A think tank (Damn! I need to join one of these and rest my head) just popped up saying that computers should be sold without operating systems.
Let's bypass the fact that Microsoft clearly has a massive image problem right now and follow this thinking on its own "merits" for a moment. This think tank has concluded that because Linux isn't on many PCs, the reason isn't because Linux isn't competitive, it's because people don't get their choice of operating systems.
Gee, I think that car and tire companies are too close, so let's sell cars without tires, or you know, window companies have a lock-in on houses, so let's sell houses without windows, or doors, or carpets, or . . . When did it ever make sense to address a competitive issue by forcing consumers to build their own products? Let's see, who has the most loyal customers? Apple. Who would this hurt the most? Hint: It isn't Microsoft.
Seriously, who would this policy hurt the most? Microsoft, which actually does sell a class of OS products for the home PC builder; a small PC manufacturer which includes all the stuff to work with a massive variety of hardware devices; Apple, which doesn't have a hardware-independent OS; or Linux, which the home-builder market appears to avoid like the plague? (And which has this little overabundance of distributions issue).
And this comes at a time when one of the biggest initiatives in the industry I know of is to use Linux as a pre-boot OS under a hypervisor, which would, if it succeeds, actually make Linux the default OS on hardware, but be illegal under this proposal.
I'm seriously starting to wonder whether there are too many people in Europe whose elevators don't go to the top floor.
Why People Who Don't Know PCs Can't Get It Right
A few years ago, Larry Ellison of Oracle and Scott McNealy of Sun came up with the idea of a thin-client platform to displace Windows on the desktop. It was actually a good idea, and if it had someone with the remotest clue how the desktop market actually worked been behind it, we probably would have a massive number of thin clients in the market now as opposed to a tiny fraction.
The problem is people get so excited about their ideas, they forget that to move on a market, they have to understand the existing market dynamics. Part of those dynamics are an existing robust ecosystem made up of people who don't actually work for Microsoft, but live symbiotically off the infrastructure that company has created. These folks specify and support the hardware they and their companies use and, if you can't capture them, you can't move on the market.
In addition, you can't move in under just a cost-savings message. If that were the case, a lot more of us would use second-hand hardware or really old stuff. You also need to provide some key benefit the user wants, so they will make the switch. That is why Apple appears to be gaining share competitively; Apple is providing an ease-of-use/reliability/coolness benefit that pulls customers to its platform. Where Apple misses is that for corporations, you need at least two competing vendors using your platform so the corporation doesn't have to go through the massive pain of sole sourcing. (Intel had this same problem and helped AMD in the beginning to get into the same business). Finally, you need an out-of-box-experience that is acceptable to the market.
For thin clients, there was a massive up-front cost, no two solutions were alike, and users generally hated the things once they had them. This has mostly been mitigated over the years, but thin clients, blade PCs, and other alternative PC platforms largely stillhave the competitive bidding problem keeping them from mainstreaming.
Why Bare-Bones PCs Won't Work
Simple: drivers and out-of-box experience. To get around this, you'd still have to get the OS from the OEM, as much of what goes in a new PC is, well, new. Linux drivers (assuming the buyer even knows which distribution will work) tend not to be first on the priority list and probably aren't even in the Boxed Windows OS on the shelf. Then, if you load the OS and the result breaks, who do you call? It wasn't the OEM's image, and while you can call the source of the OS, if the machine was never actually loaded with an OS in the first place, how does that company do any remote diagnostics?
I can picture the call:
Me: Hello, PC Vendor. My machine won't boot. Dell: This is a recording, hit the number for the OS you installed for help. Ah, you need to call xxx and get help from the OS Company. OS Company: This is a recording: If your PC won't boot, please call the OEM for assistance as it is probably a problem with its drivers.
And so on...
Machines are largely tested without loading the OS now, so this is actually easier to do these days (old PCs without an OS were originally preloaded with something like DOS). But think how the out-of-box experience would go. You'd get the PC, which might come with a driver disk -- you are kind of screwed if you have a laptop without an optical drive. Then you'd take your OS of choice and try to load it, putting in the driver disk when asked. Now you could put the drivers on the hard drive, you say. Which file system do you use and what happens if the OS formats that partition when it installs? The word "screwed" comes to mind.
Now, there is no way the OEM has matched up the version of the OS (and it will be patched over time) with the drivers, particularly if the PC is either new and the OS has been on the shelf for a long time or, inversely, the PC has been on the shelf a long time and the OS is new. There will need to be some kind of Garanimals matching system. Hey, I just had an idea, maybe a 3-year-old came up with this idea?
These will not be happy moments, and you'll either get things that don't work or screens of death (once again, Windows is designed to deal with these problems and has a massive driver library, but even it likely often will have issues with new hardware). This means the user is basically completing and QCing their own PC.
For retail, they will likely step in and load and test the OS you want for you, and clearly have incentives to sell you the one that generates the most cash. So, let's see, which vendor has the strongest retail position and best incentive model. I doubt people in "think tanks" study retail much.
How Could it Work?
Well, if the market does move to the pre-boot VM model I was talking about, the hardware can be abstracted and then the driver problem largely goes away. In addition, networking is up in the pre-boot, along with advanced diagnostics, so the systems can be better tested before delivery. Then, if networked, the OS of choice can be downloaded from the OEM, which can still ensure the experience. The strange thing about all this is that this is pretty much exactly where the industry is trying to go in the first place. But this sure isn't the way to get there and someone still needs to complete Linux so it has an adequate out-of-box experience, and the OEMs are looking to do this themselves like Apple did with BSD UNIX. It will likely take a year or so to get the work done though.
If Apple were forced to support other hardware, you wonder whether the result would kill Apple like cloning almost did (recall Jobs killed that when he returned). I still think cloning could work for Apple if it were done right and the Apple OS is cooked. (The problem before was Apple hardware wasn't cost competitive; it could be this time around).
If PCs had to be sold without operating systems, the result wouldn't work, and both Linux and Apple in particular would be hurt the most long-term, not to mention the OEMs that likely would pass the not-insignificant additional support costs for the "alternative" operating systems back to the buyer.
The problem with Linux isn't that OEMs don't want to use it; Linux isn't ready for the desktop. The platform does not provide the required out-of-box and support experience. Dell is trying to use it, and so is Lenovo, but it's neither a big seller nor is it particularly profitable. Someone has to market Linux at some point, and it doesn't throw off enough profit to cover a good marketing program.
Here is a hint: When Apple took the MacOS and allowed cloning for a short time, its market share increased sharply for the OS (it suffered in hardware, but that was a different problem that doesn't apply here). Apple took UNIX and in a short time created a truly competitive OS -- the problem is that it's tied to Apple hardware. All the Linux folks have to do is do what Apple did, but don't tie the result to one hardware vendor. It's foolish to run around destroying the user's experience before hard OS work is done. You can't force a solution on people; the fix is to make the solution attractive enough so people want to use it.
Maybe the new European Linux motto is "Getting the hard stuff done is a pain in the butt, so we get our users to do it for us."
Yep, this will work. Lord, what do they put in the water over there?