The story I wrote this morning on Medium (see details of my experiment) has blown up. I’ve been looking at referrals and stats and tried to determine where the traffic is coming from.
The initial response I got was probably due to my own Twitter account and Designer News. But that accounts for fewer than a thousand of the ~9,000 views the post now has. Twitter has now referred far more than that, but it wasn’t the source, it’s the result of the number of readers.
I can’t confirm, but I think that the majority of the views are coming from Medium’s own recommendation engine.
I get emails every day from Medium that tell me what folks I follow on Twitter are recommending on Medium. I probably read that email more than the average promo-email I get, if not as much as my staple newsletters I rely on every day.
In addition to the emails, other people on Medium itself get “Read Next” notices that relate to them, and the Medium home page is catered based on who you follow on your connected social networks.
Finally, the “Top Stories on Medium” further drives traffic to stories that are already performing well.
A recommendation network for WordPress.org
How could this look for WordPress.org? That’s what I have started thinking about. Obviously WordPress.com can do something similar (and does) with their “Likes” functionality.
But I would be really interested in something similar for WordPress.org. In all likelihood, this would require someone like Automattic (aka in Jetpack) to drive it, as it would require significant server resources.
I think the WordPress dashboard could be a really interesting place for this kind of functionality.
Essentially, the dashboard could be a news feed similar Medium’s home page. It could be catered per user based on who they follow in various social accounts.
This is more than a Reader, where it would only show my friends’ posts. It offers broader reach, because it shows me what my friends are posting plus what my friends are recommending (and reading).
So, now I can have a pretty highly catered reading experience right there in the dashboard of my own websites.
In a world where RSS is falling off, social networks are often too noisy to keep up with constantly, and we’re hungry for content curated for us, this could serve us quite well.
Additionally, it’s a no-brainer for publishers to opt-in to a system like this. It helps you get the word out about your posts. Think how much faster a new blog (self-hosted) could gain an audience if it could have a network effect as big as Medium, Tumblr, or WordPress.com.
And the front-end cost is minimal. All I need is a recommend button on my posts.
I’m just spit-balling, but I haven’t seen a really great example of something like this for self-hosted websites. I’m curious what we could dream up,
This is a neat idea and something I thought about back in 2007 when a few different services launched aimed at linking together independent sites. CLIQ and BlogRush are two such services, both of which are in the graveyard.
Jetpack already has a related posts module and we know that it works by mirroring a site’s content on the WordPress.com servers in order to do the relationship processing. Do you think the module could be extended to show related posts network wide? Maybe it would be better off as its own module?
Then again, if it’s powered by WordPress.com/Automattic, that limits the amount of related content to WordPress.com and .org powered sites, is that enough or too limiting?
I’d imagine something closer to the Likes model than the related posts model.
“6 of your friends like {x post}”
A grid of posts prioritized by who in your network liked them, available in the dashboard.
I dunno… just a thought 🙂
When you say “who in your network liked them”, which network would that be? People you follow on WordPress.com which in many ways, is a social network? How would you follow independent people or sites? You can add .org sites to the reader but that’s to see every post by that author or on that site.
I’m just brainstorming along with you 😛
I much more imagine it coming from Facebook / Twitter, which I’d have connected to my blog (and Jetpack). But theoretically .com could be as well, it just wouldn’t have the reach… because I’m less likely to actively “follow” someone on WordPress.com than Twitter.
I can totally see something like this getting rolled into Jetpack, but it’d be nice if someone took a crack at throwing something independent together.
A MVP shouldn’t be terribly hard to construct. Just back the service with an Elasticsearch cluster — which, if I recall correctly, Automattic is using for its related posts feature in Jetpack & on WordPress.com — and you’ve got a stew going, baby.
After seeing you tweeting about this yesterday I started thinking about the possibilities and I worked out that this would very possible to do using the JSON REST API that is being worked on at the moment. All you would need is a site where people could submit posts from their .org website and the post content, comments, meta data, etc. would be pulled in via the API. This would mean that no content is stored on the site (other than cached post content) and all the content would be dynamically generated via the API. You could even allow commenting on the post via the API so the conversation all continues in a single location.
Once you have the post content and all the related data (taxonomies, etc.) you could easily build out a recommendation engine similar to Medium’s based on categories, tags, what your friends are posting, etc.
An MVP wouldn’t actually be too tough, but a solid product that works as well as Medium and has equally consistent quality would be a lot more work.
I’m keen to work on something like this is anyone else is interested in collaborating.
Might be worth talking to ManageWP.org to see if they’d be open as being a base of content to pull from, if we were going to take Hugh’s approach.
Keep in mind, I’m not talking about WordPress news. I’m talking about it just being in WordPress. All sorts of blogs would have their own dashboards catered to their social audience.
I had similar feelings, about a recommendation system, or a platform where we could refer products & good articles. A little bit research ended with results like ManageWP.org but there was still a gap in the product’s space. So, I went ahead and built WPMetaList.com to serve the purpose.
But I’d love more innovative ideas.