Monday, August 9, 2010

Follow up on the “geek post”

I never realized I had quite so many super-smart and geeky readers.  The quality of comments – especially those that came through my inbox – was amazing. 

This post attempts a synthesis.  That is difficult because many of the super-smart and geeky readers had diametrically opposed views.  This was typical:

The bottom line is none of this stuff matters enough to get religious about it. If you want to go through a four week learning curve to use Ubuntu, have it. If you want to use second-rate office software because it's good enough, by all means do that. But if you're preaching for a mass conversion to happen, don't hold your breath waiting. In all likelihood your son will soon want a Windows machine so he can play the same games his friends are playing.

And I have to say he is right.  Ubuntu is faster – and it is nicer for some things – but I have such huge capital embedded in Windows (years of use) that I will use Windows for many things.  Moreover getting Windows games to play on Ubuntu is (frankly) not worth the chop (even inside “virtual windows” – explanation below).  Also my son does have a windows partition – and the only thing he runs on it is Cyberlink Powerdirector 8.  Games get run on the Wii (or Playstation) – but a non-linear video editor is essential for him in these days of YouTube.  My son communicates with movies…

My purpose was not to learn to use Ubuntu.   My purpose was to work out where I stood as a stock analyst.  It makes no sense whatsoever to change unless you have very specific needs [the usual need being to maintain servers]. 

By contrast I was told a story about an elderly (and non-techie) woman in Lithuania who was given a brand new laptop with Windows 7 on it.  The first thing she did was wipe the disc and install Ubuntu.  Why?  Because she was used to it and knew how things worked.

And that is the point: these things have inertia – and the inertia is real.  Windows is useful primarily because it is there.  In particular:

(a).  The easiest system to use is the one you are used to.  If you use Windows at work it is likely you will use it at home.  I feel uncomfortable on a Mac because I have not used one since I gave away my Mac Plus.  [That I guess dates me…]  The woman in Lithuania felt uncomfortable with Windows because that is not what she is used to using.  The cost of changing (four weeks) is just too high. 

(b).  The price does not really matter.  The OEM cost of Windows is about $50.  The laptop lasts three years.  There is no way that anyone is going through 4 weeks of hard intellectual work of learning a new system (any new system) for saving of about $17 a year.  That is simply a non-starter.  This defends market share.  I hated going from my Mac to a Windows machine at work – but once I crossed that divide all my future machines were Windows.  However the young are back buying Macs and that will weaken the Windows incumbency.  There are a few places where Ubuntu has serious share (the Baltic states spring to mind) but that is rare.  The proof that price does not really matter is the sudden willingness of people to buy Macs at price points up to $500 higher than current PC offerings.  They do this because – well – the Mac is nice…

(c)  Lots of software is written already for windows – especially games – but also video drivers etc – that do not run as native on linux.  Moreover because the desktop share of linux is so small nobody wants to bother porting their software.  You use Windows because everyone else does.

(d).  Microsoft had a really useful suite of “developer tools”.  This meant that many businesses wrote software for their businesses that works in Windows and does not work elsewhere.  This also means that they have a reluctance to change.  [By contrast most business software these days is written with web-interfaces.  We have written a database for work – and our developer wrote it to work on our linux server – but the interface is our browser…]

Microsoft has not come up with a really good “must have” product in a while.  That said – Windows 7 is quite a good system – and even my tech friends who are linux devotees – confess an admiration for the new-found stability.  This will drive some sales as sensible people did not move from XP to Vista – but will move from either XP or Vista to Windows 7. 

Inertia matters and this bodes well for Microsoft.  The tailwind (developing countries) remains strong for some time.  Inertia saves the Western business momentum.  The cash-flow is robust for at least a few years – and will probably grow.  I even purchased some Microsoft stock.

Some thought I was overstating virtualization as a future because it is reliant on very fast connections.  There are a few responses to that.  Firstly within the enterprise (which is the core Microsoft market) virtualization is a reality now.  I know a 50 person financial firm whose computers are really two mondo-powerful servers.  The 50 staff all have a “virtual box” with the computing capacity shared on the servers.  One server would do the job – but two is for redundancy.  There are an exact copy two servers in a remote location about 50 miles away – that is the disaster recovery.  This is superior in lots of ways.  There are no distributed hard drives which makes data much safer from theft (you cannot walk out with a hard drive).  There are no USB keys or other ways to pilfer data in the field.  The computers are mobile – someone can log on from anywhere in the world.  If they move desk they do not need to move their computer.  The disaster recovery is really simple – and is an exact duplication of the work machines.  No decentralized data to lose.  No problem with mirroring. 

