Is the Salmon protocol tasty enough?

Conversations on the social web are mostly performed through comments. But comments are so fragmented! Consider this example:

  • Publisher  publishes a blogpost
  • A regular reader of the Publisher comments on the blogpost
  • Someone else reads the post in  Google Reader and shares it
  • Another comments and reshares the Google Reader item
  • Another decides to share it on Facebook
  • Another comments on above  the Facebook link
  • Another submits the link on Digg
  • Another comments on the Digg link
  • Publisher  has a Friendfeed account and the post appears in his FF stream
  • Another user comments on the FF stream item
  • etc

Obviously the post has stirred some interest and generated a conversation. But the conversation is dispersed in many different places. Publisher looses track of many aspects of the conversation around the post. Commenters also mostly ignore what happens outside their area of interaction with the content: Facebook users ignore the FF commenters etc.

This situation has fired some intense debates. Many publishers think this situation is not  in their best interest as potential traffic to their sites is deflected to an ‘aggregator’. Especially publishers that have a financial interest in their site traffic and do not just want their opinions spread, find this particularly not appealing.

To mend this situation, a group in Google is working on a new protocol that will allow comments to ‘return’ back on the original publisher site. The protocol is called Salmon

Salmon protocol logo
Salmon protocol logo

and you can get a basic idea of its workings  from this  presentation.

Salmon does bring back the comments to the publisher site but it does not solve the publishers’ problem.  As you can see from slide 4, once a comment is back to the publisher’s site, it  is republished back to all its subscribers (including the aggregators). What this would mean is that each aggregator has a full picture of the comments around the post, regardless of origin. From the user’s standpoint there is no need to move to the publisher’s site or to another aggregator for any reason, as the  full picture will be available in  whatever site the user prefers to frequent. The publishers may object it, but in what right? The publishers’ protests imply  they OWN the comments which is hardly the case. The user owns his comments.

But let’s leave aside the publisher’s concern for  a moment. Is slamon a good thing for the user? I would argue it is. He can have access to a discussion in its entirety  without much hassle. And therefore he might be tempted to engage or engage more.

But there is something still missing: the user does not have easy access to his own comments for ALL pieces of content he has interacted with. And he has no control either. They can disappear with a site that closes down. Or in the simplest case, the can be deleted by the site moderators. This is the problem that systems like disqus, intense debate and JS-Kit are aiming to solve. But they won’t. Because it is very unlike that one of them will become ubiquitus.

I think the problem should be approached from another angle. A comment is a piece of content. There is no distinction in form from any other piece of content. They are both text (or audio or video in some cases). What subordinates a comment-content to the original post-content is notional and semantic: the post-content preceded the comment-content and actually the post-content was what aroused the commenters interest in the issue. But the same applies to a post that pingbacks to another post. So a comment is a piece of content and should have independence.

The question is how?

The issue is related to our digital identities: if in the web -to-come we can  have a unique independent central point for our digital identities, this central point could be the originator and hoster of our comments.

A modification of the salmon protocol could easily let this happen: whenever a user comments on a publisher site, the site will send the comment back to the users digital identity home. Likewise, whenever an aggregator receives a user comment, the aggregator sends the comment back to the user home, as well as to the publisher.

I do not think this is difficult to implement although I can predict the frictions about who controls the user’s  digital ‘home’. But that’s another issue.

Read also Louis Gray’s post on Salmon

An idea: blog comments the way

For sometime now, it bothers me that I cannot track the comments of the people I care about and value, in other blogs. Neither can I have a unified picture about what have I commented and where.

This was the main reason I chose to switch to Disqus in my greek blog just a couple of months ago. Since a fair amount of blogs are utilizing disqus too,  the picture looks brighter but far from ideal.

To make things worse, there are competitive solutions to disqus, as well as solutions that do not attemot  to replace the comment system altogether but track the traditional blog comments in one place instead(co.comments comes first in mind).

The problem with all the above solutions is that they are not universal, nor can they be: they are competitive services that want to keep any competitive advantage they have  for themselves. Quite understandable. Especially since no cross commenting standard is to be found.

Yesterday, I came upon this  post of Shey Smith, where he calls for a unified commenting standard, having in mind primarily the cooperation of systems like Disqus, IntenseDebate and SezWho. But these systems are niche play compared to the standard commenting systems of Blogger, Typead and WordPress.

As I played with, it came to me that the microblogging federation idea can also be applied to comments. To see how, let’s examine first what comprises a comment system first:

  • A form of identification (that can support anonymous comments too).
  • A comment text of arbitrary length, related to a post
  • Replies from one commentor to another (in the case of disqus, replies become threads).
  • A feed for comments
  • Moderation of comments

If we were to replace the traditional blog comment systems with a microblogging platform like (where each blog would behaves like a separate server), what we would actually get is something that would accomodate for :

  • A form of identification without anonymity
  • A short text comment/post/update
  • An informal reply mechanism
  • An informal reply tracking mechanism
  • A feed per user and friends, or a total feed.
  • A way of blocking users (this is from twtter, not from

To bring the two worlds closer, we would have to expand the text from 140 chars to any length, bind updates to posts by making every new post create a new update with the trackback  url in it.

Around each blog there is a community of regular, not so regurar and casual users. Most likely each one of them reads or owns other blogs too. If someone owns a blog, his/her blog becomes her place where the user identity is registered/created. For users with no blogs, this can be any other blog of their choice, that will server from then on as their commenting system provider. Each blog owner could choose to follow whomever he likes and therefore track his comments in the blog comment systems federation. Likewise for a user without blog. If a comment is originated from someone that the blog owner does not follow, it will show up in the replies, as replies from users we don’t know can show up in our twitter replies tab.

The ultimate merit of such an implementation is that it converts each blog into a social network too.

What do you say?

Note: As I was about to finish the writing of this post, through a tweet, I was informed of an alternative suggestion too.