This blog post mainly comprises of a message I just sent to the Salmon Protocol mailing list, but I have expanded on it a little.
Salmon Protocol is actually one of the most exciting things that will be used by Google Buzz & theoretically could dramatically make huge changes to the way communication flows around all kinds of web content, whether comments on blog posts, votes & likes on Youtube videos, maybe blog posts themselves, video replies etc.
It designates the way notifications are sent “upstream” just like Salmon swimming up a river to a specific end point.
Many pundits are looking on Google Buzz as some kind of final application, whereas in many ways it is purely an introductory interface to many information protocols that in many cases are still being roughed out by programming geeks who will define how the web will work (the way various apps talk to each other) for the next decade.
I believe what I have stated about end points is important, but might need an example.
- Someone comments about one of my blog posts on Buzz
- I write a blog post linking to someone
- That person has a video from Youtube embedded in their post
- That YouTube video is a reaction to someone else’s YouTube Video
- That YouTube Video was just a syndicated copy of a video originally posted on a blog post as a mp4
If that comment on Buzz is a “Salmon” how far up the stream should it flow as a final destination?
In theory the chain might never end, wheras the one definitive end point is a profile owned by the person making the comment.
Here is what I wrote last July which might give you an enhanced perspective about what I feel is exciting about “buzz”.
The Future Of Commenting And Aggregation
An even more radical approach would be to totally get rid of “comments” as a unique entity, and many other social sites for that matter, and have only unique personal streams of media, long or short form, video, pictures, text or a mixture, and what appears on other sites, whether on a blog as a comment, or on Twitter, Youtube or an social site would just be a syndicated copy of your original content. Just one permalink for the original content, with full ownership and privacy controls over who could see it.
In many ways Youtube is just a video feed reader where you syndicate your unique video, and you should link back to the original source, and get the original source ranking :)
It is not just “Salmon Protocol”, there are all kinds of other things going on underneath that need to be understood, and to be honest my own level of understanding of many of the intricacies is a little vague.
Here are a few reference links
Here is what I posted to the Salmon discussion group (I will add a link if it gets through the moderators)
I noticed this Self-Salmon proposal
From a content ownership perspective, the default endpoint should be SelfSalmon and all other endpoints should be for syndication purposes, under user control.
If a user migrates from one endpoint to another, say from a Google service to a Facebook/Yahoo/MSN service, their Salmon moves with them.
A user may even access their Salmon through multiple interfaces of interlinked IDs
I would also like to reference a few conversations from the search community
Do content creators own the comments?
How about the people the content creators link to?
Or the people they link to?
Or the YouTube Video that they embed that they are all discussing?
Salmon could flow upstream with no logical end point.
Ultimately the only logical primary endpoint is the reference that is unique, the comment creator = SelfSalmon
There is in many ways also an issue with multiple identities & privacy which I discovered from user complaints about Disqus
I am not sure whether this is still an issue, but Disqus was tying together anonymous comments based upon an email ID and displaying them in public, thus it was possible to tie comments on a Political blog under one pseudonym with those on a tech blog using another.
Thus a need for personal end points being able to provide anonymous private notification keys for each interaction.
I realise that Magic Signatures is intended to address this problem, but can it also in some way be extended as a unique permissions mechanism for email delivery?
If an email address was actually a unique Magic Signature, it could also then easily be revoked by the owner similar to providing one-time disposable email addresses.
From a purely email context, a Magic email address would work for only the designated sender to the designated recipient. If sent from a non-designated sender, such as a replyall to a CCed email, there would need to be some kind of authorization/permission & update layer.
These thoughts come from looking in some way to use the provision of email addresses by Facebook through the API and the onerous need to update sending permissions on multiple email service provider platforms, whether correspondence, sales messaging, transactional email or just a comment subscription.
Within the offical spec for Salmon there is effectively a canonical reference for the comment originator
POST /salmon-endpoint HTTP/1.1 Host: example.org Content-Type: application/atom+xml <?xml version='1.0' encoding='UTF-8'?> <entry xmlns='http://www.w3.org/2005/Atom'> <author> <name>John Doe</name> <uri>acct:firstname.lastname@example.org</uri> </author> <content>Yes, but what about the llamas?</content> <id>tag:aggregator-example.com,2009:cmt-441071406174557701</id> <updated>2009-09-28T18:30:02Z</updated> <thr:in-reply-to xmlns:thr='http://purl.org/syndication/thread/1.0' ref='tag:example.org,1999:id-22717401685551851865'/> <me:provenance xmlns:me="http://salmon-protocol.org/ns/magic-env"> <me:data type='application/atom+xml'> PD94bWwgdmVyc2lvbj0nMS4wJyBlbmNvZGluZz0nVVRGLTgnPz4KPGVudHJ5IHhtbG5zPS dodHRwOi8vd3d3LnczLm9yZy8yMDA1L0F0b20nPgogIDxpZD50YWc6ZXhhbXBsZS5jb20s MjAwOTpjbXQtMC40NDc3NTcxODwvaWQ-ICAKICA8YXV0aG9yPjxuYW1lPnRlc3RAZXhhbX BsZS5jb208L25hbWUPHVyaT5hY2N0OmpwYW56ZXJAZ29vZ2xlLmNvbTwvdXJpPjwvYXV0a G9yPgogIDx0aHI6aW4tcmVwbHktdG8geG1sbnM6dGhyPSdodHRwOi8vcHVybC5vcmcvc3l uZGljYXRpb24vdGhyZWFkLzEuMCcKICAgICAgcmVmPSd0YWc6YmxvZ2dlci5jb20sMTk5O TpibG9nLTg5MzU5MTM3NDMxMzMxMjczNy5wb3N0LTM4NjE2NjMyNTg1Mzg4NTc5NTQnPnR hZzpibG9nZ2VyLmNvbSwxOTk5OmJsb2ctODkzNTkxMzc0MzEzMzEyNzM3LnBvc3QtMzg2M TY2MzI1ODUzODg1Nzk1NAogIDwvdGhyOmluLXJlcGx5LXRvPgogIDxjb250ZW50PlNhbG1 vbiBzd2ltIHVwc3RyZWFtITwvY29udGVudD4KICA8dGl0bGUU2FsbW9uIHN3aW0gdXBzdH JlYW0hPC90aXRsZT4KICA8dXBkYXRlZD4yMDA5LTEyLTE4VDIwOjA0OjAzWjwvdXBkYXRl ZD4KPC9lbnRyeT4KICAgIA </me:data> <me:encoding>base64url</me: <me:alg>RSA-SHA256</me:alg> <me:sig> EvGSD2vi8qYcveHnb-rrlok07qnCXjn8YSeCDDXlbhILSabgvNsPpbe76up8w63i2f WHvLKJzeGLKfyHg8ZomQ </me:sig> </me:provenance> <title/> </entry>
The canonical ID for the content would be
The problem comes when destination sites lay claim to ownership of a particular comment and control how it is syndicated.
