With the semi-recent approval of OOXML by the ISO, an issue has come up over how well Office 2007 implements Microsoft’s own standard, OOXML. Griffin Brown tested the compatibility of the latest Office 2007 with the latest OOXML standard and found that the results were less than perfect. My first reaction, of course, was outrage. How on earth are competitors supposed to implement Microsoft’s “open” standard if even Microsoft itself cannot implement it correctly? As I read through the original post, though, I believe I may have been too fast to jump to that conclusion.

Griffin Brown tested Office 2007 against two versions of the OOXML standard - the transitional and strict versions. When tested against the transitional version (a less strict version) Office 2007’s implementation returned 84 error messages, which Brown calls “relatively few.” Many of the errors, he says, are “all of the same type.” Additionally, in the example he gives, the Office 2007 version uses “on,” which was used in a previous version of OOXML, instead of “true,” which is the current version. Judge for yourself, but this sort of thing seems OK to me as long as Microsoft corrects it soon.

When Brown tested Office 2007’s implementation against the strict version, however, it received many more errors. Apparently, 17MB (around 122,000) of error messages were output. This seems much less justifiable to me. Seriously, 17MB???

Before I make a judgment, I want to know two things:

1. Which version is more important, transitional or strict? Groklaw says about the transitional version that it is “one no one on the planet will be using,” but this statement is not backed up or justified.
2. How common is it to have errors in format implementations? For example, if most implementations by programs such as OpenOffice are perfect, Microsoft’s implementation would seem inexcusable. On the other hand, if almost all implementations contain many errors, Microsoft’s implemenation makes more sense.

There is, of course, an argument that those questions don’t matter and, since OOXML is Microsoft’s own format, it should be implemented perfectly. This, however, seems unfair to me. Instead, I will wait to see how others react to Brown’s findings. I just want to make the point that, no matter how many other flaws OOXML has, Brown’s findings might not be as damning as they seem.

Note: I want to make it very clear that I am not excusing Microsoft, I am simply waiting for more information before judging.

I recently came across a YouTube video created by Blogeee.net which shows both the Windows Eee PC and the Linux Eee PC (I assume each with the same specifications) starting up, launching Firefox, and shutting down. It is very apparent in the video that the Linux version is much faster, but, since only the startup is done at the same time for both machines, it is hard to tell just how much faster. I timed each part (starting up, launching Firefox, and shutting down) to see what the time difference really was. Here is what I found:

Startup
Linux: 30 seconds - Windows: 54 seconds
Launching Firefox
Linux: 4 seconds - Windows: 16 seconds
Shutdown
Linux: 6 seconds - Windows: 68 seconds

(Note: Numbers are approximate. Firefox start time is from clicking on the Firefox icon to the dialog about starting a new session and then from the end of that dialog to when the windows appears, or, in other words, everything but loading the web page and waiting for the user to click on new session or resume session.)

Those numbers by themselves are impressive, but they get even more impressive when you look at how many times faster the Linux version really is.

Startup: Windows takes 24 seconds longer. Linux is just less than twice as fast.
Launching Firefox: Windows takes 12 seconds longer. Linux is four times faster.
Shutdown: Windows takes 62 seconds longer. Linux is slightly more than 11 times faster.

Look at it, Microsoft. Vista would barely even run on these things and Windows XP is much slower than Linux. This is very likely to be Linux’s first major success on the desktop.

Recently a diagram that depicts Microsoft’s “empire” competing or fighting its competitors, particularly open-source. The diagram (older versions), while probably mostly intended for fun, shows two important realities for Microsoft:

1. Microsoft is huge. They will never just disappear overnight or over a year. If they die, it will be very slow as they gradually lose various parts of their bussiness to competitors.
2. Microsoft is surrounded on all sides. They are facing serious competition almost everywhere they go. No longer are they the only player in anything.

Although these two points may seem simple, and to some extent they are, they are important. If Microsoft wants to continue to survive, it must remember that it does not have a free ride. Whether or not Microsoft will be able to beat back the competition remains to be seen, but so many (myself included) seem to think Microsoft will fall.

EWeek is reporting on some unusual good news from Redmond, which is a rather nice change from the usual flow of negative Microsoft stories. Microsoft has released some ASP.net development tools for Linux. Although I can’t claim to understand all the technical details of exactly what they released, it appears that it is a development tool for Linux developers.

Although I suspect that not a whole lot of people will care about this particular piece of software, the precedent this potentially sets is great.

In the past, when Microsoft has said “cross-platform” what they meant was always “it runs on Windows and it at least sort of runs on the Mac.” Now that Microsoft has started releasing software for Linux, if only a small development tool, perhaps that definition will change to “it runs on Windows, Mac, and Linux.”

It is also possible that this is just an isolated incident of releasing something for Linux and it will not become regular, but I am glad to see some good news from Microsoft and I hope we see more in the future.

Computer hardware is constantly getting  better. Intel and AMD are constantly trying to have the better, faster, cheaper chips. Graphics cards achieve higher benchmarks and lower price points every month. 2GB, or even 4GB, of RAM is becoming commonplace. Everything is getting faster and costing less at the same time and this will probably remain true forever. Microsoft and most Linux distributions view these same hardware improvements differently, though.

When Vista came out, one of the big complaints about it was that is was so resource demanding. Although perhaps not quite as extreme as with Vista, it seems quite reasonable to expect that with each new release of Windows, you will need a slightly, or perhaps not so slightly, better, faster computer than before.

