email: ceberhardt@scottlogic.co.uk
on Google+
Around one year ago I made the prediction that Silverlight would have an overall adoption of 81% by the end of 2011. The adoption statistics are now available and indicate that my predictions were wrong by more than 10%, with Silverlight adoption reaching just 67% at the start of this year. So what went wrong?
In my previous blog post I assumed that Silverlight adoption would follow a sigmoid curve as per the Diffusion of Innovation theory. Taking the adoption statistics from StatOwl, I used Excel to fit a logistic function to the data in order to predict the future adoption, arriving at a projected adoption figure of 81% at the end of 2011. However, in my curve fitting it looks like I made one important mistake – I assumed that Silverlight adoption would eventually reach 100%!
By adjusting my Excel spreadsheet to allow a sub-100% adoption I achieved a much better curve fit and the 2011 predictions were reduced from 81% to 75%.
However, data from throughout 2011 seems to indicate that the rate of Silverlight adoption is falling (I think we all know the reason why). Including last years data shows a very good fit to the logistic function, with the rather gloomy outlook that the adoption is asymptotically tending towards 73%.
The graph below shows the adoption statistics reported via StatOwl from 2008 through to 2011, together with my (corrected) predictions based on the data up to 2010, and my predictions based on the complete set of available data.
This doesn’t compare well with flash where the adoption of each new version is faster than the last. With Silverlight adoption statistics tending towards 73% the outlook doesn’t look too great. Is this a technology we should still be investing in?
Personally, I think the answer to this question is “yes” for a couple of reasons:
Whilst the Silverlight plugin might be losing relevance, Windows 8 Metro and Silverlight for Windows Phone 7 are both XAML languages with a similar API. The skills you learn whilst developing Silverlight or WPF applications are certainly going to be relevant in the future.
Businesses have an investment in the skills of their employees, but just as importantly, they invest in the code that they write. Whilst there is some comfort in the ability of their employees to easily adopt new technologies, the cost of migrating a code-base is always going to be much greater. It is clear that Silverlight is no longer going to be the Flash-beating plugin that would be the de-facto choice for interactive content on the web (Flash, and the plugin model in general, is in trouble). However, Silverlight is now a very mature platform that is an appropriate choice for delivery of complex business functionality either within an intranet, where you can control plugin installation, or where the business value of the application being delivered ensures that the 30% without the plugin will install it in order to use your application.
Now, more than ever, is a time for developers to diversify and learn new skills.
Regards, Colin E.
Tags: adoption, News, silverlight
[...] however, I cannot refuse to acknowledge Silverlight’s future. But nor can I justify the higher development costs of HTML4/5. Sure, the naysayers have had their [...]
Thinking from the global picture, I think that businesses will hit a wall in finding skilled people to support all the types of technologies that get born, shine and die before people can spell their name correctly. Businesses are accumulating data and information in more and more ‘new’ technologies without thinking who will be able to read that code 8 years down the line. It may be true that programs are not always the driver and that DATA is king, but the amount of ‘business logic’ invested in the code is not to be ignored. I guess what we really need is a new technology that has a solid core and transparently replaceable ‘flavors of the day’. Those ‘flavors of the day’ should be dumb and hold zero business rules. The real meat should be developed in the solid core.
Hi Emmad,
+1 to this. The sheer number of different technologies is growing faster each day. There is nothing wrong with the ‘new’, I am all for innovation. But businesses need a solid, stable base to build on.
Colin E.
What are all these apps being developed on? Ipad is not developed on an Ipad… the desktop/laptops are where they are developed. The person who wants to view the info on Ipad can use html(sharepoint). Hopefully all that will happen is we carry on developing C#/xaml – and a wizard will create wpf/Silverlight/winrt/knockout. I use a tablet – its a big phone. Completely useless for development as is an ipad/phone or any Tablet.
[...] The Gradual Decline in Silverlight Adoption [...]
Silverlight is a GREAT technology if all you want to do in the enterprise is run on PC’s and Windows Phones. But this is not going to be the case. Enterprise adoption of the iPad and other non-Silverlight compatible devices is on a steep incline. Imagine having to explain to your boss’s boss a year from now how that great new Silverlight app that you and your team spent the last 12 months developing won’t run on his iPad. Microsoft’s failure to push Silverlight onto every platform (including it’s biggest competitors’ like Apple and Google) have spelled it’s demise as a future all-encompassing solution, similar to Adobe’s failure to get Flash on iOS.
