Searching for the Linux of Mass Appeal

As a computer service volunteer and Linux advocate, how can I get the average computer user to adopt Linux?

This has nothing to do with what I happen to like, but what I find actually doing the job. Having to work around minor issues is no sweat for me, but the average refugee from Redmond isn’t gonna like it. Just getting past the inertia of moving away from what comes bundled on the PC from the store means we have to be significantly better. Gamers and hobbyists already know what to do, but most home users have too many other concerns in their lives. We want them; we need that mass appeal to overcome hindrances we have faced since the days Linus first posted his ideas on Usenet. We have to offer a compelling reason to switch.

Simple security is not enough. Everybody knows Linux is better at security, if they even know what Linux is. I remain firmly convinced neither the Russian Mafia nor NSA will find a typical Linux home PC an easy target for cracking over the wire. This is because of fundamental differences in structure, not just market share. There is a huge difference between a system which starts with all the gates closed (Linux) and one which has no fence (Windows). But the people who bring home a Vista box from Wal-Mart will simply run what comes on the system, even struggling with ill-considered security measures, will not install something different unless it’s worth more to them than the presumed value of that bundled OS.

What they want is easily determined from anyone who serves these same people when the security measures fail, or when something simply doesn’t work. I can convince them to be more careful in selecting their optional hardware add-ons, because they are used to being ripped off via false advertising. I can’t convince them their entire OS is broken because they can see it running on their computer. What will move them is something which works with the hardware which came in the box. We are doing a lot better in that department, and I’m really impressed with the proliferation of projects grinding out working drivers for just about everything. We’ve also done a great job with Linux gaming software. We will probably never capture the insane gamers who live inside their PCs until we capture more market share and convince game developers to add a Linux port, but that’s the whole point. We have reached out to every marketing area except one: the mass of users who simply want their system to work better with less interruption. They consider a computer essential, but not at all central, to their daily lives.

These people will all too quickly revert to Windows, demanding it loudly, if the daily activities for which they have been using their system is in any way hindered. They’ll forgive a drab and boring desktop if it they can find everything and use it pretty much as they did in Windows 98. Over the past three years I’ve been testing systems based on what I see my clients and friends doing on their PCs at home. Please note, we’ve already covered Linux distros with rolling release or frequent releases, and which refuse to support any previous release longer than one year. Those are out of the question. I find a 3-year support cycle is the absolute minimum. As I compare what they see with what I experience in the Land of Linux, here are some observations:

1. GNOME is slightly ahead of the game over KDE and others. More stuff works, more consistently, right out of the box. They include things like the default CD player, screensavers, applets, and frequently font rendering. People don’t give a rat’s patootie about clutter or simplicity as a matter of aesthetics, but once they have it set up, it better work first time, every time. GNOME has a better track record. It also has a bigger share of the common use applications interfaces. While plenty of KDE stuff works better technically, and is more familiar in terms of general appearance, Windows refugees are more likely to find GNOME usable. As a side note, I find KDE 3.x still broken, but it seems abandoned now. At the same time, I doubt 4.x will be usable for at least a couple of years, particularly from the standpoint of common users.

There is one issue so large I can’t overstate it: Adaptive Technology. Right now, KDE is very poorly covered by this. I’ve seen a range of add-on software for Windows, from the cheapest to the most expensive. I’ve seen how it works, and how the vision-impaired want it to work. I haven’t tested the Linux version, but that will begin next week. Fully blind users are another issue, but in my reading I’ve seen nothing to compare with the Orca Project in Linux, and it is based on GNOME. The built-in stuff with KDE won’t do it, though it covers motion-impaired users well enough. A significant part of my service is to vision-impaired people, and I can’t afford to invest time in something which isn’t even close to readiness.

2. Hardware detection is everything. It’s not just a question of, “Will it work in Linux?” It’s a question whether the installation process of your chosen distro will minimize the pain. Frankly, Ubuntu/Knoppix and friends are loosing ground on this front where I’ve tried it. I won’t pretend to know what’s behind it, but the last few releases have given me fits with display setup. The GUI config utility is entirely too simplified, and there is no means for simply choosing the monitor from a list, nor even setting the hardware specs if known. There are at least a half-dozen systems where the boot process froze, and one or two where all the expert boot incantations in the world can’t get past the hardware detection freeze. On the same boxes, RedHat clones or SUSE do just fine. I’ve had to quit offering the former CDs, because they just don’t work.

3. There is such as a thing as “too secure.” Here is where CentOS and other RH clones break down. Security should work in the background. The number one complaint against Windows I hear is the way security software dominates the system resources and too often fails in one way or another. The primary consumer complaint against Vista is the obtrusive security measures which serve only to make it hard to use. In a hundred little ways, those server distros do the same thing. Yes, we want security on our servers, and server distros are the primary source of stability over the long haul. However, persuading CentOS to loosen up for a different mission is quite a large task. We can add all sorts of third-party goodies without breaking a sweat, but I have found, in consultation with developers and sysadmins, all that tweaking of the underlying assumptions is probably not worth it. Somebody needs to step forward with a different model, or we’ll never gain that wider acceptance.

4. For now, openSUSE is probably the best bet. Complain all you like about compromise with Leviathan, and apparent bloat, SUSE on the desktop is the current leader for the mass market. This, even though their latest release is badly broken in some places. I’m going to be offering to my clients and friends the DVDs from 11.0, particularly those with 64-bit hardware. People who could care less about the technical differences between 32-bit and 64-bit are often the first to notice the difference in multimedia played on them. It’s no longer just databases and heavy graphics rendering questions; 64-bit sound and video playing is noticeably better than 32-bit. I’m surprised how little this is noted in the Open Source articles I’ve been reading lately. SUSE was quite early in the game of releasing 64-bit desktop stuff, and naturally holds a lead in some aspects.

It won’t matter I’m not happy with the way SUSE tends to run a little more slowly than other distros. It won’t matter I prefer KDE (3.x) to GNOME, the latter is better for my clients. I’m finding the difference between 3 years’ support and 5+ is just small enough I can probably sell it, at least from the way my clients and friends react. Should there ever come a day when Novell wises up and releases the update SRPMs for SLED in a more timely manner, they’ll steal a huge share of the non-paying adoption market. Yes, that won’t make them any profit directly, but it will make a huge difference when they grab a much bigger market share because of the direct feedback on improvements which keeps RedHat cooperating with the likes of CentOS and Scientific Linux. For now, Novell’s policy is to restrict access to update SRPMs, and allow non-paying customers only an annual “service pack” ISO. A year is a really long time in the field of computer technology. While patching and recompiling from source is easy for folks like me, a build-ready SRPM is far simpler and quicker to deploy on a large scale — that’s the whole point behind having RPMs.

So far, no distro is a shining star. SUSE is just barely enough better than Windows I have some hope of selling it. I’ll be running it as my primary work station only so I can keep track of little things which make a difference to clients. I still prefer CentOS for my own use, but I don’t have enough computers, nor enough space for more of them, to make like some commercial operation.

Update: Notice the follow-up to this post — No Paradise