With the bickering about what Dell will and won’t do to provide Linux on their desktop machines, it seems to me there’s a much easier way to introduce GNU/Linux into the world. Scrap it!
GNU/Linux is a great project, with fantastic underlying technology, but its name conjures up images of geek humor, stuffy Unix-like machines from a 1970s university, and—worst of all—servers. Not desktops. Therefore, I propose we set up a business that will fork the kernel into a new project, let’s call it MegaOS, and install that onto a set of new machines, each branded with our corporate MegaOS logo.
Next, we ensure that MegaOS will remain stable for many years by releasing only security fixes. After all, we’re looking at the desktop world where users neither know or care about how the memory manager works, or if one version is better or more “structured” than another. Windows has many years between each major iteration and this has served them well. I counted six versions in the last 12 years (Win95, Win98, ME, Win2K, XP, Vista) which limits the number of times users have to go through “driver deprecation hell” as companies struggle to produce updated drivers for the new operating system. Also, because MegaOS would not change the driver mechanisms as often as Linux, the amount of driver recompilation is reduced. This eliminates the hardware obsolescence that can occur when minor drivers aren’t updated and recompiled for the latest kernel. This is an end-user machine, remember, and they are not expected to understand automake, let alone gcc compile options.
And, speaking of drivers, this leads me to the next part of my solution. Sell MegaOS only on known hardware. Not necessarily custom, but known. It can (and probably will) have a PC architecture underneath. But if the BIOS looks visually different from what PC users expect, it will appear like a completely new machine. So, find a set of decent, home-user-ready, hardware for the various components (such as audio, video, and networking) and get those free software drivers working perfectly together. Then, enter into a business partnership with those manufacturers so that chipset XYZ123 is always chipset XYZ123 make sure they will be available for two years, and will be fully compatible with the drivers. Dell, in the past at least, have been lacking in this regard, and shipped components with identical part numbers but significantly different internals, causing drivers to break. Remember, of course, that if we control the hardware, the number of drivers and driver combinations in the wild will be comparatively small compared to the GNU/Linux of today.
Naturally, with only custom hardware in the original box, we would like to restrict those who can produce third party hardware, which would break our MegaOS system and becoming nothing more than a poor Windows copy. In reality, we don’t. We just need to make the end user aware of the value of buying approved hardware with a “Designed for MegaOS” label on the box. By using a trademark (yes they can be evil, quiet at the back!) there is legal recourse for abuse. However, the rule for gaining the right to use this label would be very simple: any third party hardware manufacturer must create free software drivers and make them available. This, by implication, ensures these drivers will be available for back-porting into GNU/Linux proper, and the rest of the community.
Part of the MegaOS budget will be spent on aesthetics. As a technical person, I’ve always consider this an easy job, but looking at the beige boxes cluttering my studio I guess not! Therefore, we’ll hire someone to design a nice shiny looking box, stylishly coloured, and feature it at the end of adverts where sexy 20-somethings dance in silhouette to popular music... for example.
But hardware aside, what does the average user want? Basically, they want applications, and they want them to work. Preferably together. This is already available with either of the two major desktop solutions. Which? I don’t care! MegaOS will simply pick one, choose a suitable window manager, create some clever new icons and graphics to give it a unique feel, and release that. Face it, most non-technical users won’t change the defaults anyway. Except for, perhaps, the background/wallpaper image.
From here, the software can be chosen according to the whims of the MegaOS company. A media player, an office suite, internet tools and CD burner software is enough for version 1. Since we’re working on known hardware, compatibility testing will be easier, and would focus the development of Windows-compatibility tools (WINE, CrossOver Office, et al) ensuring a migration path exists for Windows users.
Games might still be a sticking point, but much of the industry is moving towards console and mobile so that’s half the market covered. The other half of the budget can be spent wooing developers and porting WoW and SecondLife to MegaOS. Because we control the OS, we can ensure closed source companies can remain closed source until they’re ready to switch. By which time, the users will have experienced enough free software to force their hand.
At no point are we abusing any of the free software ethics, as we will be adhering to all license agreements by supplying the source from our web site. Nor are we taking control away from the user. We are limiting their initial choices certainly, but not removing them. After all, the traditional end-user consumer won’t change the defaults because they don’t need to, or they have a fear of breaking things and will work within whatever limits the system has.
So there you have it, a modest business proposal. In short: fork GNU/Linux, ship it on known hardware with working drivers, put it in a pretty box, pick a friendly UI, bundle principle lifestyle and productivity software, provide a migration path, and give it a snazzy name.
Remember, it worked for Apple...