Blern is pretty cool, it’s a web-page recommendation tool that attempts to learn from your RSS subscriptions, social bookmarks etc. to provide you with articles that are likely to interest you. I’ve only just signed up and so far it thinks I have an unhealthy interest in subversion, but that’s OK because, compared to most people, I do.

So I got my email of my daily recommendations from Blern with four out of ten posts that covered version control systems (not a problem, I realise I still have to train it) and one post caught my eye. Jim Priest has posted a great article on SVN that I clicked through, read it and wanted to bookmark it for later in del.icio.us.

This was Blern’s first triumph! An article they had recommended that I wanted to read, then wanted to keep and share. Go Blern!

But Wait, It’s Wrapped In A Frame!

I don’t want to bookmark Blern’s recommendation, I want the recommended page itself.

This is the kind of trick that was used to build a “sticky site” in the 1990s and resulted in the kind of bloated portal pages that we still see at Yahoo today. Luckily, as a Mozilla Firefox user, I could right-click and select This Frame -> View Only This Frame to get to the page I wanted. Still very annoying, though.

There Are Lots Of Fixes...

The reason for the frame is obvious - Blern want to know what I like and dislike. So they put some buttons at the top of every page with voting options. I want to give them this information, otherwise they cannot improve their recommendations for me.

Using an IFRAME is not the way to do it. The typical Blern user must be web-savvy since they need OPML files and/or del.icio.us/FriendFeed data to get any hope of having a good guess at recommendations in the first few weeks. This is a demographic that is willing to experiment, it’s willing to give away personal data - it’s GOLD!.

Two obvious alternatives to the IFRAME are a Firefox plugin and cross-browser bookmarklets. Either of these could also be clicked to let Blern know when I like something that wasn’t recommended by them. More information makes for better predictions.

Blern knows which links I have clicked through a link on their email (or my Blern page) and from that it could infer that I like the title and snippet. If I do not click the like or dislike buttons, it could ask me next time - but only for a few items, please! This kind of feedback shows concern for the user.

Blern Learns

The tagline for the website is “Blern learns”, I hope they do. I want them to succeed. They seem to be aiming for the right results without being evil, and that’s exactly what the semantic web needs right now - results for the user.