This is not at all true of Linux. Phoronix recently published benchmarks comparing Ubuntu 7.10 (the current version) to Ubuntu 8.04 Beta (the beta of the version that will be out very soon.) What did they find? Ubuntu 8.04 is only very slightly slower than Ubuntu 7.10. Even more impressive, when the did comparisons with even earlier versions, 7.10 had actually improved on former versions.

What all this means is that Windows is becoming more and more resource reliant while Linux essentially maintains its requirements. Microsoft is already seeing the effects of their sloppiness in bad reviews of Vista and having to extend XP’s life, but unless they change soon, they will see it even more, and pay for it too.

In the future, it seems likely that a computer with Windows will cost far more than a computer with Linux, not because of the price of the operating system (since hardware manufacturers are constantly pretending Windows is free, when in reality it is not) but because the hardware required to run Windows is so much more expensive than the hardware required to run Linux.

Returning to the present, Linux is already well on its way to removing all the road blocks to switching. Even when there are no road blocks left, though, Linux will still not win mainstream adoption, until it gets a simple reason to use it: it’s cheaper.

Of course, Linux is cheaper right now. It’s free, while Windows costs hundreds of dollars. Well, as you probably know, most places like Dell get it for much less than that. I don’t think anyone knows for sure (and it almost undoubtedly varies for deal to deal), but a safe guess would be that places like Dell get Windows for around $40 (just a guess). In addition, we repeatedly see cases of an identical Windows and Linux machine costing exactly the same price. It is not fair to give Windows an advantage like that, but it seems to be the reality. No longer will it even matter, though, when a Linux machine costs so much less than a Windows machine, just because of the hardware.

Just by continuing to use minimal resources, Linux is constantly becoming a more and more attractive option, and eventually this may be the single most important factor in winning Linux’s mainstream adoption.

Recently Best Buy posted a price for the upcoming Windows XP Eee PC on their website. The price is $399, the same as an Eee PC with Linux and the same hardware. What is wrong with that? Lots! You can read my full post on this subject for more details, but the very short summary of what is wrong with pricing the Windows Eee PC the same as the Linux one is that either Asus or Microsoft is absorbing the cost of Windows. If Asus is absorbing the cost, it is not their business to determine the pricing of other operating systems or if Microsoft is pressuring Asus to absorb the cost, that is almost certainly anti-competitive,  and if Microsoft is absorbing the cost itself, that too is almost certainly anti-competitive. (Not that I am a lawyer.)

So basically, if the otherwise identical Windows version and Linux version cost the same, I will be furious and you will almost certainly be seeing more about it here in the future.

You may be wondering why I am not already furious. After all, hasn’t Best Buy confirmed the price is the same? (Best Buy’s page for the Windows version, Linux version) The reason is that I do not trust the numbers on the Best Buy site for two reasons.

First, the picture on the Windows page shows the Linux interface and the Linux page shows a blue screen of death. (Really just a blue screen without the death part.) Second, and much less significantly, the Linux version has an offer for a deal to save $10 on Norton 360, a security product for Windows.

For these two reasons, it is possibly nothing more than a mistake made by Best Buy. So until the pricing is confirmed, I will withold judgment.

News that another school system or government has switched to Linux seems seems to come  very frequently now. The latest appears to be news that 9000 Swiss computers are switching to Ubuntu, according to Duvet Dayz.

This trend, similar to the trend to cheap UMPCs, represents a significant future problem for Microsoft, because students will grow up and be familiar with Linux and presumably use it at home and possibly at work too. In the case of governments switching to Linux a similar thing will presumably occur, except that (hopefully) most of the workers are already grown up.

I said before that this trend is similar to the trend towards cheap Linux UMPCs and that is true, but there is one important difference. Microsoft has seen the hole that Linux UMPCs are creating, but they have not yet shown they have seen the education and governments hole. Even if Microsoft cannot compete with Linux’s advantages, they can use FUD (fear, uncertainty, and doubt) and force to accomplish a lot of things, but not if the don’t pay attention to where they are losing dominance.

The cheap Linux UMPC trend will continue to help Linux, though Microsoft can be expected to resist it. They may or may not win that fight, but they will do damage, through real competition or through FUD and force. The education and government hole, however, is currently being ignored or not seen by Microsoft. This gives it a chance to really take off, just as the Eee PC did before Microsoft started paying attention.

As I see it, there are two ways in which something can become a standard:

1. By general consensus: people use it so other people use it and pretty soon it is a “standard,” even though it is not necessarily an “official” standard.
2. It is declared to be a standard through some agreed upon process.

Although the ISO just approved OOXML as a standard, it is not a standard by either of these definitions.

By the first definition, OOXML would have to already be widely accepted and used. Although it is very difficult to scientifically prove that OOXML is not being used, but if it were truly a standard you would see most office documents in OOXML, which does not seem to be the case. I have only seen a very few of them in OOXML. So OOXML is not yet a standard by general consensus.

By the second definition, it would seem that OOXML is a standard, because the ISO says it is. Not so fast. I said it has to be a standard based on an agreed upon process. There is a process for ISO standardization (all the countries vote and if enough of them vote to approve the standard, it gets approved.) This process, however, was not followed. There have been many incidents of suspicious (to say the least) things going on during the process, including events that seem like 100% cheating. Unless it can be shown that the standardization process was fair, OOXML is not a standard by this definition either.

In conclusion, until OOXML is truly adopted by the majority of people (preferably not through Microsoft forcing it on them) or the ISO shows us the process was really fair, ISO standardization means nothing and OOXML is not a standard.