Calling ASCII a primary source of inefficiency that needs to be "fixed", in the presence of so much rubbish (advertising) being force fed to users by so many websites, is a red flag that this is not truly concerned with efficiency for users' benefit. It may be more concerned with efficiency for those serving up the rubbish. There are other sources of inefficiency that one could argue are much larger sources of inefficiency for end users. Unwanted advertising could be at the top of the list. Resolving domain names of ad servers that serve the user no intentionally requested content, day after day, might be another. True "efficiency" would be embodied by the user receiving only what she requests (the content), and nothing more. A question to ask might be: will a binary protocol make this easier or more difficult for users to achieve?
In a vacuum, absent any surrounding context, binary does trump ASCII for efficiency in transmitting and receiving information over a wire. No contest.
But in the context of the world wide web, which is a visual medium to be read by humans, moving to a binary protocol does not guarantee that the web will become more "efficient" for end users. If I am an advertiser and can more efficiently insert rubbish into the content the end user is seeking, does that make the web more "efficient"?
And who is it that is responsible for 95% of the revenue of the company that designed SPDY/HTTP 2.0? End users? Guess again. The answer is: advertisers.
He doesn't have to make a substantive criticism because (unless I'm mistaken) that's not what he's saying. What he's saying is that it's not a good thing for an advertising company to "control" ("steer" might be a better word) so many aspects of the internet and web[1], one of which is the design and implementation of the HTTP protocol.
But, to feed your unrelenting desire to argue even when you have nothing worthwhile to say, here's a substantive question (I'm not familiar enough with the draft yet to crticise it): the protocol combines several things that are each improvements in their own right, irrespective of the other, namely multiplexing, pipelining, encryption and (header) compression. If we could subtract three of the four, which one would you keep? What is the most needed improvement to HTTP?
Note that the later three of the four I mentioned have been available for many years (pipelining, encryption and compression). But we've seen over the years that none of them has achieved "sufficiently widespread" use.
1. Cache of the web, web browser, mobile and notebook OS, DNS resolution, video codecs, programming language, etc., etc. While they are a very different sort of company, their overall strategy is much like Microsoft: acquire all the world's talented developers and businesses potentially competing for user mindshare, early and often. How can such a monopoly be good for users in the long term? MS tried vary hard to thwart the standards process. Why should we believe that Google's involvement in the standards process should not be similarly suspect?
In a vacuum, absent any surrounding context, binary does trump ASCII for efficiency in transmitting and receiving information over a wire. No contest.
But in the context of the world wide web, which is a visual medium to be read by humans, moving to a binary protocol does not guarantee that the web will become more "efficient" for end users. If I am an advertiser and can more efficiently insert rubbish into the content the end user is seeking, does that make the web more "efficient"?
And who is it that is responsible for 95% of the revenue of the company that designed SPDY/HTTP 2.0? End users? Guess again. The answer is: advertisers.