Cloudflare offers $100,000 for prior art to nuke networking patents a troll has accused it of ripping off Cloudflare today offered $100,000 for evidence of prior art to kill off a bunch of patents it is accused of infringing. Sable Networks, which took over the intellectual property of failed "flow-based router" startup Caspian Networks in the mid-2000s, sued Cloudflare and five other companies in March, claiming they ripped off …
"needs to be beaten soundly with a very large stick."
The current patent office is so long in the tooth it needs to be taken out behind the barn and put out of it's misery before it does any more damage to the things around it in it's senility.
The whole thing needs to be figuratively burned down and rebuilt. It has gone far beyond a joke.
Part of the problem is the requirement for a patent to be novel, and have an inventive step. In the US, patent cases are jury trials. Over the years juries (who are not specialists in intellectual property law) have been persuaded that the amount of novelty and inventiveness should be set lower and lower. The USPTO has responded, and allowed patents with lower levels of novelty, since the legal system has decided that these will stand up in court. It's a bit of a chicken-and-egg situation too, since lawyers can argue that "of course the patent is valid - the USPTO wouldn't have granted it if it wasn't".
When we asked about patent actions in the US, the advice was 1) don't get involved in patent actions in the US, and 2) get through the first (jury) trial as quickly as possible and take it to appeal. Appeal cases are heard by judges only, and are more likely to get a just result.
The USPTO is part of the legal system, and stating that it is useless without examining the wider flaws in the whole system isn't going to help.
It's not just US patent law. A good fraction of my job is being in charge of IP for the division of the company I work for. It's a mid-size company, just shy of $1B/yr revenue. I have 8 issued patents myself, and a bunch more applications.
You only have to deal with WIPO/PCT for 5 minutes to realize the EU, JP, KR, CN, etc patent offices are not any better.
The challenge is that to invalidate a whole patent, you have to beat back every single claim, which usually means at least 20 claims per patent (and does in this case too). Having my name on three or four patents in the same general area, I looked to see if there was a quick $100k for me here, and sadly there isn't. It would take days of work just to translate the claims into plain English, and then you have to stretch Google pretty hard to look for around 100 hypothetical pieces of prior art. If the patent attorneys did their job, two or three of the claims in each patent would survive this process (because the art of writing a patent is making the successive claims narrower and narrower, so that even if the main claims are thrown out, something survives).
If I was seriously looking for the prior art, I'd read Larry Roberts' own publications for the few years from about 1996 on. All the ideas seem pretty obvious with my excellent 20/20 hindsight.
Also, unless Cloudflare has gone into the router and switch hardware design business, it stretches credulity to suggest that they have implemented even a single one of all the claims in those patents. These are not things you do in software.
All the ideas seem pretty obvious with my excellent 20/20 hindsight
The basic idea seems to be effectively to treat individual transport connections as units of routing by labelling the packets based on source/destination network+transport address and then routing on the label rather than the destination ip address alone. I know flow-labelling was discussed in the early development of DECnet Phase V - though largely as a means of header compression, though QoS-based routing was also considered. I'm pretty sure it wasn't published though and it hardly seemed revolutionary at the time - it's little more than a table-based lookup. Having established the principle of flow-labelling, the remaining patents seem to just to be routing as we know it but using a flow label rather than an address.
Fortunately, the patents to which my name is attached have long lapsed - by the time the lawyers had finished with them I had no idea what they were actually claiming. Sounds to me, too, that this is one best left to the professionals.
"Also, unless Cloudflare has gone into the router and switch hardware design business, it stretches credulity to suggest that they have implemented even a single one of all the claims in those patents. These are not things you do in software."
Is this not how new patents are created. There was the peak of taking an existing patent and adding "on a computer" and voila, new patent. Since that peak has passed now the key phrase is "implemented in software". A rising star these days is "in the cloud"
If I could only foresee the next magic phrase I could patent a new business method of creating new patents with it.
After all everything that will be invented has been already.
"(because the art of writing a patent is making the successive claims narrower and narrower, so that even if the main claims are thrown out, something survives)."
But isn't that exactly the point though? Patents so often are overly broad, covering a whole raft of unrelated applications and theoretical uses and need, nay MUST, be narrowed down. And they mustn't be allowed to be renewed with minor tweaks. They have a specified lifetime for a reason. The patent in question in the article are near as dammit 20 years old now.
