An enhanced Hunch feed
Starting today you’ll notice a new look to the activity feed on Hunch. The “recommendation feed” combines several longstanding Hunch features — an activity feed, personalized recommendations, machine-based predictions and a follower system — into one comprehensive feed that also includes several new social features.
The Hunch Feed helps you discover personal recommendations made by people you follow, and lets you share your own recommendations among people who follow you. The recommendation feed supports embedded video and audio, hi-res imagery, and a threaded commenting system.
You can “re-recommend” something which appears in your feed, which will in turn push it out to your own followers, creating a potential viral wave of recommendations. You’ll also notice the introduction of a commonly-requested feature: the ability to save something for later. So if you find a book, movie, recipe, handbag, or sprocket that seems intriguing, you can save it for later when you have time to read, watch, make, wear, or buy it.
Hunch’s machine learning algorithm continues to be the core that powers your recommendations
As always, recommendations on Hunch are never “just” recommendations. The recommendations you see in your feed will draw upon Hunch’s Taste Graph to show an algorithmic prediction for how much Hunch thinks you’ll like them. This means that if you’re following Jane, and you love her book suggestions but aren’t so keen on her taste in movies, Hunch will come to understand this over time and will adjust its predictive ratings (and the visual prominence of Jane’s respective recommendations) accordingly.
The previous version of the feed, which we introduced in March 2009, was designed to recognize initial seeding and training of Hunch. This helped populate the site and train the Taste Graph to the point today where it has more than 20 billion correlations, or “edges” in tech-speak.

The previous activity feed focused on populating results by topic
Now that Hunch’s Taste Graph is well-populated and trained, the new version of the feed places emphasis on positive endorsements and more social features:

The revised feed is focused on positive recommendations
But…whom to Follow?
For existing Hunch users, chances are that you’re already following some other people on Hunch. So your existing followers will form part of the basis of your recommendation feed.
The first time you use the enhanced feed, you’ll also be asked to choose some broad areas of interest. Hunch will then propose new people to follow based on those interests. Some suggestions may be Facebook friends, others may be users who are influential ‘taste makers’ in your areas of interest. Soon you’ll also begin to notice some high profile and influential bloggers, companies, brands and reviewers coming on board to make recommendations about their passions, products, causes, and content.
How to recommend something to your followers
There are two ways you can ‘push’ a recommendation to your followers: 1) you can use the ‘Recommend’ icons at the top of your feed, or 2) you can use the bookmarklet to quickly add something to Hunch while you’re browsing another web page.

How your past ratings and pros/cons fit into all this
The enhanced Hunch recommendation feed is designed to highlight positive recommendations/endorsements that you would want to provide to other people. At the same time, we know that many users rated content on Hunch before this change was made. So we’re in the process of merging these two approaches using the following guidelines:
1) If you previously rated something on Hunch as ‘like’ or ‘favorite’ AND you provided some pro/con text, then the item will appear in your followers’ feeds as something you previously recommended.
2) Not yet implemented but coming shortly: If you previously provided pro/con text to an item on Hunch, but you didn’t ‘like’ the item (either because you didn’t rate it at all, or because you rated it ‘dislike’), the item won’t appear in your feed. However, your historic pro/con will still appear on the result page for that item as a ‘review’. The key here is that we are distinguishing between the helpful information in a historic pro/con, vs. the positive endorsement of a current, active recommendation. At the same time, we’re retaining all the valuable content that contributors have added to Hunch over the last 3 years.
Going forward, pros/cons have been replaced by two ways of sharing your opinion about items listed on Hunch. First, Hunch will always ask you to provide a written reason why you’re recommending something to your followers. Second, you can comment on things recommended by other people.
Browse
In addition to using the revised Hunch recommendation feed features, you can still browse Hunch as you always could in order to find specific recommendations on any number of topics.
One thing you’ll notice is that we’ve combined previously separate topics into broader topics, which allowed us to eliminate duplicate results which appeared in a number of topics (a common request).

Streamlined categories and topics on the browse page
We’ll be working to improve the integration of recommendations which appear in the feed vs. recommendations which you’ll find in the browse sections of Hunch. This will include harmonizing how the result pages look. In fact, there’s a lot we’ll be continuing to refine and improve based on your feedback.
As always, we value your input and suggestions. Once you’ve tried the revised Hunch recommendation Feed, we’d love your feedback via a quick, 5 minute online survey. You can also leave comments, questions, or suggestions in the Hunch forum.