Totally agree, Silverlight is great, but has failed to keep up with the changing way in which we interact with the internet.
“Imagine having to explain to your boss’s boss a year from now how that great new Silverlight app that you and your team spent the last 12 months developing won’t run on his iPad.”
Is it really right to expect every kind of app to just run on a tablet device? Sure, many simple apps should ideally be able to, and parts and pieces of more complex apps could be displayed in a tablet-friendly way. But I would argue high-productivity apps, without having a modified user experience, wouldn’t be used on a tablet. In the case of my boss’s boss, he’s going to what to consume information, but not necessarily manage inventory transfers between facilities, create POs, manage the weight of truckloads, etc. So should the limited needs of consuming some kinds of information dictate the technology used for desktop-scenarios? I’m asking this with sincerity because I want to learn…I may not be thinking about this correctly.
Following this argument, couldn’t one could make the case that you shouldn’t develop an iPad app in Objective-C because it wouldn’t run on WP7 or Android?
I have developed on MS platform for the last 15 years. It has been my experience that it is not wise to stray away from MS core products (VB, C#, Windows ops, ASP, SQL Server). MS has this bad habit of getting all pumped up about some new technology, pushing it, and then abandoning it. Usually MS rushes the product to market. The product is not fully baked and it takes several releases before it become truly useful. Building a LOB app means that you can expect the app to be long lived ( > 5 years). Therefore you cannot afford to get hooked into the MS conveyor belt of products. I liked the SL idea but was reluctant to adopt it for development because I did not believe that the product was going to be around for 10 years or better. I think that longevity is one of the key features of the Linux environment.
Microsoft has already committed support of SL5 through the end of 2021. While apps do have a tendancy to stay around longer than that (like the 16-year-old VB6 one we’re getting ready to re-write in SL), I personally think that 10-year window is sufficient to get great use out of a SL LOB app.
IMHO – If Netflix wasn’t using Silverlight for their instant-play delivery the numbers would be much lower.
The problem with SL, is that I’ve had trouble installing it on 64bit MacOSX…
My (non-software) company is getting ready to tackle a re-write of our core sales order entry (with-a-little-of-everything-else-mixed-in) application. The current version is in VB6 and is way overdue for replacement. I’m pushing hard for SL in spite of the negative publicity of the “SL is dead” whining. XAML will be alive in well, like Colin stated, and a re-write in SL will only help prepare our dev team for future Windows development. Of course, it helps that we’re a Windows-centric business.
I’ve also heard someone make the case that if SL5 is the last version, that might actually be a good thing from a practical perspective: having a mature, stable product allows developers to be highly productive with it without chasing all the new features every 9 – 12 months. Think about WinForms over the past 5 – 7 years…not much has changed, but that has made it a good and stable platform with which to build business apps.
Of course, it goes without saying that while my team enjoys the benefit of a rich, stable platform like SL, we also need to be engaged in learning and adopting newer technology. Colin – thanks for sharing some good, reasoned thoughts on these matters.
Hi Tim,
Have you seen the White Paper I wrote a few months ago, “Flex, Silverlight or HTML5, Time to Decide …”, the conclusions of that paper are still quite relevant today.
If you can guarantee that you will not have a future requirement for your sales order entry application to run on tablets or smartphones, you will find that SL is a *much* more productive framework for creating this application compared to some JavaScript equivalent.
However, if there is a chance that in the future you have to run this on a tablet, your stakeholders will not thank you for choosing Silverlight!
Now, more than ever, choosing the right technology has become very difficult to do. Even setting aside bias, whether you are pro-Microsoft, anti-Microsoft, or like myself, a pragmatist, making teh right choice is *hard*!
Good luck
Colin – I had read your white paper and it was very helpful…in fact it was one of the things that helped me consider continuing the SL path for this project. There IS a need to allow *some* functionality on tablets…not right away, but in the next couple years. However, right or wrong, my take is that the use case for a tablet user will be different from the use case for an office/desktop user. Meaning, even if SL was truly platform independent and could run on every device, we’d likely be designing a different user experience for tablet users vs that of the desktop users. If the tablet experience will be a limited subset of the desktop experience, is it really that bad to expect to create an HTML5 client when the time comes that exposes pieces of the SL functionality? Yes, there is some re-work but I’m not sure that is a huge deal (when considerating the points below). What are your thoughts on this?