This is a sophisticated and superior set up.  But it is cheaper than the existing set up – and easier to maintain.  That is a winner.  What works there will work elsewhere.  I mentioned this to a UBS executive and he just said they were miles behind on that.  But hey – even UBS will catch up!  It will happen in the home too – maybe earlier in countries like Australia (which is rolling out the National Broadband Network) – but it will happen more generally.   I have gone personally from opposing the NBN (a costly government project) to being supportive (it will allow us to save huge amounts in IT infrastructure).  That said – if it works in Australia it will happen everywhere with sufficient population density.  Anywhere that you can deliver high definition TV on demand has more-than-sufficient bandwidth. 

Linux as the dominant operating system

So lets get to the big point.  I am going to put it in bold because it – at first look is such an outrageous statement – but I think it is inevitable.  Within 18 months the world’s biggest selling operating system (that is the one in the largest number of new devices) will be linux.  It is not even going to be close.  Within three years linux will be utterly dominant – maybe a 70 percent share. 

The thing is all that share will not be on laptops or desktops.  It probably won’t even be on tablets or pads.  It will be on telephone handsets.  The handset market is about a billion pieces per year.  The computer market maybe a fifth that.  The dominant phone operating system will be Android – and Android is just a cut-down linux.  Who cares about the front-end of Ubuntu (which is adequate).  What we should care about is the front-end of Android – which is a winner.

That has implications – and if you were a PC maker or a Microsoft investor you better think them through.  But I will outline a couple. 

(a).  You can run a windows machine or even a Mac on a linux machine using virtualization.  This makes possible Windows as an app – or for that matter Mac as an app.  This will require more computing power than a handset currently has (but give it time).  It will also require some form of “docking stand” into which you plug your keyboard, mouse, printer, very fast internet connection, sound system and maybe a hard drive for mass storage.

(b).  Virtual machines can be spread across multiple platforms and can share computing capacity with the platform.  So you can have your “windows as an app” which uses the handset for simple tasks (such as word processing etc) and uses the cloud’s processing power for complex tasks (such as rendering an MPEG4 movie for YouTube). 

This machine can be far better than the computers of today.  Here are a few drivers:

1.  Software only from repositories.  One of the reasons why Linux has far fewer glitches than Windows is the software loading process.  With windows I go to a site and download a .exe file and click (although there are even drive-by infections on the web).  Nobody polices the sites I get software from.  With Linux most software comes from “repositories” where the software is downloaded from vast servers.  (Downloading software from outside repositories is surprisingly difficult…)  When the software comes from “trusted sources” it is far less likely to contain a virus or trojan.  The Apps store is just a repository – and refusing to allow software purchases outside the store is a protection mechanism.  When we do cloud-software the service provider will probably also provide repositories.  Cloud computing increases security – and the absence of trojans will increase privacy.

2. Machines are mirrored in the cloud.  If you lose your phone you buy another one.  You then download your old computer from the cloud.  No data is lost.  The new “owner” of your phone will not have access to your machine without some serious identification – probably your thumbprint and a password.

So here is the question: can Microsoft – or even Apple – win the “pc as an app” war?  Is incumbency enough protection?  In the corporate area it clearly is enough for now.  The financial firm I described above is fully virtualized.  The servers run linux – but the (thin) clients are all Windows.

I learnt a bundle from the comments – so more requested.







PS.  I mentioned that the virtualized PC does not play games well.  There is a reason – I run a laptop.  To play the game well the virtual computer would need to take direct control of my graphics card.  If it does that the host computer (ie the linux box) loses control of it – and hence loses its screen images.  So I can’t play even simple 3d games in my virtual computer.  Every business function is fine – but the “fun-stuff” is not.  The graphics card companies (ATI, Nvidia) can probably solve this by having dual-chip graphics cards or the like.  But no matter where I looked I discovered that graphics cards were the biggest problem with my set up. 

