Darwin and FriendFeed’s “Bounce to the Top” Algorithm

I really enjoy seeing the flow of new content come through FriendFeed. And the best feature is that people’s interactions make content resurface at the top of your FriendFeed page. Human filtering to make the most interesting stuff bounce to the top.

We all enjoy the Likes and Comments that make the most interesting content in our social networks pop back up to the top. And it seems to be pretty simple, right?

{Like} or {Comment} = Content Bounces to the Top

Well, it’s actually a bit more complex than that. The FriendFeed “bounce to the top” algorithm has some interesting behind-the-scenes rules. There is a certain amount of Darwinism to it.

Having noticed that the “bounce to the top” function is not as straightforward as simple Likes and Comments, I posted this comment on FriendFeed:

Sometimes I don’t understand the FriendFeed “bounce to the top” logic. I just added my Like to Corvida’s blog post about email (http://friendfeed.com/e/68e566…), but I didn’t see it pop back up to the top of my FriendFeed screen. What drives the logic?

In typical FriendFeed fashion, there was a really interesting conversation that followed. Mark Trapp, who has really impressed me with knowing stuff, had this to say:

I’ve been tracking the bump logic for a month or so now, and this is what I’ve found out: likes will stop bumping a story after a certain amount of time, and comments will always bump to the top if either a) it’s your story, b) it’s someone you’re subscribed to’s story, or c) it’s a FoaF AND a friend of yours comments on the story AND it’s the first time your friend commented on the story. Everyone else’s comments won’t bump it. For scenario c, Paul had mentioned to me that a fix for it was in the works.

From Mark’s response, one can see a few design principles behind FriendFeed.

1. Comments Are More Valuable than Likes

Comments always bounce something back to the top. Likes stop doing that after a period of time. This makes sense. Likes are easy. They are quite valuable as signals for the “interestingness” of content.

But Comments are King in the FriendFeed system. Comments mean someone has taken the trouble to express themselves. And that engagement draws the content creator and others into the conversation.

As reviewed in an earlier post here, Comments don’t necessarily mean the content has information that’s good to know. Some bloggers are just really good at stirring things up.

But the supremacy of Comments in the “bounce to the top” algorithm clues you in to the FriendFeed founders’ orientation. Conversations are ranked higher than good content discovery.

I Like that.

2. The Value of Likes Decreases with Time

In my comment, I referenced that my Like of Corvida’s blog post failed to bounce her entry back up to the top of my FriendFeed. I believe the FriendFeed entry of her blog post was only an hour old at the time.

My initial thought for this is…why? Doesn’t a Like mean others should check out the content, even if it is older?

Good content doesn’t have an expiration date.

But there are other ways content resurfaces in the FriendFeed system. Google Reader shares. Tweets. Direct posts. Bookmarks. Unlike comments, which only become visible when the entry to which they’re attached bounces to the top again.

So the multiple ways content can resurface reduces the need for Liked content to bounce to the top again after a while.

Wonder how long Likes have an effect on the “bounce to the top” algorithm?

3. Comments = More Reputational Skin in the Game

This is the conclusion I draw. Likes are easy, and have multiple purposes, as described by Mike Fruchter. If someone were to call you out for an errant Like, you could always say, “I was just using the Like to bookmark the entry to my personal feed.”

But comments are more visible, and you are much more accountable for them. When people post comments, they are adding to or withdrawing from their reputation account. And in social networks, reputation is huge.

So the emphasis on comments makes sense. The threshold for adding those is higher, and its effect should be greater.

Survival of the Fittest

Only the strongest content seems to survive in the FriendFeed Galapagos Islands. An initial rush of Likes puts the content into the stream of many, many users. Strong content will get this initial rush.

Then the content has to evolve. If it’s going to get that ongoing attention, Likes are no longer going to cut it. The content has to attract Comments. These Comments sustain the content in a sea of new, competing content.

Of course, just like Darwin’s animals, eventually even the well-adapted content will perish. Fossils for our future searches.


See this post on FriendFeed: http://friendfeed.com/search?q=%22Darwin+and+FriendFeed%E2%80%99s+Bounce+to+the+Top+Algorithm%22&public=1


About Hutch Carpenter
Chief Scientist Revolution Credit

5 Responses to Darwin and FriendFeed’s “Bounce to the Top” Algorithm

  1. marco says:

    Hutch –

    great stuff man, reminds me why I love FriendFeed so much.

  2. markdykeman says:

    Great analysis from you and Mark Trapp. I had been wondering how the “bounce” works, too.

  3. Wow, that’s some deep thought going into the design, although not unlikely for the level of app friendfeed is. Love your explanation of it by equating it to darwinism

  4. Michael says:

    Thanks for the feedback about my question on Likes on FriendFeed. I didn’t know so much was going on behind the scenes.

  5. Andrea Hill says:

    Wow, I don’t spend enough time on friendfeed! 🙂

    The “comment outweighing likes” makes sense to me. It’s pretty easy to like things discriminately, whereas actually leaving a comment takes time, as you say, and makes the site more engaging. Good strategy on friendfeed’s account!

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

%d bloggers like this: