Originally Published: Tuesday, 25 September 2001 Author: Rudy de Haas
Published to: opinion_articles/opinion Page: 1/1 - [Std View]

Pessimism at LinuxWorld Misplaced

In this startling essay author Rudy de Haas lays out his strategies for the future of Linux and open source technologies both at home and in the corporation. Haas exhibits an uncanny understanding of the factors that motivate companies and individuals, and this thought-provoking essay should be required reading for anybody who thinks of themselves as an open source advocate.

The Future of Linux

Many of the comments emanating from LinuxWorld in San Francisco at the end of August seem subdued, even depressed, as if somehow the money is gone, the excitement is over, and only the hopelessly loyal are left to fight on. That's nonsense, the future of Unix, including Linux and FreeBSD among others, has never been brighter with all of the needed technological pieces now in place for a Unix takeover of most corporate computing.

That Unix is better, cheaper, faster, and more reliable than Windows does not suffice as the basis for nation-wide defenestration (the throwing out of Windows). For that to happen we need a cultural change, a massive social movement in which non technical people shift their allegiance from one set of ideas about computing to another.

The emotional basis for that change, the anger and resentment users feel about dealing with shoddy products, ever escalating costs, and companies (or systems departments) that couldn't care less about the customer, is solidly in place; but the alternative vision, the set of ideas that crystallize the issues and precipitate action, is not.

Earlier this year I wrote a book - The Unix Guide to Defenestration - on using Unix to deliver increased value from corporate computing but it wasn't until I read a quotation attributed to Jeremy Allison at the latest LinuxWorld to the effect that Linux, to succeed, must take over the client, that I finally understood what a comprehensive alternative to the Windows monopoly might look like.

I think he's right, Unix does need to take control of the corporate desktop, but I don't think we can do it with Linux because better, cheaper, faster, just isn't enough of a motivator for most of the decision makers involved.

The Linux desktop scenario, in which users are offered a choice between a low cost, high reliability system that works and a high cost, low reliability one that often doesn't has been played out before. In January of 1985 buyers were confronted with two main personal computer architectures:

  1. the MacXL, at a list price of $5,495, offered 1MB of RAM on a MC68000 16/32bit chip at 7.54Mhz with a 10MB disk, a 720K 3.5" floppy, the MacOS GUI and a full suite of graphical applications on a 480 x 640 screen; but,
  2. for only five dollars more, the IBM PC/AT offered a maximum of 256KB of ram on a 16bit chip running at 5.77Mhz with a 10MB disk, a 360K 5.25" floppy, PC-DOS, a BASIC interpreter, and a 24 x 80 green screen,

and not only did they overwhelmingly select the PC but then they massively endorsed magazines that argued that the PC was cheaper [!] faster [!] and came with more software [!]. (And they're still doing it, see the note that the end of this article)

Today's Linux PC (or Mac G4 - now also a Unix PC) is in the same position: yes, it's better, cheaper, faster; but no, corporate personal aren't buying it and for the same reason: there's no compelling personal motivator for them to respond to and business reasons just aren't enough to balance out the personal risk of being seen as stepping outside the bounds of majority opinion.

That the alternative vision must include better technology, better management, and obtaining a better return on the systems dollar is a given but these things are clearly not enough. That's because, at least in my experience, people justify their decisions to themselves and others in terms of these kinds of publicly recognized values but make them more in terms of the effect they expect each choice to have on their perceived position relative to the people they know, or know about.

An Alternative Vision

Right now, for most people, making the Linux desktop decision has a downside - personal responsibility for any negative consequences - and no comparable upside. Our job as Unix evangelists, therefore, is to articulate an alternative vision that gives decision makers a powerful personal motivator for buying in.

I'd like to propose one: put smart displays, backed by big Unix servers, on most corporate desktops and put Linux, or FreeBSD, on most home machines with full integration for the MacOS X application shell and easy connectivity for the remaining Windows users.

The technology for this is the easy part: products like Sun's SunRay and NCD's NC900 let you replace the desktop computer with a much simpler machine that does a very good job of exactly one thing: interfacing between users and server based applications like Peoplesoft, OpenOffice, mail, browsers, multimedia conferencing, or any of thousands of other products.

The most important thing about smart displays is that they're completely maintenance free on the desktop. From a corporate perspective this means no Windows support people, it means getting off the upgrade treadmill, it means centralized administration, better security, near perfect reliability, and it offers a five to one IT staff reduction along with far more freedom to innovate.