It is probably good for me to have a work computer that cannot play games.  (Saves much time wasting.)  But it is not ready for general prime time.  Again solutions cannot be far away – but they are not really viable for a non-geek now.  [My computer will accept a second graphics card through a PCI slot.  This is precisely what is envisaged by the manufacturer when they put that option in the BIOS.]


狂猪 said...

Microsoft has three simple strategies against Android phones

1. Make each new generation of MS Windows more bloated (aka more feature rich)! This will make running the latest Windows inside virtualization on low power machines (android phones) a poor experience. Historically, every new generation of MS Windows have made the current generation of PC hardware obsolete. This is a very profitable trend a whole industry depends on. Will this trend not continue?

2. Microsoft can continue to evolve the Windows Mobile platform to match Android and IPhone in design and function. Microsoft will be able to offer a better Windows as an app experience on the Windows Mobile platform than anyone else. The simple reason is the engineers of both technology platforms works at Microsoft. If a user really wants to run Windows software on his phone, which phone will he buy?

3. Microsoft may also have a strategy that requires little more than a drop of ink. Just write it in the End User License Agreement that MS Windows cannot be run within Android.

Ultimately, if Android is to overtake MS Windows, it must succeed on it's own. It cannot succeed just by running Windows apps. Android as a phone platform is well on its' way to beating Apple in a few years. It is a replay of a 80s movie. Let's see if Android can repeat MS DOS' success. First baby steps than world domination!

Robert said...

Virtualization does not always need to be dependent on a connection to a remote host. For an example of what I mean, I suggest looking for a video demonstration of XenClient running on a laptop. XenClient is a bare-metal hypervisor, which is a step up (or, in a more literal sense, a step down) from things like VirtualBox. Hypervisors allow for things like live migration of virtual machines from one hardware platform to another, all without disturbing the operation of the VM. More importantly to consumers, bare-metal hypervisors allow guest operating systems to make direct use of the graphics hardware for applications such as games or rendering images.

I would also like to mention that downloading software outside of the repositories on linux is not particularly difficult. It's installing it and keeping track of it that's a headache (this often involves compiling the software yourself.) But you did hit upon the key advantages of repositoreis: they're secure and convenient. Software in the repositories has been pre-compiled and (one would hope) tested to work properly. In addition, the packages are cryptographically signed to prevent any tampering.

As per your last question: I would say that incumbency is only enough protection if your platform is the only one that can run essential applications. If all platforms could run all applications equally well, operating system usage would change to reflect the merits of the OS itself, rather than the applications that run on top of it. said...

iirc, intel have been working on virtual direct hardware access. The biggest issue is that of memory addresses, as they need to be handled so that the driver thinks its writing to one address set, while its actually writing to another, and the hardware swapping them around as requested by the VM software.

however, so far this means that you need to be running a specific combo of recent intel hardware and compatible VM software. And this means intel can sell their hardware at a premium vs competitors that lack support of this feature.

F. Heinsen said...

The estimated “$17 a year” in saving is a misleading figure, because it considers only the cost of the OS.

We should also consider the saving from switching other components across the software stack – not just the OS, but also the applications.

Switching to free, open-source software is not an all-or-nothing proposition; it can be done partially or even gradually, e.g., by remaining on Windows but replacing the browser, then email, then the office suite, etc.

Many customers routinely pay Microsoft hundreds of dollars a year per desktop in software licensing fees. The license-fee savings per desktop are considerably greater than $17 a year.

Otherwise I largely agree with your views.

William Huynh said...

I have resisted commenting on a number of topics here for some time as I wasn't sure I had much value to add. This time, I think John has gotten the hammer right to the head of the nail.

To give some context, I'm a relatively tech-savvy mid-20-something investment banker in Canada that has an interest in technology and is a huge fan of Gizmodo, 2daytechnology and other techno-blogs. And I was having a similar discussing with some friends last night.

1) Gaming - Much to the chagrin to many people, the number of computer gaming successes (barring WoW, Starcraft 2, etc) are becoming more and more limited. By all accounts it's a dying or very ill medium. That said, rising momentum in Mac sales means that inevitably all "hot" titles will be compatible with both PCs (windows) and Macs.

