Building a better feed August 15, 2011Posted by David Card in Uncategorized.
Tags: David Stewart, enterprise social networking, real time, real-time feeds, SocialText, user communications, user interface
Last week, Twitter retooled its site to offer two Facebook-like activity feed channels aimed at increasing and broadening user participation. Facebook itself did some tweaking to its feed and is reportedly evaluating a bigger overhaul, and Digg updated its news feed. Real-time feed-based user interfaces (UIs) are becoming one of the most important ways of presenting information online, and they are critical areas of competition in social networking and search.
Different companies using feeds reveal UI implementation strategies that tend to focus either on active (user “control panels”) or passive (algorithms) techniques. The winning approach will probably be a blend that leans toward passivity. Consider the following:
- Twitter’s new tabs show activity around the user (mentions, favorites, retweets) and the user’s followed friends. Twitter wants to boost usage by mainstream users and encourage favorite-ing as a simple way to engage users who aren’t necessarily in the mood to post or reply.
- Facebook countered Google’s new social gaming thrust by fine-tuning how players and games communicate (a ticker and less throttling of messages in the feed). Earlier it introduced a new feed “story” type that groups actions based on natural language analysis of related topics.
- Like other enterprise social networking from SocialText, Jive Software, Salesforce.com’s Chatter and SocialCast, Yammer drives user communications via an activity stream. Its feed emphasizes “ambient consumption” of info that’s surfaced to users based on an algorithm that evaluates topic and relationship data for relevance. Platform VP David Stewart told me that tools to embed that stream in other enterprise applications that were announced in May will be available in beta later this month.
- Venerable link-sharing site Digg introduced “Newswire” that enables users to filter and sort links appearing in real time based on things like recency, topic, format and who posted or voted on them.
The best approach: Balance user control with algorithms
Mathew Ingram doubts that Digg’s new features will be enough to help it regain the audience it lost to Reddit and others when it did a poorly received redesign last summer. He’s probably right, but Digg’s latest moves illustrate that adding controls and filters to a feed is mostly for power users. Making mainstream users take active control of information presentation is extremely challenging, usually resulting in adoption in the 5 percent (Facebook Lists) to 20 percent (Yahoo customization) range.
It’s “easier” — from an adoption if not technology perspective — to rely on passive personalization via algorithms that analyze feed content and promote it by guessing it will be relevant to users. That’s what Facebook does with its social graph–powered EdgeRank, and that’s what Yammer is doing, although Yammer doesn’t do any natural language interpretation. Rather, Yammer incorporates user curation by encouraging topic tagging. If Twitter gets users to choose favorite tweets more often, it will have more curated data to power potential feed sorting and prioritizing schemes it might develop.
Meanwhile, advertisers and app developers seek to reach audiences within the feed, where most user attention is directed. Sites that accommodate that desire gracefully aren’t merely caving in to marketing pressure; they’re enabling social media communications that many users will find valuable. But they have to enforce relevance by monitoring user reactions and weighting their algorithms appropriately to avoid crossing the line into spam.
Companies using feed-based interfaces need to strive for a balance between algorithms — which can produce odd results — and user controls that may require too much work from the masses, like lots of tagging or advanced search pages. Simple actions like a Like or +1 button will likely be more popular and are the easy entrée into curation. And the data they produce can, in turn, be funneled back into a relevance algorithm.