Not that I would normally post when someone covers us. But today on techcrunch some interesting things said. Greg Linden of Blogspot did a great job of pointing out the difficulties in dealing with attention data in real-time. Let me

In late 2000, my team started working on an internet marketing technology designed to efficiently collect, organize, analyze, and pull metrics off of hundreds of thousands (to millions) of meta-data transactions per minute. That system was up, fast, scaled nicely, and proved its point by providing marketing lift and supplying outstanding real-time metrics without harassing the user! But… right technology, wrong time…

Today, that system (and its dozen man years of head start) is what we have based our attention streams on. The foundation of our system (client through server) is based on attention streams. We use it on the client to synchronize (and soon client side prioritize) articles among the multiple views (IE, Outlook, more…) we provide. We use it on the server side to synchronize client to server (to mobile). These efficient attention streams are what we will use (with permission and privacy protected) to aggregate users to make relevant recommendations (at the feed and article level), and add community based prioritization of your existing articles, and both timely enough to support today’s requirements.

T.

Findory has been watching what you read and what you click on and then prioritizing feeds and posts according to the reader’s interests since we launched in Jan 2004.

The hard part is the personalization. Generating high quality, relevant recommendations in real-time at scale is a real challenge. Most of the work on scalable personalization algorithms is still in the research stage.

Attention is a huge problem and only going to get worse. We are overwhelmed with information flooding at us in our daily lives. We need a way to filter, prioritize, and focus. Personalization can help. In the future, all information streams will be personalized.