Microsoft Tafiti

» see tafiti.com as it lived, on the wayback machine

~ the obituary ~

i missed tafiti. by august 2007 my browser tab was already google, not a silverlight demo of live search. tafiti shipped at tafiti.com on 22 august 2007 as a way to show off what silverlight 1.0 could do in a real product, with microsoft's web-search api wired up underneath. the word means "do research" in swahili. that was the only thing about the project that felt like it had a clear narrative.

the product itself was a thin layer of animation over a search engine almost nobody in europe was using. the page split into two big visual regions. on the left there was a vertical carousel you could spin through different result types like web pages, images, news feeds or books from the live book search index. on the right there was a virtual wooden bookshelf where you could drag any result onto a shelf to save it for later or share the shelf with a friend by url. ranked results sat in the middle as an animated card stack.

anyway, the technical context was that silverlight 1.0 had just gone gold in september 2007 and microsoft needed showcase apps to convince web developers that the plugin was a real alternative to flash. tafiti was one of three or four big public demos that came out around the same launch window. it had been built by the platform incubation team inside the windows live search org, mostly as a what-if. the marketing rollout treated it as a serious product, the engineering reality was a hackathon prototype that microsoft dressed up enough for public release.

~ the case file ~

~ what it tried to do ~

the design pitch was that search results should not be a static list. you should be able to spatially arrange them and come back later, then share the set with someone else by passing them a url. the bookshelf metaphor was meant to make that obvious. each shelf had its own url, either public or private, with thumbnails of the result cards you had dragged into it.

the deeper bet was that browser-side animation could change how people thought about search. if you could spin a carousel between web pages and images with one fluid gesture, you might compare result types in a way the ten-blue-links page did not let you. nobody really used the product long enough for that hypothesis to get tested. honestly, the kind of person who landed on tafiti.com in 2007 had probably typed it in to try a silverlight demo, not to actually look something up.

~ how it actually ended ~

this part you had to be inside the microsoft search org to even notice. in july 2008 microsoft released the tafiti source code on codeplex under the ms-pl license, which was usually a signal that the team was done iterating on it and wanted developers to use it as a silverlight reference codebase. the bing rebrand of live search shipped on 3 june 2009. the live search api that tafiti depended on rolled into the bing api over the following months. by mid-2009 the tafiti.com domain stopped serving the app and started redirecting to bing. nobody at microsoft wrote a shutdown post.

~ what carried over ~

the codeplex source survived until codeplex itself closed in december 2017, at which point it migrated to github as a frozen archive. silverlight kept going as a real microsoft platform until microsoft stopped pretending to care around 2015. the search-as-spatial-canvas idea did not really come back. anyway, tafiti is the kind of project that mostly exists in microsoft press releases from august 2007 and in a small set of silverlight reference books. nobody at the bing team in 2026 would remember it without checking wikipedia.

@}-->--

              )
             (_)
            .-'-.
            |   |
            |   |
            |   |
            |   |
          __|   |__   .-.
       .-'  |   |  `-:   :
      :     `---'     :-'
       `-._       _.-'
           '""""""

« previous  ·  random  ·  next »