Changing my mind on Microsoft

Seldom have I looked at something that is a major long in the portfolio, changed my mind, sold the entire position and continued selling to go short (albeit in a small way).

I just did that on Microsoft. The immediate trigger was Windows 8 - but the thinking has been longer and harder than that.

This post is to run through my thinking - and maybe generate some comment. (Smart readers - and you are smart readers - are a great testing-board for my theories...)

Quick background to the Microsoft story

The background to Microsoft is well known. In the late 1990s Windows developed huge market power. Whilst not strictly a monopoly the company had plenty of monopoly characteristics. Sure you could buy a Macintosh - but that market was so small that people did not develop software for Macs and hence Macs were for people who did not need a wide range of software. You could also load a computer with Linux (although despite being a sometime-geek I could not imagine doing except for a server). In those days Microsoft even dominated the server market.

This was - for all effective purposes - a growing global monopoly with very low marginal costs producing for what was fast becoming one of the most important industries in the world (personal computers).

Moreover it had huge pricing power. I purchased a computer from Gateway (remember them) and spent about $1200. The $50 operating system was embedded - a small cost embedded in a large cost. I assure you Microsoft made more from the transaction than Gateway. Gateway even ran a store to sell me that beige box - the store ultimately being run for the benefit of Microsoft.

The company had a virtuous circle. People developed software to run on Microsoft using Microsoft developer tools. They did not bother developing for other platforms because those platforms were economically irrelevant and the Microsoft developer tools worked. Because all the software you might want to use ran on Windows you were effectively compelled to run a Windows machine thus perpetuating the cycle.

Microsoft used its power for better and for worse. My personal gripe was how poorly Microsoft thought about and handled security issues. My Linux computer is as far as I know virus free. Apple only had to remove their virus-immune claims recently. Windows security issues are everywhere and it did not need to be so. The first computer virus I ever saw was in 1989 and it was on a Mac - these problems - if not entirely soluble - were controllable. Microsoft did not control them. Whether this was arrogance or incompetence or monopolistic-disinterest I do not know though I have heard arguments for all three propositions.

Still Microsoft gave us acceptable if not brilliant product and I never quite bought the "evil-empire" line. In my view they were a large, increasingly fat, slightly disinterested monopolist that had stopped thinking clearly about users.

Developers: the key to the Microsoft virtuous circle

The key to Microsoft's market power was the virtuous circle whereby people used Microsoft computers because software was developed for Microsoft and people developed software for Microsoft because people used Microsoft computers.

Making all this work required that Microsoft make available easy to use "developer tools". A large developer tools business would include tools for software development and training tools to train future developers. If the developer tool business ran at (say) a billion dollars in loss that was perfectly acceptable so long as more and more software was developed to be Microsoft platform specific.* Indeed the developer tools business never played much of a role in the sector-breakup of Microsoft - but that did not diminish its importance.

If you are not convinced that developers are the key to Microsoft's lock-in look at this classic video of Steve Balmer:




Balmer is sufficiently worked up that the wags captioned this video with the text deodorant, deodorant, deodorant. However the extent to which he is worked up tells the Microsoft crowd what they should be focussed on.

The rise of platform-agnostic developer tools

The Microsoft virtuous circle is now dead. Two related things killed it: the rise of platform agnostic developer tools and the rise of alternative operating systems (Linux for servers, iOS and the "Big Cat" series for Apple, Android).

To my way of thinking the platform-agnostic developer tools came first - though this is a chicken-and-egg problem. The first really important platform-agnostic tool was Java. Programs written in Java run on Linux computers precisely the same way as they run on Apple computers or Microsoft computers. If you developed something on Java you could run it anywhere and you thus undermined the Microsoft virtuous circle.

Developing things for Java became widespread when people downloaded programs (applets really) from the internet. The person writing the applet had no idea what the customer computer set-up was and so had to write in a platform-agnostic fashion. Interactive Brokers for instance writes its software to run on Java - and they do this because it is a complex piece of software that has to run on many different flavours of client computer.

Over time Python developed as an even more important platform agnostic developer tool.

Nowadays nobody under thirty writes anything on Microsoft developer tools unless they are demented or brain-dead. Firstly the kids out of the colleges know the platform agnostic stuff well. Secondly when half the computers leaving factories either run iOS or Android (that is are smart-phones) nobody sensible will write in a way that does not allow easy porting to these platforms.

Microsoft's developer tools business and the customer lock it created has had a bullet through the brain. The body is lying on the floor - and most the users who have never developed anything and did not know that there even was a developers tool business have not noticed the blood-soaked victim.

The lock that Steve Balmer worked himself into a frenzied sweat over is dead.

The lock is dead: long live the lock

An asset management firm I know well has 100 thousand or so customers. The customer relations system for the firm is proprietary. They developed it themselves and it integrates with their business practice.

And it runs on Microsoft. It was developed by people who are now over 35 - and hence used Microsoft developer tools.

