Zembly

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

~ the obituary ~

In 2008 the social platform was briefly the thing every large software company believed it had to have a story about. Facebook had opened its application platform the year before and Google had answered with OpenSocial. The assumption across the industry was that a huge population of casual developers was about to appear and would need somewhere to work. Sun Microsystems built Zembly for those people and put it into public beta in June 2008.

The pitch was that you never installed anything. You opened a browser and got an editor with a runtime behind it. Whatever you wrote then ran on Sun's own cloud, with no server to provision and no hosting to sign up for. It targeted Facebook apps first and picked up Meebo and the OpenSocial containers after that, along with web apps for the iPhone, which was a sensible target list for that specific year. Collaboration was built in at the file level, so two people could work on the same app the way two people worked on the same wiki page.

Sun announced in November 2009 that the site would close on 30 November, and the notice said plainly that once it was shut down every application and service on Zembly would be unavailable. There was no meaningful export and the platform code was not released, which produced the reaction you would expect from people who had built on it. One developer put two and a half years of work in the past tense in a single sentence.

I was finishing an engineering degree while this was happening, on ttnet broadband, compiling against a JDK with Sun's name in the header of every source file I read. Sun was ambient to anyone writing code in that decade. Zembly was not: it was a product for people shipping Facebook apps in 2008, which was a specific job I never had, and the first time I looked into it properly was to write this.

~ the case file ~

~ the slide ~

The timing was fatal in a way that had nothing to do with the software. Oracle agreed to buy Sun in April 2009 for $7.4 billion, and the deal then sat for nine months waiting on the European Commission, which was worried about MySQL. During that stretch Sun could not start anything and had every reason to stop things, while the company was losing money quarter after quarter and cutting thousands of jobs. A browser IDE for third-party social apps with no revenue attached to it was never going to survive that kind of review.

The market assumption underneath it also failed. The wave of casual Facebook app developers did arrive in 2008 and mostly left again by 2010, because Facebook kept changing the rules and eventually shut off the viral invite mechanics that made those apps spread at all. Attention moved to the iPhone's native store, where the price of entry was Objective-C and a Mac rather than a browser tab.

~ what it was right about ~

Everything Zembly promised is now normal. You write code in a browser tab against a hosted runtime you never provisioned, and somebody else can be editing the same file while you do it. That description fits Replit and CodeSandbox, and by now the built-in editors of the cloud providers themselves. Zembly shipped the idea in 2008 from inside a company that was being sold, which is how being early ends up counting for nothing.

Oracle closed the Sun acquisition on 27 January 2010, two months after Zembly went dark. Zembly at least got a shutdown notice with a date on it, which is more than some of Sun's other cloud work managed.

@}-->--

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

« previous  ·  random  ·  next »