% fortune -ae paul murphy

Re: "7 Reasons Why Linux Won't Succeed On The Desktop"

Last week information Week editor Alexander Wolfe asked readers to respond to his "7 Reasons Why Linux Won't Succeed On The Desktop". Well..if he's trolling for readers, my thought is that getting a third regular reader wouldn't hurt the discussions here either, so...


Dear Mr. Wolfe:

To quote your seven headlines, the Linux desktop won't succeed because of:

  1. Prohibitive application porting costs

  2. The Fanboy alienation factor, or how Linux's biggest supporters drive away potential new users

  3. You can't make money on the operating system

  4. Resistance from average users

  5. Linux is "simple"; Windows "just works"

  6. There are way too many Linux distros

  7. No powerful evangelist for Linux comparable to Bill Gates or Steve Jobs

In some of these the headline misstates the argument and, partly because of this, some are redundant, one is either sad or funny, and you missed the underlying reason for the present state of desktop Linux acceptance entirely.

To see that, lets take it from the top.

You describe reason one as excessive application porting costs, but that's just plain wrong. In reality porting costs are relatively minor, especially if you already have a MacOS X version.

What you're really arguing here is that the market is too small because of reason six: "There are way too many Linux distros" and reason three: "you can't make money off the operating system." It's a common line of argument, but I think it assumes that the balance of revenue to cost doesn't justify the investment, not because costs are too high, but because the off-setting revenues are too low.

Here's your summary:

One can well understand why garage software shops might avoid Linux and attempt instead to mine the more populated Windows user base. But why does one suppose that major vendors like Adobe and AutoDesk avoid Linux? It's because the payoff isn't worth the trouble.

Even if a company's marketing department can be confident that it'll sell a lot of Linux software, qualifying Linux apps is a logistical nightmare and a costly mess. This will remain a stumbling blog for large and small vendors alike.

The reason is, there's no such thing as a single "Linux." If you want to qualify an app on the open-source OS, you've got to test and verify that it runs on a bunch of specific distributions. Commercially speaking, this means a minimum of four distros

Contrary to your conclusions, however, the driver for those costs isn't distribution variation or Linux market fragmentation - it's that Linux kernel change happens frequently and unpredictably. Thus for a company like Adobe to sell a single copy of one of its products on Distribution X, Kernel Y requires that company to support Distribution X, kernel Y for far longer than Torvalds et al will -and that's both why garage software shops don't get into the game and the argument that would be made against Linux at bigger companies if the far more important Wintel relationship issues smothering decisions like this ever let the discussion get that far.

Look, furthermore, at companies which do offer commercial Linux software and you can see an underlying cause for the industry-wide failure to adapt to frequent kernel change. Bear in mind that Linux is really GNU/Linux and recognise that fragmentation over time is driven from kernel change but fragmentation over distributions is driven, except for the gimmicks companies like Red Hat hang on the kernel, mainly from externals that are easy to compensate for in package management and set-up software.

In other words, Linux is Linux regardless of who's name hangs on the ISO and where they choose to put various files. That's why packages like Perl and Apache are easily ported across distributions, but require revision with each new kernel release. So why isn't commercially licensed software equally portable across distributions? or - to put it more sharply, why do companies like Oracle voluntarily incur the costs that go with tying themselves to a few well known commercial releases like Red Hat's? Simple: the lock in code is written to support co-marketing agreements whose purpose is to help the players sell to each other's installed bases - i.e. to lock in each other's customers by fragmenting the market.

In other words your argument puts the consequential cart in front of the motivational horse: you're saying vendors are reluctant to commit monies to a market that's highly fragmented across distributions when in reality it's highly fragmented only over time with those co-temporaneous differences that do exist largely driven from vendor decisions to artificially lock in their customers.

Your reason number 2: The Fanboy alienation factor, is both unfair and simplistic.

On the surface this seems unfair because Wintel fanboys are at least as obnoxious as ours, but the deeper problem is that the opinion is both often cited and completely wrong.

It's wrong at two levels. First, you write about Linux proponents "reinforcing the geek stereotype which has been the main impediment towards open-source adoption among non-technical computer users" but, in reality, the stereo-typical young techno-geek is a Microsoft marketing invention aimed both at bolstering the self images of its fanboys and at preventing the end user from wondering why he needs a horde of techno-geeks to use a product suite marketed as trivially easy to use.