This firm is very progressive with their computing structure. All internal computer now run as virtual machines (not desktops) running on two mondo-powerful Linux servers. The virtualization platform is Citrix. Nobody has a functional box under their desk any more.

However on top of this enterprise cloud is 65 virtual Microsoft machines all running Windows. The company has got rid of the desktop computers entirely (sorry Dell and HP), it has a hugely powerful internal network system (currently provided by Cisco but in the future provided probably provided by Nicira). Disaster recovery is a mirror of the two mondo-powerful servers 100km away.

In other words this is the enterprise computing platform of the future.

But they still use Microsoft as if they had the computing platform of 1999.

Why?

Because they used the developer tools of 1999 to build mission-critical enterprise software.

Nobody is locking new stuff up in Microsoft but there is huge amounts of intellectual capital built up in Microsoft's old platform and that intellectual capital continues to force people to use Microsoft. Some of this property is trivial (I know to shut down the computer I go to the "start" button). But things like the front end for a customer relations system for a largish financial firm - that is non-trivial and it is very sticky.

Why I owned Microsoft

When I purchased Microsoft I was well aware of the death of the lock (developer tools). I knew the future for Microsoft would not (unless they were very lucky and well managed) be anything like as glorious as the past - but there were two really decent trends in favour of Microsoft.

Firstly as enterprises moved their computing platforms to enterprise clouds the Microsoft computers - rented as virtual computers - would be pervasive. Microsoft was going to be able to charge rents for a long time. People would upgrade their computer (meaning the physical hardware running linux and Citrix or VMWare on top of that) but they would still pay rent to Microsoft.

Moreover these Microsoft dead-enders - locked in by the enterprise software that they wrote a long time ago - are very sticky. It is expensive to redevelop proprietary systems - and so they were likely to use Microsoft for decades. The pricing would be to lease seats to virtual computers and those lease fees could be high and increasing.

Second, there was one beautiful tailwind for Microsoft which has alas disappeared. In 1999 if you purchased a computer it was probably a beige box. (Laptops were prohibitively expensive and underpowered.) If you purchased a computer in India it was a beige box it came loaded with a hot (ie pirated) version of Microsoft.

By 2007 if you purchased a computer it was likely a laptop. Boxes have become objects for gaming enthusiasts, developers and dinosaurs (I say this as the proud owner of a couple of boxes). The computing power you need can (mostly) be put in a smaller package at a reasonable cost. It is almost impossible to buy a laptop which is not pre-loaded with an authentic version of Microsoft. That was true in India too. I do not even use Microsoft but if I buy a Lenovo computer in Australia on their website I am forced to include a copy of Windows. The tailwind de-jour was the rise of computing in developing countries and most importantly the shift to laptops reducing piracy to almost zero. This was profoundly nice to Microsoft - but as a trend it is dead. The new generation of computers is going to be pads - they may have plug in keyboards - but they are pads. Even laptop sales are problematic.

Moreover laptop prices are falling and falling. Five hundred dollars now buys quite a nice laptop. The laptop I use day-to-day is not worth much more than that (except for add-ons like a large solid state hard drive). Microsoft once buried $50-80 of software in a $2000 computer and that made their (fat and profitable) take disappear. It is much harder to bury $50 of software in a $300 computer - but that is where we are going.

But in essence we had two trends: pricing power on dinosaur enterprise computing driven by the old customer lock (previously developed enterprise software). That pricing power would remain and turn into rental contracts as computers disappeared into enterprise clouds. And we had developed world laptops (a trend that is now turned quite sour).

A Vision of Windows 8

I had a vision of Windows 8 which addressed all of this - and I doubt that it was a vision that was very far from Microsoft's own vision.

Windows 8 was to serve a dual purpose. It was to be above all a pad operating system - one that doubled as a desktop operating system. You were going to be presented with bunch of tiles - the functional equivalent of Apple's app icons. If you used it as a pad it would have the limited functionality of a pad.

However you could take the pad, put it on a docking stand and use it with a keyboard and mouse as a desktop computer. This solves a lot of problems.

(a) it offers a distinct improvement over existing pads which are not very good for content creation. I cannot see myself editing a video on a pad or writing a blog post this long. But hey - I could with a plug-in-keyboard and mouse,

(b) it offers enterprises a chance to take their existing enterprise software and make it mobile. For example if a customer relationship system runs on Windows you could - without much further development - make it run on a Windows pad. This means there would be no incentive to redevelop it using (say) Python to run on iOS.

(c) it gets a large number of people used to the Windows system. There is a lot of human capital developed in using computer systems - trying to change - even Windows to Mac or vice-versa costs a lot of time as you work out how to say copy a file to an external hard drive or from a camera.

(d) it leads you to a world where the pad has some computing power - but if you need more grunt you connect it to a docking station in turn connected to a fast internet connection and you put the power in a cloud and rent the power out by usage. A world of semi-smart terminals - a pad if not docked, a super-computer if docked.

But the combined desktop interface has a big problem. Because desktops and pads and phones do different things they have different interfaces. A windows, icons, mouse and pull down menu interface has a venerable history because it works.

