heironymouscoward's Journal: Crash Computing

Last week saw some stimulating debate on Slashdot, especially concerning Microsoft. Inspiration strikes me to jot some random analysis apropos to Microsoft, open source, Linux, security, et al.

1. Microsoft and Sun

$2bn for Sun? Perhaps it's pocket change to Steve Ballmer, but it's a lot of money. And for what? No-one gets rich by spending freely. What can Microsoft expect to get for its cash? What would you pay Sun $2bn for? Cringely believes it's a settlement designed to ward off anti-monopoly charges from the EU et al. But $2bn buys a lot of lobbyists. Hmmm, here is what I think Microsoft wants for its money.

First, to kill Java. Why? Because MS is obsessed by platforms (think of the scores of platforms MS has produced and pushed over the years), and it feels that Java provides a platform for the competition. Every Java programmer, every Java application represents a plus for the opposition. The opposition is, of course, IBM, and Microsoft want Java dead, and all development pushed to .NET.

Sun's decision last month to not release Java under an open source license makes sense for reasons other than "we need to keep control". Perhaps it was "we are almost bankrupt, and we're going to sell our first-born child in a desperate attempt to get some cash."

$2bn is probably a good deal for Sun, but it's a sell-out that will cost the industry much more, if it succeeds. It's a deal that will bring Microsoft much more than $2bn. If it succeeds.

So, can Sun squash Java? IBM has been working very hard over the last 5 years or so to develop its own Java platforms, faster and more stable than those from Sun. But as far as I know, it does this with permission from Sun.

So, some predictions coming up: Sun will announce that it is slowing, and finally cancelling further development of Java and proposing a migration to .NET. Furor will ensue. IBM may try to buy Sun or Java. More likely, it will announce that it is legally protected to use Java for the next decade or so, and it will announce its own new platform and migration strategy. Free software advocates will say "we told you so!" and they will be right. The GPL, after all, was designed to combat exactly this kind of threat.

What else might Microsoft be after? I suspect they regret very much that OpenOffice.org escaped into the wild. They could have bought Star Office for petty change, and killed it off. Instead Sun bought it, cleaned it up, invested in it, and released it as OpenOffice.org. The final cost to Microsoft will be enormous, as OOo represents one of the only serious contenders to MSOffice: it is a solid, complete, stable, and portable office package that provides an easy migration for many people from Windows to Linux, and must represent a nightmare to Microsoft strategists.

Microsoft may appear arrogant but I'm certain that they are perfectionists, obsessed with detail, control freaks on a massive scale, and the escape of OpenOffice must have traumatised them. I'm just speculating, but I suspect that there have been many meetings and memos along the line of "how to avoid another OpenOffice". $2bn is not a lot when seen as insurance to protect MS Office.

The cat is out of the bag, but OpenOffice still enjoys significant support from Sun. If Microsoft can't remove the OOo installations from PCs around the world, it can at least try to cut the air supply.

Second prediction: Sun will announce that it is slowing development of StarOffice and stopping all support for OOo, for financial reasons. Again, furor in the industry, with hordes of free software people saying "see, thank god/RMS for the GPL", and they will be right, once again. Only the paranoid survive, and Richard Stallman is one of the best.

Can MS kill OpenOffice? No, they cannot. OOo is a powerful weapon again Microsoft, and there are many people with the means and the motive to use it. Either Novell or IBM will announce that they will host the OOo project and provide support. I suspect it will be Novell, but with backing from IBM.

Third prediction: Sun will, within no more than two years from now, either file for Chapter 11 protection, or be bought by Microsoft or IBM. They are a zombie company and are now surviving by selling their own living organs.

2. Moving Walls

Microsoft is in a trap of its own making. The trap is this: it has dominated the market by bringing Internet computing to the unwashed masses. These unwashed masses are now being eaten alive by viruses, spywares, trojans, worms, porn, spam. The industry response has been to blame the users for not applying patches, for surfing bad sites, for clicking the wrong buttons, for file sharing, and so on. Blaming the users is stupid, and useless. Most Windows users are unable to protect their systems. A new XP system is infected within minutes of connecting to the Net. Even a savy user has a hard time downloading the necessary patches.

Like Indiana Jones stuck in the corridor of death, the trap has more than one moving wall. Computing has become a commodity infrastructure, thanks to the technology curve that has brought software development costs down to almost zero. I've discussed this before, called it "Heironymous' Law": all technology follows a curve that tends to zero. Someone once answered sarcastically: "yes, this is why a Jumbo costs more than a small country", which actually proves the point. Even that most complex of machines, the airliner, costs a fraction of what it used to, 20 or 50 years ago.

Microsoft sells two basic products: Windows and Office. All the rest is fluff. It has tried hard to penetrate the enterprise market, but it's not managed to get into large companies except on the desktop. Microsoft's server products are somewhat of a joke, and it keeps them in place only by tying them closely to its desktop products, not by intrinsic value. I.e. businesses use Exchange not because it's cheap and fast, but because it is necessary if they want Outlook to work well.

Basically, Microsoft makes money by extorting it from its captive clients. All very well, one may say, this is the market in action.

But this is where I disagree. Extortion works only when there is no competition, and this is the kicker: computing has become a commodity, and there is lots of competition. It's not very well organized, it's not as coherent and agressive as Microsoft's offering, but it works well, is cheap, and is above all, safe.

There are several ways that the Windows monopoly will crash, but crash it will. There is no way to de-commoditize the market and there are no solutions to Windows' increasing insecurity.

The first way: a mass ephiphany that there is a better way, lead by the media pleading for more secure computing based on better infrastructure. Chance of happening? 5%, no better.

The second way: governmental intervention, following a mass breakdown of IT due to a worm. Chances of happening within 2 years: 50%, I guess. Computing is so vital to national interests that there will be simple regulation of the OS platform: if it is not secure, it will be made illegal.

The third way: market forces. Every firm that has allowed itself to be captured by Microsoft will find itself failing to compete with firms that use the cheaper, better, safer commodity computing. Firm A uses Exchange, IIS, Windows XP and spends 10% of its budget on IT. Firm B uses sendmail, Apache, Linux and spends 2% of its budget on IT. Guess which firm makes more profits and invests more in product research and finally dominates the market?

Microsoft are trying very hard to sell the idea that Windows is cheaper than Linux, but this is really besides the point. Firms that are stupid enough to believe the argument will fail. Natural selection will eliminate the vulnerable and promote the resistant.

Chances of market forces crashing the Windows monopoly? 100% within 10 years.

3. Time for some terminology

I propose a new term, "Commodity Computing" to replace "open source" as a marketing term for cheap, reliable, safe IT. Commodity Computing is easily defined: it is open source, it is portable, it conforms to independent industry standards.

All governments should be legislating to promote the use of Commodity Computing. Countries that do this will (like Firm B) find themselves at a competitive advantage over countries that do not.

My final prediction: the last country on earth to embrace Commodity Computing will be the US. $50bn of lobbyists buys a lot of time.