Google's SPDY protocol is a way to reduce web page load time be smart-multiplexing of the content. It has gained a strong opening acceptance among high-traffic organizations.

Despite armchair opinions that SPDY should have been leveraged with IP – could it have been?

Did Microsoft file aggressively, or defensively? Are patents like 6,604,144 and 6269403 a bad example, compared with other historical multiplexing IP?

  • This is going to be very hard to answer without more specifics on the protocol, what parts you might think are particular new (or not), etc. As you've currently framed it, it basically requires someone to read the whole protocol spec to answer it.
    – Luis
    Commented Sep 22, 2012 at 18:30
  • @Luis, interesting thought. I'd surmised that some of the people reading there would be familiar enough with networking IP to offer an answer off-hand, or need little research. Maybe it will take a while. I'll add to this question as I go along. Commented Sep 23, 2012 at 3:50
  • 2
    There are certainly protocols which can multiplex multiple connections over a single TCP connection. I believe SSH has that feature. Commented Sep 24, 2012 at 10:07

3 Answers 3


We wanted to make sure the protocol was as widely deployable as possible. We knew that patents are encumbrances on technology adoption, so of course we didn't want patents at all. Further, we knew the rules of the standards bodies, and that if the protocol ever were to be adopted, patents are just a headache and a barrier.

Given these goals - patenting SPDY or holding back the IP would have been a horrific mistake.

As an example, consider the FAST TCP Congestion Avoidance protocol for TCP . It's generally well regarded as a promising and potentially better algorithm than what is used today. But it's patented. So nobody uses it. It's not "better enough" to overcome the obstacles of the restricted IP.

  • I think this answer is valuable domain knowledge - but it doesn't address the fundamentals of whether SPDY could have been patented. No public standard is patented/enforced, e.g. bluetooth – that's a secondary matter. You also sound like someone involved in the development of SPDY. thanks for the efforts. Commented Sep 25, 2012 at 17:40

Did Microsoft file aggressively, or defensively?

Those patents long predate SPDY, so the question of "aggressively or defensively" is moot. I think Microsoft, as a participant in the IETF standards process, is simply following it's duty in disclosing any patents they own that might have a bearing on this protocol. Given how broadly the claims are written up, it seems difficult for most application protocols to not infringe these patents.

So, to answer your questions in the title, there is some broad prior art, but the specific optimizations of SPDY could still have been patentable. By now, however, I'd guess the bar date to apply is long gone.

Are patents like 6,604,144 and 6269403 a bad example, compared with other historical multiplexing IP?

Your "historical multiplexing IP" patent has almost nothing to do with the Microsoft patents. The historical patent relates to "Inverse Multiplexing", which is achieving higher bandwidth transfers by splitting up and sending data over multiple lower-bandwidth channels in parallel. This is, of course, different from the multiplexing in SPDY and the Microsoft patents, which is combining multiple response objects into a single data connection.

So nobody uses [FAST TCP].

Well, Akamai recently bought FastSoft, the company commercializing FAST TCP, so apparently, no; one of the largest CDNs ever will probably be using it.

The moral of the story, I guess, is that with any innovation your choices are mainly:

  1. to make your innovation as widely available for use as possible, you'd want to release it free of IP and other encumbrances; or
  2. to make money off your innovation, you might want to protect it with IP rights, such as patents and copyright.

Either is a valid choice, and there certainly are lots of variations in between.

  • Thanks for your clarification on [inverse]-multiplexing. I was hoping to get some perspective on he IP in that area compared to SPDY. Commented Sep 25, 2012 at 19:52
  • One of the caveats when searching for prior art or related patents is that the same term can have different meanings, especially in different contexts and in light of the specifications. This gets even worse considering that "patentees can be their own lexicographers".
    – kinkfisher
    Commented Sep 26, 2012 at 5:19

I will attempt to answer in brief.

First-off on protocols. Protocols as such cannot be patented [good reference required], just like the case of standards. Standards cannot be patented, but the catch is, technologies used in a standard could be patented. An example is the MPEG-2 standard that incorporates many patents. So, SPDY protocol as such cannot be patented, but the implementation (software) of the protocol can be (considering, we are talking about the US here).

Despite armchair opinions that SPDY should have been leveraged with IP – could it have been?

Sure it could have been leveraged as a proprietary IP, say like a tradesecret, may be in the form of encrypted or obfuscated code. The article says, the protocol should have been kept proprietary. I think there is subtle difference between proprietary and patented. As said earlier, proprietary could mean a trade secret, e.g., the famous Coca Cola formula or even for that matter Google's search algoritms, and proprietary may not necessarily mean a patent. The information pertaining to a patent becomes publicly available (so not essentially proprietary), only restraint is that you cannot implement it without the permission of the patent owner before it expires.

Also, I must say that in my opinion making it open source was more beneficial. And beneficial to us, the public, since I feel the open source community has some pretty darn good developers and open source software comes free to the user.

Did Microsoft file aggressively, or defensively? Are patents like 6,604,144 and 6269403 a bad example, compared with other historical multiplexing IP?

Honestly, I don't know. Somebody else has to take this up.

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .