This page was written in the 1990s. Today the content is partially obsolete. Some of the complaints I have on Microsoft Windows are no longer relevant, but on the other hand there has been many new reasons to avoid the system. The Orwellian named Windows Genuine Advantage for example.
So if you read this page, do it keeping in mind it was written in the 1990s.
This document is a statement of my opinions on Microsoft Windows, and my reasons for those opinions. When it comes down to it, people are usually not very interested in other people's opinions. I don't expect you to be interested in mine. Nothing forces you to read this. You may if you really want to.
When it comes to computers I prefer to use Unix systems, and most of
all Linux. From time to time I
get the question why I don't use Microsoft windows. Implicitly, or
sometimes explicitly, followed by: Everybody else (using computers)
does!
. That is like asking: Why dont you drive a Volvo like everybody else (in
Sweden)?
. Or: Why aren't you a customer of Nordbanken like everybody else (in
Sweden)?
. Or: Why do you work part-time and stay home with
your daughter instead of completely going in for your career like all
other (men)?
. Or: Why don't you vote for the Social Democratic Party like everybody
else (in Sweden)?
. Or: Why don't you speek chinese like
everybody else (human)?
.
The examples are probably enough that anyone reading this will realise
everybody else usually don't actually mean everybody
else. There is also the saying: 100 billion flies can't be wrong,
walk in the ceiling!
. (Often in a variant on what you should eat
instead.)
My point is of course that what might be right for a large majority not necessarily is right for me, or for any other particular individual. There are reasons for me to differ in all the aspects mentioned above. And this document explains one of these cases, why I avoid Microsoft Windows.
Let it be absolutely clear that I'm not any expert on Microsoft Windows. On the contrary, every time I have something to do with the system I'm strengthened in my urge to keep as far away from it as possible. If anybody find a factual error in my statement below, please do send me a remark.
The quality of Microsoft Windows is so low it crashes very often. Windows users appears to consider it perfectly normal to have to reboot the computer several times in a day. Compare this with a Linux system which, in the absence of hardware errors and with a stable (not development) kernel, can stay up for months or years. And nobody would consider that strange.
If I read the bug report Q216641 from Microsoft correctly it means that a Windows 9X-system never have been able to stay up more than slightly less than 50 days. I find the fact that such a serious error hasn't been discovered until several years after the release of the systems telling.
The lack of proper protection between user processes and the kernel in Windows aggravates this. I must point out here, though, that Windows NT and Windows 9X differ in these respects. But the fact that viruses in Windows are so powerful partly depends on the fact that an ordinary user process can modify system configurations.
The system doesn't leave any freedom for personal tastes. It is not possible to decide by oneself how to do things. I listened to a good illustration of this during a coffee break at my job. The issue was raised how to do get focus to a window just by moving the mouse pointer to it, without having to click on the mouse. The Unix and X11 user asks which window manager is being used, and refers to the manual for it, and some configuration in there which he doesn't know by heart. The one who knows Microsoft Windows don't think it is possible, but questions how wise it would be, and points on the risk for the pointer to accidentally move in to some other window while writing.
These answers are rather characteristic. In Unix there is a way to do what one wants. It might not always be possible to control from a GUI, and one might have to look in the manuals pages. But it is possible.
The answer from the Windows world when you want to do something
differently is: You want wrongly, change what you want!
.
I don't know if it in this particular case is true that it is impossible to have a focus follows mouse stgrategy in Windows. But very often the choices are very much more limited there. I definitely prefer the freedom!
The Windows system makes a connection between a particular working environment, (configuration, file setup, etc.) and a particular machine. All settings of various kinds are done globally for the machine. There is no way to make a connection between a user and an environment.
It is of course not the case that all settings should be connected to a user. What disks there are on a machine is not affected by the current user, nor is the IP number(s) it has.
But everything having to do with the individual working environment belongs to a user. Settings done in some tool, like a word processor or a web browser, belong in this group. So does the documents and files being worked on. The large majority of all choices done does.
Furthermore I want my own working environment regardless which machine in a network I use. On a typical site with many users, it is not without significant drawbacks possible to borrow workstations because of this deficiency. And/or just move around among them. In a home environment with maybe only a single machine, several users are forced to adapt to each other. And all this for no good reason.
There are many smaller deficiencies in Microsoft Windows. For example:
Often a graphical user interface is more of a burden than an aid. It is usually simple to do something with a graphical interface and the mouse if one does it just a few times, but for things one does much or frequently, the keyboard is much more efficient.
The individual disks/partitions are still fully visible in the Windows environment. Furthermore, every unit is identified by a single letter, and the possibilities to give logical names to them are limited, to put it mildly.
None of these, or a countless number of other deficiencies, are very serious by themselves. But they are almost always to disadvantage of Windows.
I'm not saying tghat everything is better on all Unix systems. As an counterexample we have Microsoft's Mindcraft benchmark. That test showed Windows NT is indeed able to give better performance than Linux in one situation.
The test has been critisised for not being a realistic example on what the load would look like in reality, among other things. The critisism is probably well motivated.