Secondly, and somewhat more subtly, your analysis omits any consideration of the audience involved. To quote from your article:

I call this one the "Linux is too cool to live" factor. My contention that the vehemence of the Fanboys is off-putting to the vast majority of people who might otherwise consider trying Linux requires no argument, only a defence. Q.E.D.


Let me explain: We all know that the statement is pretty much self-evident. We also know that posting such a thing in public will bring the attack dogs out of the Web work. However, I'd venture to bet that they'll be far less inclined to discuss my thesis than to engage in an ad hominem attack on the messenger (me).

Accordingly, let me try to head off the expected invectives at the pass. Some people consider fanboys a pejorative. I don't intend it that way. I say it's a fair, descriptive term. I'd define it as an over enthusiastic and myopic techie who brooks no disagreement with his -- let's admit that it's almost always a "he" we're talking about -- views. (Hey, if the Linux or Apple foobar fits...)

Face it, fanboys. You're not cool because you use Linux, disdain all things Microsoft (MSFT), and treat newcomers as if they have to prove something to enter your elite-in-your-own-mind club. You're just reinforcing the geek stereotype which has been the main impediment towards open-source adoption among non-technical computer users.

Now it's perfectly true that we Linux fanboys like to point out that we are, as a group, rather better informed than our wintel counterparts, but we don't generally address these comments to the general public. Thus the average PC using AP clerk or lawyer doesn't hang out at slashdot or contribute to the discussions here - and I don't go in front of general public audiences to tell them that they're idiots if they don't switch to Unix - I do what other Linux fanboys do: talk mainly to other IT people.

So, bottom line, who're we offending when we say that Linux is self evidently better than Windows? Windows fanboys - and who do your Wintel fanboys address when they demonstrate their skills by flaming us in response? Linux fanboys - and in neither case is anyone talking to the general computer using public.

Bottom line: since Linux advocacy doesn't address the "general computer user" audience, it can't be annoying that audience - and if we're scaring your Wintel friends? well, our commitment to openness means that they can play catch up at any time simply by going back to school and learning something about computing technology.

Your reason number five is a masterpiece: 'Linux is "simple"; Windows "just works"'. Wow! The reversal is amazing - to find humour, in the form of an Apple slogan applied to Windows, in such an otherwise serious effort is genius - sheer genius.

In contrast to the other three, number seven has something; well almost something, going for it. Here's your summary:

The single biggest downside for Linux isn't technical. It's that there's no single public face or advocate to drive adoption on the desktop. True, Linus Torvalds gets the tech industry's all-time good guy award. He's also a poster boy for how to keep an unruly, disparate community focused on maintaining a good technology.

Unfortunately, for all his strong points, Linus is a bust as an evangelist. If you don't agree, just compare Linus to Bill Gates and Steve Jobs.

In reality, Linux did have the Cult of Torvalds going for it with the mass media lovingly building up their own invention: a Linus Torvalds bravely standing against the titans of American capitalism to single handedly wrest control of an industry away from all those horrible Republicans.

Unfortunately even the Boston Globe couldn't fake that story after the SCO lawsuit catapulted Torvalds into IBM's embrace. So, really, if you wanted to be honest about this you'd have said that the hot breath of politics driving mass media support for Linux came and went during the nineties and left Linus Torvalds pretty much where he was before it came along: doggedly building support for his "free Unix for the 386" one geek at a time.

The bottom line on this, however, is that Torvalds earned his success doing what he wanted to and doing it extremely well, that Jobs seems to have succeeded differently but every bit as honourably, while Mr. Gate's success seems mostly due to having had the right parents and only a lawyer's ability to tell right from wrong.

Signed: Unix FanBoy.


And tomorrow.? The real reason Linux hasn't (yet) made in on the desktop - did I mention trolling for readers? oh yeah, I did... ok.

Paul Murphy wrote and published The Unix Guide to Defenestration. Murphy is a 25-year veteran of the I.T. consulting industry, specializing in Unix and Unix-related management issues.