But there's a catch - you don't get the benefits for nothing. The cost is a dramatic change in IT direction setting and management. From a senior corporate perspective the control structures and incentives which traditionally favor budget growth and staff expansion have to change to favor successful service delivery. In response IT management needs to morph from parasite to symbiot, from cost center to profit partner, while developing the trust relationships with users that underlie the partnership. That's also what The Unix Guide to Defenestration is about - adopting the Unix mindset to better manage IT investments - but this is much, much, harder than changing technology and will correspondingly take longer and endure more false starts.

From a user's personal perspective server based computing means fast, clear, graphics on big screens with instant response even on complex data crunching operations; it means noise free, hassle free, high reliability operations; it means freedom from Microsoft product churn; and, it means easy cut-and-paste access to multiple concurrent applications.

In a leadership based IT environment these benefits would be accompanied by a change in attitude within the systems organization as it shifts to an outward focus and so provides more freedom, and more flexibility, for its users. In most cases, however, management will not adapt well or quickly and users will rightly perceive the change as one threatening increased central control and a return to the arrogance, expense, and ineffectiveness of the traditional centralized systems manager.

In reality, of course, the evolving Wintel alternative is rather worse. Microsoft's current push into the data center with highly restrictive licensing and insecure software running on rackmounts stuffed with little machines is far more of threat to user freedom and desktop control than the switch to smart displays run from big Unix servers could ever be. Unfortunately most PC users don't want to understand this - and it's not exactly being stressed in the PC press - and so seem like the frog in the beaker of water being slowly heated to boiling: able to jump out but apparently unable to perceive any need to do so.

Linux at Home: The Personal Benefits

To address this the user must have a way to counter the centralizing tendency of server based smart display use. The route to that is Linux at home and that idea brings us to part two: the magic triple whammy of benefits that can make this alternative - Unix servers and smart displays at work and Linux or FreeBSD on PCs at home - compelling for reasons that go beyond cost and performance to embrace personal benefit.

The first part of this triple hit addresses the issue of user control. The perceived problem with centralized control doesn't have much to do with technology or the realities of data storage or daily operations. It pertains, instead, to the fear that systems people will extend their power over the user's desktop into power over the user. Redressing the balance while keeping the centralized architecture requires giving users power over systems people in ways that go beyond organizational hierarchies and role statements to actual enforcement of user privileges.

Linux at home provides a big part of this because the user runs the same software on his home machine - but directly under his own control. That forces operational IT decisions into the open because it is impossible to con or cow an informed user for long - and the opensource movement is all about open information. The traditional excuses for not doing something, or doing it badly, - no budget, no staff, too busy, too difficult, no benefit - are all pretty hard to get away with when users can just open a window on their smart displays to their (or someone else's) home machine to show the thing working - and can then point at a dozen sites to show what other people are achieving.

The second part of the benefit package is more directly similar to what happened when the PC beat the Mac. The first people in any office who convert their home machines, and their thinking, to Linux will get immediate and enormous social rewards by becoming the local agents for the imposition of user control on Systems: the opinion leaders their colleagues consult and whose views IT staff have to consider.

The third part delivers the additional social benefit without which the decision makers won't see advantage to themselves. That benefit consists of a relatively minor financial saving as they opt out of software churn and hardware upgrades and a major social coup as those savings let them respond in silence - a silence that screams "I'm smarter than you" - when others talk about the money they spent on their latest PC upgrades and Windows product licenses.

This triple whammy defines the motivator, the social benefit package, that lets the decision maker see the Unix decision as putting him ahead of the pack and therefore validates the suggested alternative architecture as a saleable package.

How to sell it to the masses is, of course, a different problem but I think we have to let people discover the social benefits for themselves. We need, that is, to talk up the technology and its benefits in terms of cost, performance, and reliability -the public values people cite in support of their decisions - but let people work out the social benefits to themselves -the real basis for these decisions - for themselves. Do it right and we'll unleash a wave of systems change on a scale that's never been seen before.


The ability of Wintel boosters to delude themselves and others is a source of continuous amazement to me. For example, there's a current article on Cnet.com in which the authors compare Apple and Microsoft tech support. To do this the writers picked 6 Windows issues and 7 MacOS X ones before attempting to answer them either from the documentation or via calls to support. In the end they asked Microsoft technical support four questions, paid Microsoft $70, endured a difficult and tedious access procedure, and spent 68 minutes in conversation with Microsoft staff, all to get two right answers (50%) and some bad advice.

They asked Apple representatives four questions, one of them - on Microsoft Outlook - twice; had an easy time getting through to support, got four right answers (100%) in 51 minutes (and a referral to Microsoft in another 19) and were charged nothing for the service.

This, in the Wintel mindset, evaluates as a tie.