2) Android - The crux of the matter is, despite peoples' feelings regarding MSFT, the business model that has churned out massive amounts of cash over many decades is the one being adopted by Google right now for phones.

3) Switching Costs - The caveat to this is that with the emphasis towards intuitive user interaction with these devices (computers/phones) the learning curve (cost) for switching has dramatically decreased. Now everything isn't completely automatic (or plug&play) with Macs or PCs yet - but with phones anyone can pick up and use an iPhone, Blackberry or Android phone. Easy, intuitive, no installations, drivers or "management" issues. They typically simply work, albeit with a strange throw-back, because - this may just be me - but an Apple dashboard covered in icons reminds me strangely of a cluttered Windows 95 desktop with a billion shortcuts on it.. The more things change the more they stay the same?

Anonymous said...

These are the sort of issues that plague tech stocks and why they should trade at low valuations. Many leaders are vulnerable to a takedown by some guy in a garage. That doesn't exist in most industries.

In this case however I don't think it's relevant to shareholders. MSFT is a mighty incumbent and the stock is mighty cheap. Even if you're right, it will be many many years before it matters to investors. Reminds me of your Gates/Kodak post except MSFT is not headed for that fate at all.

Regarding Droid, I agree it's good compared to most attempts to unseat iPhone. But it's still a distant 2nd, particularly is user friendliness. Ask IT departments how many more questions they have to deal with from employees after switching. Given the amount of time Google had to observe, copy and improve, I expected it would be better. Apple is a remarkably innovative and intuitive company.

狂猪 said...


Do you think the performance gap between a PC CPU and a mobile phone CPU will shrink significantly in the medium term (5-10 years)? The reason I ask is software are written to match the capability of the hardware on which it runs. The more capable the hardware, the more complex the software that runs on it. This matching of hardware to software complexity is what drive the very profitable hardware and software upgrade cycle in the tech sector.

If the performance gap doesn't shrink significantly, then there will always be a significant mismatch between the complex PC software and the limited capability of mobile phone hardware. The reason for the performance gap is due to the difference in how the hand-held and the PC is used. While it is clear that both type of CPU will improve over time. It is unclear why the performance gap should shrink significantly. After all, these CPUs are designed to work under very different operating conditions and requirements. These conditions and requirements don't vary much with time.

MohKohn said...


Good analysis in your two posts. I would like to address a few points :)

Firstly, I would like to comment on the GNU/Linux user interface. As noted, there are actually several - Gnome, KDE, and Android (which is Linux but not GNU) are the three biggest. There is also the new MeeGo from Intel/Nokia, which is based on KDE technologies.

I am a user interface designer (among other things). My assessment is that Ubuntu Netbook Remix and MeeGo are considerably ahead of anything Microsoft makes in terms of UI on their target devices. I haven't used Android, but I hear it is quite nice too.

In desktop UI, Microsoft has kept pace with the release of Windows 7. I expect the release of Gnome 3, if it fulfils its promise, to pull Ubuntu ahead of Windows in desktop usability.

Applications are a different matter. There's not much of a gap between Word and OpenOffice. Photoshop is considerably more usable than the GIMP. Inkscape is more usable than Illustrator. I admit I haven't tested any of this, these are just opinions and not to be taken as gospel.

But it is notable that in every area free software has advanced rapidly in user interface design over the last few years. I expect Gnome 3 to demonstrate that this trend can be maintained past the point where free software catches up with proprietary alternatives.

GNU/Linux distributions are taking more and more responsibility for providing an integrated experience, which will be a huge boon.

(...pause for breath...)

Secondly, application availability. You note that there is no nonlinear video editor of sufficient quality on GNU/Linux. I'll take your word for that. From my own experience, there is no music-creation program that can compete with Ableton, Logic or Reason.

However I've been geeking around this area for a decade or so now, and I can remember very clearly the time when GNU/Linux was not an option on the desktop because we did not have a word processor as good as Word. In my experience so far, free software inexorably fills these gaps.

The greatest advantage GNU/Linux has is that it has no stock price, no shareholders, and no debt. It cannot go bankrupt. Microsoft has been very effective at destroying competitors then consolidating a hold on the market. It cannot destroy GNU/Linux. The only way to compete is to innovate faster. Apple have managed this, so far.

MohKohn said...