From many of the code examples I have seen, if someone originated a comment whilst visiting my blog, the above tag would be.
Effectively content is “owned” by whichever platform is being used to create it, rather than a stated intent that the comments are somehow portable owned by the creator of the content, and published under license.
All my “tweets” on Twitter currently exist on a Twitter permalink. There isn’t an effective method to migrate that content to a different platform, though it is extensively syndicated.
The thing is, as an active participant in conversations across the web, I want to be able to control things from a single “web cockpit” which currently ends up being tons of email notifications in Gmail, many of which end up in spam folders.
It could well be that I don’t understand the implications completely, or the geeky side of programming this, but it seems Salmon isn’t yet user centric but portal centric.
User centric would allow:-
- Migration between interface platforms
- All content interactions possible, but not restricted to a single interface platform, thus I might start an engagement posting a comment on a blog, but continue all the discussion from within my current web cockpit of choice
- Complete control over syndication – if I don’t want my “Salmons” travelling upstream, possibly out of context, I should have complete control over that – I should be able to limit who can display my content on an individual Salmon basis, with the possibility to revoke access to individual sites
- Complex threading, not just branching within the interfaces – a single “salmon” whilst in response to one piece of primary content, such as a blog post, might also link through to 3 or 4 other pieces of content.
Imagine what you can do with a blog post and pingbacks/trackbacks with interlinking with other blogs. In theory that capability should be within every salmon created. In some ways that is a possibility within forum software, as in the past I have received multiple pingbacks or trackbacks from mentions of my blog posts on forums.
- Email effectively gives me a universal notification method, and in some cases even a method to respond, such as Posterous, Disqus comments, even some WordPress plugins. Salmon & whatever platforms are used to interface with it needs to have the same level of flexibility.
- Visible canonical links within the code whereever a comment is displayed, either for humans or search engines – if I happen to be commenting on a blog post, that is linking to another blog post, which embeds a YouTube video, there needs to be some kind of visible notification of the reason my content appeared where it did.
Within all of this what I don’t really get… is the “scope”.
Salmon appears to be intended for content reactions, but ultimately even blog posts and videos are often reactions. Blogs notify each other using trackback & pingback, and Salmon is in many ways the next stage.
You wouldn’t expect a whole blog post to be imported as a reaction to someone else’s blog post, thus the same could be true for a long comment, or any comment.
Surely all that is needed on a destination site where a comment is left is a link to the canonical source of any reaction with the data displayed using Ajax.
From an efficiency point of view it wouldn’t make sense to have to poll 100 different canonical storage locations for the comment information they contain, thus there is a need to store a referenced comment locally or at a hub, or to have some level of caching, but that doesn’t mean an endpoint should have a permanent copy of whatever content they choose to display from their upstream and downstream, whether it is a direct comment reaction, a blog post or video.
With a service such as Seesmic (the video service not the Twitter client) a comment reaction after all is in many ways just as valid as a tweet or a direct comment.
Any XML data structure or feed attached to a piece of content possibly shouldn’t consist of much more than a list of permalinks where the canonical version of the reaction is held.
The overall implications for search, based on what is currently within the Salmon Protocol are probably complex enough without adding my input above to confuse things even more. My hope is that we will end up with one canonical version of all content, controlled by the content creator, but with Salmon we might end up with the same comment on 100 different sites with no onus on the sites to display any edited updates, and the possibility through moderation for further editing of content.
Google might be able to work out what is happening, but comment authors probably won’t.
Also I hope long-term it will be possible to upload a video to Youtube and then define the canonical version as a video on a domain I own, and pull in reactions, but that isn’t necessarily in Google’s interests.
That is the end of my brain dump, your head can hurt just as much as mine now.
(p.s. if you get a white screen when you leave a comment, the comments are going through I just have to track down a bug somewhere)
Here is also a link to any discussion on Google Buzz for this post