A Peek at the Technology Behind Livestand from Yahoo!

Yahoo! today announced Livestand from Yahoo!, a digital newsstand that will launch initially for tablet devices in the first half of 2011. To coincide with this announcement, the engineering team wanted to give you a sneak peek at the technologies we are using to develop Livestand. In some cases we have decided to experiment with new technology, in other cases we are leveraging existing technologies. This is a new area of digital media that is still figuring itself out (as anyone who has debated Native versus HTML 5 will tell you), so we would love to hear your opinions and suggestions on our approach.

image of Yahoo! Today from Livestand by Yahoo!y-today

A great product starts with an awesome user experience. As a mobile experience, the product needs to be highly optimized for speed, memory usage, and hardware capabilities. We also need flexibility in the look and feel so that our advertising and publishing partners can easily express their brands. To solve for this, the team has developed a native/HTML 5 hybrid approach that transparently offloads certain work to native code while keeping the flexibility of the web programming model. Pulling this off, and making it fast is a challenge, but since we are using parts of the YUI Library, the performance optimizations necessitated will benefit the broader tech community. As we develop best practices for developing, debugging, profiling and optimizing using this hybrid approach we’ll share more with the developer community.

Once we decided to take a hybrid approach,, we needed to tackle how to hook it into our content. We chose YQL because it offers the most flexibility for publishers who want to distribute their content and services. YQL provides Livestand with a data transformation and caching interface to our vast digital content and services inside Yahoo! (which includes Sports, News, Finance, and Flickr), and can map to almost any web service or content source on the Internet. It gets really fun when you combine YQL with some of the native libraries we are developing for Livestand. Depending on the characteristics of the content or service, the platform can make intelligent decisions on what content to store on the device and when content should be updated. As Livestand becomes available in the first half of 2011, we will share more information on how publishers can make their content available.

Screenshot of content categories from Livestand from Yahoo!y-categories

How do we get the right content distributed to the right users? How do we let those users engage with that content? What about scale? Yahoo! has been building a powerful content management system to drive all of its experiences, and Livestand was able to sit on top of that cool tech. This underlying content management system is build on top of Hadoop, allowing us to run sophisticated content analysis algorithms on a distributed computing cluster. For implicit personalization, we are adapting the same Content Optimization and Relevance Engine (C.O.R.E) that we use on www.yahoo.com. For advertising, we are plugging in to the same advertising APIs that power other Yahoo! products, allowing us to serve more relevant advertising. Finally, many of Yahoo!’s social features (Polls, Ratings, Comments, Sharing, Reviews) are accessed using the Yahoo! private cloud, so conversations and opinions expressed in Livestand can carry over to the PC-based media experiences. Yahoo! is a leader in open source initiatives, and sharing what we learn with the broader tech community, so if you are interested you can learn more here at developer.yahoo.com.

Let us know your thoughts on the technology approach. As we continue to iterate and build out Livestand, we’ll be sure share how things are going, problems encountered and solved, so be sure to check back.