But more interesting to me was to see the reactions after the publication. Within a few weeks a couple of bugs that lowered Linux' performance was found in the system. And fixed! As of this writing, intensive work is going on to remedy the remaining difference, and of course preferably turn the relations around. Compare this to how Microsoft usually responds to remarks on deficiencies, se e.g. Bad service below.
Then of course, you can always put a humorous twist on it instead.
Like many people who dislike Microsoft Windows, (and many who don't) I dislike the company Microsoft itself too. I don't usually go around and tell everybody about it. There is no link to Boycott Micro$oft or Conrad Sanderson's interesting argumentation from my home page.
But you have choosen to read about my opinions, and because of that I also tell you why I dislike the company. And because of those opinions, I also avoid buying it's products. Some find that silly. I see it as the way for me as a consumer to try to work for a better world. Like buying foods marked KRAV to save the environment. Or like some people boycott Coca-Cola because they have anti-union policy in some countries with weak legislation in the area.
Microsoft abuses its monopoly like position in the operating system area to kill competition in other fields. There are many examples. The introduction of tests in early Windows versions to make it impossible to run with DR-DOS, only with Microsoft DOS, is classic. The best known probably is how they try to kill Netscape, since this case is subject to court trial. Maybe Apache will be its next victim?
To avoid misunderstandings, I want to point out that I can not say if Microsoft's actions are illegal. I don't know the law, in particular not the law of USA, well enough to determine that. My point here is that its behaviour is morally bad.
There are many examples how Microsoft is selective with the truth, uses FUD, and sometimes even direct lies, to hit competition.
One example is the announcement (the original, it doesn't read the same any more) of Internet Explorer for Unix which was titled
Microsoft brings the Web to Unix
Consider then that both the Internet and WWW originally was developed on Unix. Consider that Microsoft for a long time tried to fight it, among other things with their own, closed, MSNet. And, as mentioned above, the very reason for the release was to knock out the biggest competitor.
In this context it might be appropriate to give a link to NT lies which describes a number of statements about Windows NT, by many considered true, but not being so.
Some documents intended for internal use, but which was leaked and given the name The Halloween Documents discuss how Microsoft could attack Linux. As before the attitude is unscrupulous, and shows no intention to fight a fair fight. Linux might possibly be a somewhat different opponent than many earlier. Most part of the development is done by people who not primariliy, or maybe not at all, has any economic benefit out of it. But Microsoft will surely still be able to cause a lot of damage, I'm afraid.
As I've understood it, the service level delivered by Microsoft leaves
a lot to be desired. They aren't the only ones, that's certainly
true. The quality of the service seems to be inversely proportional
to the size of the company, and/or how much one pays for
support
.
A very telling example is the response to the discovery of an error in some versions of Intel's Pentium processor. Slightly simplified the processor will hang if it tries to execute an instruction starting with hexadecimal f00f. A more detailed description is available here. The case is interesting as a comparison since it
Intel presented the first recommendations from different suppliers on how to address the problem. I find it interesting to study the different categories. Linux and BSDI says
Here is a fix!
and gives an URL. Most other suppliers aren't as quick, but says in principle
We are working on the problem!
We find Microsoft in the other end of the spectrum. They say that it is true they are also working on it, but add
Blame yourself if you are hit!
The reverse of the above arguments are in principle good reasons to use Linux. A few more follows below.
It is certainly convenient with a window system many times. They show clearly the available choices, can guide you on the right track, and support your work. Provided they are well made, of course. But many of these advantages disappear as one becomes used to a tool or system. It becomes more important it is easy and convenient to do what one wants. Then a keyboard with a hundred keys or so have very much wider bandwidth than a mouse with three or less.
The user interface of an Automatic Tellerm Machine is surely good for its purpose: Anybody can use it correctly the first time. But I would not like to have it as the user interface in my work environment eight hours during a day. The window version is not always better. In Windows you have to use it. In Linux you can make your own choice.
I'm convinced free source code is good, and use as far as possible Open Source systems. They simply give me a better quality. When an error is discivered there is often sombeody who considers it important enlugh to fix it directly. Sometimes this somebody is myself. More often it is somebody else who have fixed the error before I even come across it.
It is not only about bugfixing. It is about trust and security too.
In e.g. cryptography
a method is not considered trustworthy unless it is public.
(Interestingly enough none less than Microsoft appears as an example
in that reference. Guess if it is as a good or bad example. :-)
And if the source code is open, there is no risk of suddenly being
surprised by Microsoft having placed extra spare crypto
keys. Not to mention how they manage when they try to encrypt a
Windows NT password stored under Windows CE. To quote Bruce Schneier in his newsletter Crypto-Gram:
This is so pathetic it's staggering.
An anecdot which could be fun to mention in this context is from my time at SISU. I had the opportunity to send three error reports practically at the same time, one to IBM on their OS/2, one to Sun on Solaris, and one th the responsible for GNU Make. It wasn't the same errors, so this comparison isn't too relevant, but still. IBM stgill hadn't given an answer when I left SISU a year or so later. Sun came back with an error log number a few days later, and gave a reference the error was fixed in a later version of the system soon to appear. (Which it indeed was.) And GNU Make? You may have guessed: When I came back the following morning there was a ready patch in my mailbox. Only to apply and run!