In the cellar of Westminster Abbey in London, lies a lovely little café called the Cellarium, with all of the architectural design and feel of the Abbey. Of course, as it is in the heart of London, it has good tea as well. Earlier this week I was privileged to have a fascinating and wide-ranging discussion with Adrian Colyer, which led to 2 provocative questions:
- Will Microsoft open-source Windows?
- If Microsoft had open-source Windows 15 years ago, would it have stopped Linux?
I have written multiple times about Microsoft, especially some of the recent radical changes under CEO Satya Nadella, including “Should they kill the Windows brand in mobile?“, “Free Your App“, and “Microsoft and Open-Source Part I” and “Part II“.
Given the context, let’s examine each of the two questions in turn.
Will Microsoft Open-Source Windows?
Even 5 short years ago, the idea of open-sourcing Windows would have been inconceivable (correctly used, unlike Vezzini). Not only was Windows the core of the Microsoft brand, but it also was the cash basis of the company.
|Year||Revenue||Percentage of Total|
Note: the range of revenue from Windows in the years 2005 and 2015 are due to Microsoft’s lack of reporting of Windows desktop and Windows Server as individual line items, instead burying it in divisions like “Devices & Computers” (including Windows Phone) or “Server Products” (including Exchange and SQL Server).
In the most recent year, the maximum possible percentage of revenue from Windows is 29.6%; likely it is less than half that.
While no one wants to walk away from billions in revenue, it is not inconceivable for a company as large as Microsoft to walk away from 10-15% of revenue, especially in a staged manner, if it suited its strategic interests. If Microsoft saw an acquisition target that was critical to its future, it is wouldn’t hesitate spending billions of dollars to acquire it.
Having determined that they can open-source it, should they?
While the disadvantages of open-source for a commercial software venture are straightforward, some of the advantages are important as well:
- The “DuPont Effect”: People will find ways to use and expand the product far beyond the inventors’ original vision.
- API: An open source product has a clear and trusted API, leading other partners, such as ISVs and services providers, to feel far more comfortable building upon them.
- Add-On Services: While you cannot charge directly for an open-source product, when the cost of adoption is zero, customers are more open to purchasing add-on services, many of which are recurring revenue, which, in turn, lead to more stable cash flow and higher valuations.
- Undermining Competition: When your product is free and open, customers feel more comfortable working with it, undermining potential competition, and allowing you to co-opt some.
In a world where Linux is dominant in servers, mobile and IoT, and investors refuse to back companies built on a Microsoft platform, open-source can go a long way to instilling confidence and expanding the product to the point where it could be useful.
Is it too late? I do now know. It appears close to it.
Would It Have Killed Linux?
Back in the 1990s, if you wanted to run servers, you had 2 choices:
- Rock-solid, stable, but complex Unix variants – Sun Solaris, SGI IRIX, IBM AIX, HP-UX, etc. – running on premium proprietary hardware.
- Windows running on commodity servers.
The philosophical mindset and tools of Windows at the time were at odds with running mission-critical systems, or managing any type of system at scale. The Unix philosophy of well-defined, clear components, with each item, system or tool doing one job extremely well, made it the choice for financial systems, health care, high performance analysis.
Linux, with Unix-style operating methods, semantics and philosophy, while working on commodity (i.e. inexpensive) hardware, at a price-point of zero had instant appeal.
Ignoring the inability of Microsoft, at the time, to survive a revenue loss of that size, what would have happened if Microsoft had responded to Linux not with the infamous Halloween Documents, but instead with, “let’s open-source Windows” and get this community working for us.
Would the community that invested in making Linux production quality and feature-rich instead have focused on making Windows more suitable to high performance and scalable use? And would it have done so in time to head off Linux?
There are some reasons to believe it might have. The core of Microsoft Windows, especially the NT operating system, was solid. It was architected by Dave Cutler, key architect of DEC VMS. While the philosophical fights between Unix and VMS remain legendary, people knew VMS as solid.
The open-source community had at hand a reasonably reliable operating system, with all of the major architectural components, that was engineered to run on commodity hardware. It would seem to have been easier to tweak Windows to have the desired Unix semantics and modularization to support lightweight deployments, than to build up a new operating system from scratch.
On the other hand, many technologists and even customers viewed Microsoft as the “Evil Empire” of the day. Despite the easier path, many might have chosen to attempt to kill Windows with Linux rather than make Windows better.
Water under the bridge.
Now all we can do is wait to see what Nadella decides to do with Windows… and build great products while we wait.