around patents.
To start with, OSNews wants to claim that there are no such thing as submarine patents anymore, relying on a very specialized, nuanced definition of submarine patents. Their definition is… well, it is internally consistent, but I’ve never heard the term ‘submarine patents’ used that way before, and if you define it that way you run the risk of thinking that submarine patents are no longer a problem. This is sadly not the case.
Most people define submarine patents not as patents which are unknowable (because of the PTO’s process), but as patents which are unknowable or effectively unknown and therefore can’t be dealt with effectively.
The problem here – with software patents in particular- is that they are so numerous, so broadly worded, and so inconsistently worded, that searching for them is like searching for a submarine in the ocean. It is incredibly difficult, incredibly expensive, and very frequently ineffective to look for the ones that could torpedo your software product. And so most of the industry doesn’t bother- they just cross their fingers and hope.1
Patent pools like MPEG-LA’s are an attempt to avoid this problem, not by searching the ocean, but by bribing the submarines to surface and getting them to agree not to use their torpedoes. So they do reduce the risk of submarine patents, but they definitely don’t eliminate them- each company will still have to do their own risk analysis when they sign into a patent pool, to make sure they are comfortable with the risk from patents outside the pool.
It is worth noting here that patent rights are like copyrights, and not like trademarks: you can let them sit as long as you want without enforcing them (generally speaking.) This is another part of what makes submarine patents messy- merely using the technology in a very public way (like many companies do with MPEG) does not necessarily guarantee that there are no risks; it only means that if there are risks, they haven’t surfaced yet.
So, bottom line: if the OSNews article made you more comfortable about submarine patents, get nervous again. Using their technical definition, the risks are zero, but using the more common (and more reasonable) definition the risks are usually low but they definitely aren’t zero.
On the other point: Gruber said yesterday that Google, as an MPEG-LA licensee, would be protected if Ogg violates an MPEG-LA patent. This is possibly correct, but highly unlikely. Companies who give their patents to patent pools don’t actually give them up completely- they typically only promise not to use them against very specifically described technologies. If you’re not that specifically described technology, the patent owner is completely within their rights to track you down on their own.
In the case of MPEG-LA, the patent license is almost certainly for implementations of MPEG codecs, not for implementations of any random video codec you want (like ogg.) So Google probably has some other reason they feel safe about ogg- it may be that they’ve done thorough research on the codec, or it may be that they have other cross-licensing agreements outside of MPEG-LA, or they may just be unusually tolerant of risk. Unfortunately, we can’t know, and they’d be crazy to tell us.
Again, like yesterday, I haven’t seen the MPEG-LA licensing terms; it is possible that they do in fact cover implementations of any random codec. But that would be very unusual.
- One of the many, many ways in which software patents are broooooooken. [↩]