The announced launch of Silverlight 5 has got the developer community all excited about improved media capabilities, MVVM support, printing and 3D, but how will Silverlight adoption evolve throughout 2011. In this blog post I look at historic data and use this to predict a 76% adoption of Silverlight 5 by the end of 2011.

The Silverlight Firestarter event last week (December 2nd 2010) saw Microsoft's Scott Guthrie use his keynote speech to announce the launch of Silverlight 5 which will emerge at some point in the first half of 2011. The new features, which my colleague Gergely details in a recent blog post, are geared towards media and business applications, which in my opinion is a good thing. This further re-enforces Silverlight's positions as a plugin for application development, making it easier to differentiate it from HTML5.

New versions of Silverlight are being released at quite a high rate (approximately 1.5 VIPAs), however I feel that feature completeness isn't really an obstacle that prevents people from using Silverlight, as a technology, it feels quite mature already. I think the main obstacle that prevents people from choosing Silverlight is the adoption statistics, i.e. the proportion of people who have the Silverlight plugin installed. With the announced release of Silverlight 5, I thought I would look at the current Silverlight adoption and make some predictions for the future.

There are a couple of prominent sites that collect plugin adoption staticstics, riastats and StatOwl. Currently they give adoptions statistics for Silverlight of 67% and 58% respectively, with the difference between these values most probably being due to the demographic of the sites which they collect statistics from. StatOwl provides pretty detailed historic statistics, using this data the chart below shows the adoption of various Silverlight versions over the past couple of years:

I would expect the total adoption to follow a sigmoid, as per the diffusion of innovations theory, and you can just about see that the adoption is starting to decelerate. To make future predictions I have fitted a logistic function to the total adoption. What I find quite interesting is the adoption curves for each new Silverlight version, which are the components of the total adoption. Each starts with a rapid sigmoid conversion, followed by an exponential 'tail'. Fitting a curve to the total adoption, then normalizing reveals this shape more clearly:

If I take a guess at the Silverlight 5 launch date, based on Scott's announcement that it will be in the first half of 2011, I can extrapolate the above chart to give the adoption curve for Silverlight 5:

The shaded area in the above chart is the extrapolated future adoption. De-normalizing the above gives the following total adoption predictions:

The above chart predicts that Silverlight 5 will have a 76% adoption by the end of 2011, and that the total Silverlight adoption will be 81%.

I will be interesting to see how good these predictions are!

Regards, Colin E.