Thirdly - and, I believe, most importantly of all - there is an issue of political economy.

If you will indulge me in flirting with Marxism for a second, this has three components - marginal cost, ownership of the means of production, and the mode of production.

The marginal cost of reproducing information (including software) is so close to zero now that it would be difficult to calculate. This is the single biggest challenge to all information-producing companies, from the movie and music industries to Microsoft themselves. Apple are a little safer because they also produce hardware.

With this fact in mind, 'piracy' (sharing) is no longer morally abhorrent, but simply rational economic behaviour. The rate of profit for selling information is falling, and will continue to fall until it barely exists.

With the commodification of computer hardware and software, the means of production of software are in the hands of developers themselves.

This is one of the places where Marx was spot on - as a result of this change in ownership of the means of production, a new mode of production has emerged. The online collaborative, "user generated content" mode, which was first pioneered in the free software world.

This is unprecedented in human history - thousands of highly skilled people, many volunteers, many employees of competing companies, all working together to create a product which they then give away.

The implications of THAT will turn out to be far far larger than the success or otherwise of one software company.

James Moore said...

In my experience, young gamers are all about consoles. None of the early-teenage gamers I know care about gaming on a PC.

F. Heinsen said...

If I may, let me add that even though I largely agree with you that a mass migration of the desktop & laptop markets to Linux is unlikely to happen quickly, there are nonetheless possible nightmare scenarios for Microsoft shareholders under which adoption of Linux as a desktop OS could hit a tipping point and suddenly accelerate beyond anyone's expectation.

For instance, what would happen if a truly prominent U.S. business (say, a corporation like Wal-Mart) or a large government body (say, cash-strapped California) announced that they will follow Google's lead and switch all desktops to Linux over the next few years, claiming they're doing so to reduce costs, improve security, and lessen vendor lock-in? What would be the impact of such an announcement? How about a few such announcements?

In fact, precisely this kind of large-scale, wholesale desktop software replacement is already happening in some corners – here are just a few examples:

(The two videos above should be required watching for Microsoft shareholders.)

Are these odd, isolated, one-time incidents, or the proverbial canaries in a coal mine?

Trieu said...

It's a bit odd to conjoin Android and Linux. OSX is based on Darwin, which is about as similar to Linux as Android is. If you're going to count Android phones as Linux machines, then you may as well count iPhones. And Android is in many ways as closed of an operating system as is OSX. Don't be fooled by the PR; Google owns Android and controls it fully.

This is really to say that the relevant lines aren't so much Linux-Windows-OSX, but rather Google-Microsoft-Apple.

Trieu said...

Quick follow-up note: I think of iOS (the iPhone/iPad/iPod Touch OS) as basically a stripped down version of OSX (because it is).

Mitch said...

To restate part what MohKohn said, in a way that should be more appealing to people who think of themselves as capitalists (the Econ 101 version):

Marginal Revenue = Marginal Cost, right? If marginal cost is zero, therefore marginal revenue should also be zero.

NickL said...

To address a previous comment, gaming is far from dying. Total time spent gaming has continued to increase pretty rapidly. It's just moving online. See the success of WoW, Starcraft, Tencent, etc. In fact, the most recent wave of growth is happening purely online and mobile (no client download) through Zynga, DeNA, etc.
John has correctly identified gaming as a barrier to virtualization and linux but it is one that is declining as more game play takes place online and on mobile.

Anonymous said...

lots of major FIs are using Citrix MetaFrame--greater security and ease of administration.

Anonymous said...

xI'm a little more tech-savvy than your average person but don't really have the time to keep up. With that in mind, this talk of virtualization seems a bit ironic since it reminds me of the first computer set up I worked on way back when - using a terminal to dial-up a network somewhere yonder. We did this because we couldn't afford the processing power. But when PCs came around, everyone knew that having one's own machine was infinitely better than depending on someone else to keep the server upgraded and the apps current and information secure.

I guess we really have come full circle.

Anonymous said...

Frankly, I think gaming and high end video/audio editing are the only reasons a sensible person has to upgrade a desktop computer nowadays (other than if the computer just falls apart due to old age).

I honestly think a Pentium 4 has the power to run just about every program a non-gamer or video editor would need to run, and do it well.