It just takes a lot of work to find it There isn't much that is really new. Do you have to have actually documented and/or implemented this, or do you just have to find evidence of prior art existing? This could be almost as lucrative as bug hunting.
Pretty well just find evidence of the prior art existing - if it existed, then it was implemented at least once before the patent was invented (or filed in the rest of the world).
A case often used to prove this is Windsurfing International. See:
https://en.wikipedia.org/wiki/Windsurfing#Patents
Their 1970 patent got invalidated because a licencee who resented the high fees discovered a similar invention made by a British schoolboy in 1958, and reported in a local newspaper. The courts decided that key parts of the invention had been made earlier, and invalidated the patent.
Despite the myth that everything has been digitized & put on the Internet by now, it can be hard to find prior art. My own company, which has made a large number of acquisitions over the decades, had a recent patent scare where one of our competitors filed a patent on something we were also about to use in a product about to be released. Yet it was one of those things that in the more general technology was conceived 100+ years ago, and in the modern implementation in our field, still expected to have been patented (but not implemented due to economics) decades ago. Long story short, after over a year of us & our lawyers searching for prior art as part of our freedom to operate due diligence before product release, we found the idea had been patented in the 70's (and expired in the '90's) by a company we'd acquired in the '80's. Why did all of the searches across many patent and journal databases come up empty? Because the old patent didn't contain a single keyword that any searcher used, even by those of us who were working in the '70's. And why was that? Partly because terminology had changed, and partly simply because the patent was written using first principles descriptions rather than assuming the reader knew the buzz words. It was a wonderfully well written patent from a robustness standpoint. Unfortunately, it was also way ahead of it's time to have been put into production way back when. But it did allow us to have as kryptonite in our back pocket if the competitor ever went after us. (i.e. Thank you for your cease and desist letter. We are just using an implementation of our own expired patent ####### from the 1970's. We'll let all of your competitors know about it as well. Have a nice day.)
It doesn't have to be a coincidence that it's hard to find. The "submarine" patent is a well-known tactic if you think you may be sued by a competitor. File a patent in a faraway country and don't bother to translate it into English. When you get sued, announce "This is an implementation of Peruvian* patent number nnnn" and go on your merry way. Some countries also have a reputation of being able to expedite the paperwork, and return a nice early filing date, in exchange for "additional fees"...
The tactic of threatening to post copies of the prior art to everyone else in the business is also well-known as a method for getting competitors to leave you alone.
*Other countries beginning with P are available.
Click links to find meaningful expressions,
US6954431B2
US8243593B2
US7012919B1
US6977932B1
Ignore Google and go to https://worldwide.espacenet.com/patent/
Enter numbers. Look for either the Global Dossier link which will give you a trail of documents. Look through documents for things like non-final or final rejections or opinions etc. See if they contain mention of any prior art that might be resurrected in order to challenge.
Also look under the patent family list and do a similar search on the numbers that crop up.
Unfortunately a cursory look does not appear to show much meat however generally you find that where an application has attempted to file under WIPO or EPO the reports tend to be more comprehensive and the applicant, whilst they have a US patent, gives up trying to pursue them under WIPO/EPO
However in this case only US8243593B2 gives a WO (WIPO) application
https://worldwide.espacenet.com/patent/search/family/036595594/publication/US8243593B2?q=US8243593B2
WO2006069044A2
WO2006069044A3
https://worldwide.espacenet.com/patent/search/family/036595594/publication/WO2006069044A3?q=pn%3DWO2006069044A3
https://register.epo.org/ipfwretrieve?apn=US.2005046092.W&lng=en
It's not much but they are going to reject claims 1-40 on the basis of lack of novelty and lack of inventive step, Page 4) of the document, 1) Statement. They also give their reasons same page 2) Citations and Explanations.
"Claims 1-40 lack novelty under PCT Article 33(2) as being anticipated by ZIKAN et al. in US Patent No 6,310,881 B1, hereinafter referred to as ZIKAN."
https://worldwide.espacenet.com/patent/search/family/022642825/publication/US6310881B1?q=US6310881B1
Best I can do you.
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2024