The Ubuntu Unity failure


Microsoft is not the only party that sees a convergence of pads and computers. Ubuntu - by far the leading attempt to make a workable Linux desktop for a very large market - did a complete revamp of their desktop changing from a Windows type interface (Gnome 2) to a Mac/pad type interface (Unity). They had large, immovable buttons - just right to use with fingers. Pull down menus were dramatically reduced in frequency and importance.

This transition was a mess. Utterly horrible. You are not convinced: google the phrase unity sucks.

However - and this is fair - Linux desktop users (we are a small tribe) are probably the most motivated to learn new systems of any group on the planet. Moreover Unity did get better through time.

Apple also knows that combining the interface is very difficult. That is why they have never taken their Mac interface and put it on a pad or vice versa.

You could have worked out the difficulty just by trying to use Unity (but I doubt too many people in the Microsoft development team tried that trick - because it is Linux and not developed there).

I told the whole sorry Unity story to a senior former Microsoft employee and he thought they would not be so stupid to try the combined desktop. He thought that the interface would change dramatically when the computer was docked - looking more like Windows 7 (a good system) when docked and more like a Windows phone (also an adequate system) when mobile.

What Microsoft has done

Microsoft have tried what I originally thought impossible or at least stupid. They did not change the interface of Windows 8 much to deal with the different ways you communicate with that interface. (Fingers versus keyboard and mouse for instance.)

Here is the video which had me selling my Microsoft stake. Its a computer reviewer filming his dad trying to use Windows 8.




I watched this and the pain of problem recognition came over me.  This was exactly how I felt when I first used Ubuntu Unity.

This was a predictable problem. It is a problem that every user of Ubuntu suffered through. This is a management stuff up of the first order.

What Microsoft has done to its business

Firstly Microsoft has not understood its real franchise. Its real franchise is computers on which people do work. They don't play. They write stuff. They enter data. They manipulate graphs. They might even edit a video.

These computers are tools and the operating system is just the air they breathe. On a day to day basis they don't think about the operating system - they only think about it when it changes.

What they should have done is kept something close to a Windows 7 interface when the computer is docked and something close to a pad interface when the computer is mobile. Instead they forced people to relate to a pad interface via a keyboard. They assumed their users were as motivated as (say) Ubuntu users - whereas most their users don't give a fig about learning a new system. Changes to a proven interface either have to be incremental (so you bring your audience along with you) or so self explanatory that the audience learns in 20 minutes (thank you Apple). Windows 8 is neither.

Prediction: this will wind up with a lower corporate take up rate than Vista (ie next to none).

Prediction 2: this will accelerate, rather than slow down, the rate at which enterprises take their enterprise specific software into platform independent programs

Prediction 3: by stuffing this up Microsoft has just about lost its bet on moving the retail computer market into docking cloud computers. Apple will do this. And they will do it by stealth.

Apple, the forthcoming death of the Mac Pro and cloud computing

Bob Cringely laid out Apple's plans a while ago - commenting on all things the lack of an upgrade of the motherboard of a Mac Pro during the latest round of Apple upgrades. The Mac Pro is the most powerful Mac and the only Mac on which users can add their own components through expansion slots.

Being the most powerful Mac it is beloved by power users. The definitive power users are people like video editors. These people want to download huge amounts of data to their (expandable) machine and hence like the fastest download protocols. The fastest current USB protocol - used for say getting material from high definition camcorders - is USB3. You would think that USB3 would be standard in a Mac Pro.

But the motherboard does not have it despite USB3 being a few years old.

And when they did not upgrade the Mac Pro to USB3 Cringely rightly asked what Apple would do about their power users. Here was his conclusion:

Apple will eventually have to explain to those folks [power users] how less is more and how this new world [no expandable computers] is even better for them. I think I know how Apple will do it. 

When the Mac Pro dies for good Apple will replace it in the market with a combination of Thunderbolt-linked Mac Mini computing bricks backed up by rented cloud processing, all driven from an iMac or MacBook workstation. 

I just wonder when they’ll get around to telling us?

Apple's balance sheet is consistent with this vision. Apple has been developing huge cloud computing facilities evidenced by the vast expansion of property, plant and equipment in their balance sheet (which has been piling this stuff on in the billions).

Windows 8 - a product that gets people used to and software developed for a pad that docks - was Microsoft's way of getting used to the idea of portable computers with rented super-computer cloud space.


And it will be a failure because Microsoft, not for the first time, have lost their view of real users.

I held Microsoft for 18 months (and it was not a bad investment). But last night I gave up.

For comment.



John


*There are other lock-ins at Microsoft - for instance my business partner irregularly writes Visual Basic algorithms for spreadsheets. Visual Basic is Microsoft proprietary and these spreadsheets lock us into having at least one Microsoft machine in the office. I complain regularly about this - but Simon is over 40 and teaching old-dogs new tricks is hard. He has a lot of human capital invested in his ability to crank out something in Visual Basic.