Part of the equation for me is the following:
- we are a development team that has very little HTML/Js experience (we won’t stay this way…we are learning, but we’re simply not at the place to be able to produce an enterprise LOB app with HTML5/Js);
- everything I’ve heard regarding HTML5/Js development is that it is very difficult to develop an enterprise LOB app;
- the tooling and available productivity for SL vs (my perceived) lack of tooling/productivity that exists in the HTML world.
- the availablity of tools like Xamarin allows me to use my existing C# business logic components in the iPhone/Droid environments for making an initial entry into those areas.
I’m not trying to be like those who clung to VB6 when .NET was ushered in…HTML5 is something I want to get experience with. But I’m trying to think practically about it with the current time frame (a re-write of our primary aplication delivered before the end of a 2013) given the points above. I’d love to hear suggestions.
Hi Tim,
I think it is an important observation you have made that tablet applications typically do *not* have the same functionality as their desktop counterparts. It is unlikely tablets will replace the PC as the general business workhorse that many of us sit in front of throughout the day. In that case, I agree, a HTML5 or native client that targets a tablet sounds like a cost-effective plan. On that note, just because you create a HTML5 web-app, it does not mean you get a tablet app for free, the touch-based tablet interaction is something you have to consider and design-for, as is the smaller screen size (and higher resolution).
I entirely agree with your HTML5 / JS observations that is not the most productive approach for LOB applications. However, have you tried Google Web Toolkit? I guess your team are .NET developers – however, if you do have Java skills, it is definitely worth looking at. Google used it for Wave and Analytics.
Xamarin, MonoTouch , MonoDroid are technologies that I really want to spend some time with in the future. It does look like there is scope for significant code re-use from your Silverlight web app to tablet / phone apps. You also get the benefit of a fully native UI. It is definitely on my radar!
Regards, Colin E.
Thanks, Colin. Yes, we are primarily .NET with little Java experience. Good point about the HTML5 web apps not containing tablet-specific functionality out of the box. I have heard but not tried Google’s web toolkit. I will have to check it out. But at this point I would be afraid to do a major LOB app while learning the ropes. (It is challenging enough as it is with SL/.NET.)
I tend to agree. At the end of last year, we did an appraisal of what technologies to use to provide a web based LOB. The only solution was Silverlight for a number of reasons.
Even though it is not the Microsoft flagship it was, it is a very mature product that can leverage a vast scope of the .Net framework. The look and feel is simple to implement. The speed and ease to get up and running is phenomenal.
It is a pity that Microsoft have changed direction as they have already built a winning solution
Maurice
Hi Maurice, totally agree, Silverlight has a sweet spot that it works really well for. The problem is that the perceived change in focus to HTML5 makes it hard to recommend Silverlight – even when it is the best technology for a specific problem.
[...] The Gradual Decline in Silverlight Adoption (Colin Eberhardt) [...]
Brilliant read
Stats aren’t far off but u have to allow for concentration and seed funding efforts to jack the data up artificially. In that we spent a fortune in china to bump the numbers so they have the appearance of ubiquity and download metrics
You’d be suprises what a few 100million installs cost per month
Thanks Scott,
Surely a cheaper alternative would be to find the sites that StatOwl and RIAStats monitor and stage a DoS style attack from a small group of computers with Silverlight installed
Colin E.
It is really sad to see technologies like Silverlight & WPF losing steam and ultimately dying (For me, no development, stale project = dead). The same goes for many MS technologies. I do not develop cross platform applications at all and all my work is done using C# in .Net/Win. I love WPF as I did love Linq2SQL but sometimes I tend to think that investing in any MS tech is a complete waste of time and effort because most of the decisions taken by MS in these areas are hardly driven by programmers. I hope someone at MS realizes this and keep these awesome products alive. I won’t even mind them getting clubbed in one product.
Great post… I think developer enthusiasm is dwindling as uncertainty about the technology grows.
Also, it just feels like Microsoft has stopped beating the Silverlight drum and moved on to HTML 5 and Javascript.
Personally I think SL is a great technology and so it is sad to see these kind of adoption numbers.
Thanks Mark, I agree SL is a great technology, which for a number of reasons will probably never realise its full potential. Have you seen my untimely demise of plugins blog post? It sums up my feelings regarding HTML5 vs. Silverlight.
Colin E.