Viewzi

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

~ the goodbye ~

Viewzi belongs to a short window when serious people still believed Google could be beaten on presentation. It was built in Dallas by Brandon Cotter and Christopher Mancini, who filed the idea as a patent in February 2007 and spent roughly two years building the thing around it. Beta arrived in April 2008, invitation only at first, on a little over two million dollars of seed money.

The pitch was that a text list is one answer shape being forced onto every question. Viewzi would read your query, work out what kind of thing you were asking about, and choose a layout to match. There were more than a dozen of these layouts and the company called them views. Search a band and you got cover art with related musicians. Search anything general and you got a wall of page screenshots rather than titles and snippets.

Then the money ran out ahead of the traffic. Development stopped some time in 2008, less than a year after the public beta, but the servers kept answering. For two more years viewzi.com sat there running code nobody was touching. On 28 December 2010 the site was replaced by a notice admitting the search engine had not been worked on since 2008 and was now closed.

By 2008 the search question in Turkey was already settled and had been for years. Everything went to Google, in Turkish, on a page that loaded on a slow line because there was almost nothing on it. Alternative search engines in that window reached me the way they reached most people, as a link on a tech blog you opened once. That is the honest position on Viewzi. I saw the genre. I never made any of them a habit, and neither did anyone else, which is exactly what killed it.

~ the chronicle ~

~ what a view actually was ~

Each view bundled an interface with a set of data sources and some processing in between. The MP3 view listed audio you could play in place. The album view pulled cover images. Photo and news had their own lenses. The system tried to guess which combination applied, so a celebrity name would return something built around images while a medical term returned something built around utility.

That guess is the whole product, and it is harder than it sounds. Getting it wrong is worse than returning a plain list, because a plain list at least fails quietly. A wrong layout announces the failure at full size across the screen.

~ why nobody came back ~

Screenshot walls are slow. Every result has to be fetched and rendered before you see anything, and in 2008 that meant waiting on a page that Google would have finished in a fraction of the time. The comparison was not close and everyone made it within the first week.

Underneath the speed problem was a worse one. Viewzi was a way of displaying results it did not produce. The index came from elsewhere, so the only thing on offer was the arrangement, and arrangement is the part of search people notice once and then stop noticing. Searchme had the same shape and the same ending. Cuil tried to solve the index problem instead and died faster.

~ the patent outlived it ~

US7899803B2, "Multi-view internet search mashup", was filed on 19 February 2008 with Cotter and Mancini as inventors and Viewzi Inc as assignee. It was granted on 1 March 2011, roughly two months after the site went dark. A patent office does not care whether the company still exists.

It now belongs to Microsoft Technology Licensing, and it runs to December 2028. So the legal description of Viewzi will outlast the product by eighteen years, sitting in a portfolio at a company that never shipped anything resembling it.

@}-->--

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

« previous  ·  random  ·  next »