This value chain is slowly dying.

If virtualization becomes the norm for "home" desktop computers as well as enterprise computers down the road, companies like HPQ and MSFT or even AAPL will be in big trouble.

On the other hand, maybe mainframes will come back into fashion :)

Anonymous said...

Hi John,

this reminds me of evaluation of buying BT group around 18 months ago.

Large mature business, the vast bulk of revenues are from a cash cow based on obsolete and declining technology (landline phones), minimal revenue growth to slight decline overall as new businesses (data) are not growing fast enough to offset declines in core business.

At the time BT was $12 and price to FCF ratio was around 1.3x. Of course the shares are now $23 and i'm kicking myself.

My guess is Microsoft will be a very profitable company for years to come with minimal topline growth, so you want to buy them at a compelling cash flow multiple.

However Microsoft is not cheap enough yet as shares still trade at 12.5x fcf

It all comes down to your assessment on how quickly the windows and office business as well as their business software (exchange & SQL server)will decline over time as all of the rest of their businesses are money losers.

Another key factor is they are currently a distant 5th in smartphones/mobile computing with little prospect of catching up to the leaders.

I wouldn't buy MSFT unless the shares drop below $15 which is unlikely in the short term unless they have a quarterly earnings miss and the stock gets heavily sold.

Jeff Matthews said...

John, it isn't just the young buying Macs again. A grandmother and friend of ours just got one for her 70th birthday...and loved the service (especially the classes at the Apple store).

Cheers and thanks for this thoughtful, well researched and well written series on an important topic.

Jeff Matthews said...

And getting back to the very first comment here, regarding Microsoft's "three simple strategies"...

This is the heart of the problem with MSFT: everything they do is driven by the need to extend the Windows OS. Hence the failure of MSFT's efforts (to date) in any new non-desktop-related product EXCEPT for X-box.

There is a reason Vespas are not powered by Chevy engines, and so long as MSFT insists on using Windows to spearhead its new product efforts, it is doomed to fail.

Those "three simple strategies" are driving MSFT out of the minds of consumers and into the arms of Apple, Linux et al.


Huss said...

Most important thing to me is the 4 week learning curve period...the vast majority of western demographic doesn't have the patience for a 4 HOUR learning curve, let alone a 4 week learning curve...this is not to say Microsoft will forever dominate and hold share in the PC space but for those discussing the detailed benefits, cost savings, and all around great aspects of Linux, that's all well and good but the large blocks of humanity don't give a damn, my dear.

Till said...

For casual users, which most Microsoft Windows users are, Ubuntu can be learned in a few minutes. No four week learning period is required. Not only that, but Microsoft Office Word, Excel and PowerPoint documents can be read, edited, created and saved in Open Office, which installs with Ubuntu.In the dual boot installation I created, my Microsoft documents are all available to Open Office in Ubuntu; so nothing needs to be transferred or migrated.

Learning some of Ubuntu's specialized features such as its text editors and UNIX terminal Window does take some time, but most people will never have a need for this.

Ubuntu is so much faster and easier to use, I won't use Windows anymore. The only thing holding Ubuntu back at this point is the availability of off-the-shelf computers that already have Ubuntu installed. Once they become available, Ubuntu will gain market share very rapidly.

chrispycrunch said...

To the poster who thinks the pc game market is shrinking, the stats say otherwise:

General disclaimer

The content contained in this blog represents the opinions of Mr. Hempton. You should assume Mr. Hempton and his affiliates have positions in the securities discussed in this blog, and such beneficial ownership can create a conflict of interest regarding the objectivity of this blog. Statements in the blog are not guarantees of future performance and are subject to certain risks, uncertainties and other factors. Certain information in this blog concerning economic trends and performance is based on or derived from information provided by third-party sources. Mr. Hempton does not guarantee the accuracy of such information and has not independently verified the accuracy or completeness of such information or the assumptions on which such information is based. Such information may change after it is posted and Mr. Hempton is not obligated to, and may not, update it. The commentary in this blog in no way constitutes a solicitation of business, an offer of a security or a solicitation to purchase a security, or investment advice. In fact, it should not be relied upon in making investment decisions, ever. It is intended solely for the entertainment of the reader, and the author. In particular this blog is not directed for investment purposes at US Persons.