| Greg's thoughts on the Microsoft Space |
| Greg's rants |
| Greg's home page |
| Greg's diary |
| Greg's photos |
| Greg's links |
|
|
|
|
Other things: reboot at the least provocation.
I've been using the term “Microsoft Space” for a while now without really defining it—even to myself, probably. It's about time I did.
First, I'm talking about the way humans interface to computers. I've never meant the term to mean only Microsoft—thus the word “Space”. But Microsoft has built up an approach to interfacing with computers that is—unfortunately—creating a lot of mind share, notably from Apple and large parts of the Linux space. How should I define it? Here's a first cut.
The design is apparently intended to make life “easier” for people for whom it makes life easier—the so-called “user friendly” approach. From my point of view, it seems that it does this by hiding details that the designers of the software considered contentious. For example, it seems that a majority of people don't want to see the boot messages produced when a computer starts up—instead, they're shown a splash screen.
Other things that apparently worry people—at least software designers—are things like file names and directories. I believe that this is behind the stupid renaming of “directory” to “folder”, and the truncation of file names, replacing the “extension” (in itself a silly idea) with some kind of image that makes it even more difficult to guess the purpose of a file.
It also seems that the designers of Microsoft-style software are overwhelmed by the concept of directory trees. They do everything to hide this structure, leaving people like me completely baffled when I try to find a file. Instead, you see countless cases of utter stupidity when programs go looking for files in inappropriate directories, programs that recall the last directory they accessed and require you to step through a directory tree to change to another one—even the current working directory, which exists in Linux and Apple, but not, I think, in Microsoft—possibly because they don't understand what use it is.
The approach is clearly aimed at people with limited typing skills. This is a valid thing to do: most people can't type as well as they can interface by other means, but that's very definitely a limitation. By all means make life easier for people with limited typing skills, but not by making it more difficult for people who can type. I think that typing has stood the test of time when it comes to an accurate interface. People still write things down—maybe by hand—when they want to formulate something accurately.
The alternative concept of voice input is still with us, but I don't see as many claims being made for it now as there were 10 or 20 years ago. The Wikipedia page on that subject gives a number of reasons why not. Anybody who has tried dictating any lengthy document to a stenographer will understand why: it's very difficult to remember what you said, and when you want to change anything, it's difficult to explain. And that's with an intelligent being; with a computer it would be much worse. I suspect that a combination of keyboard and voice input might be useful, but I haven't seen many developments in that area. And every time I have to talk to a voice interface on the phone, it drives me mad. Australian phone companies insist on demonstrating their lack of usefulness, infuriating their customers in the process.
The chosen alternative in the Microsoft Space is the mouse. Mice are good for some things, notably choosing from a small, fixed number of alternatives or for manipulating graphic data like photos. But it's very difficult to use when you have multiple level menus; I discussed this issue years ago with my paper “Why I hate Open Office”.
Selecting things with a mouse only works if there are relatively few choices. Also in my “Why I hate Open Office” paper, I noted that I had about 1800 programs on my laptop. The Microsoft Space approach to running programs is to click on their icon; but there's no way you can display 1800 icons on a screen. Even the 30 odd that I have on my Microsoft box seem to be deliberately ordered in a way that doesn't conform with my wishes, and the pictures mean nothing. With a two-level menu you can probably get a couple of hundred programs, but no more. So this approach has a limiting influence on how many programs you're likely to use. Apple is even more restrictive with its program start menu: a single line, shared with other icons whose purpose I keep forgetting.
On all systems there's a workaround: bring up a little window and type in the name. That's so useful that you could leave it up there all the time—in fact, make it bigger so you can type other things in there as well, and maybe get output in the same window. But wait, hasn't that been done already? Why isn't it used in these environments?
The lack of support for text input also means that there are no good text editors in the Microsoft Space. Most programs offer their own WYSYWIG text manipulation, all slightly different, all almost completely lacking any but the most basic editing concepts. This is, I suspect, one of the reasons why almost all people in the Microsoft Space can't use email effectively.
The “don't show any more than necessary” attitude doesn't just apply to boot messages and file names: windows are made too small, names are routinely truncated, and things are omitted if the program doesn't like them. A good example is Apple's “Finder”, which opens with a window about half the width and breadth of the screen, along with a scroll bar. If you manually mount an NFS file system on one of the mount points in a “Finder” window, it will obligingly lose the icon, moving it to the top left pane, where it will be given an unqualified name possibly matching one already there.
Window sizes aren't the only thing that programmers set according to incorrect concepts; text size is another. It seems that most Microsoft Space programs confuse pixels (the dots which make up an image on the screen) and points (a typesetter's unit, approximately 1/72" or 0.35 mm). They don't do this for printing, but for monitor displays it means that it's very difficult to change a monitor resolution without changing the size of the text on the screen. Strangely, Apple seems to be very bad about this. I consider this to be the main reason why display resolution doesn't conform to Moore's Law.
All GUI environments offer multiple windows. You normally move between them with the mouse—a good approach in this case. But in most cases you must then click on the window you want, and it raises the window, whether you want it to or not. That drives me mad, and it's completely unnecessary. I'm pretty sure that the Linux imitations can at least be configured to stop this ridiculous behaviour, but I think they usually have it enabled in default installations.
Why? In the early 1990s X did this too, and there was a good reason: changing focus on mouse movements took up a lot of CPU time. But modern processors do conform to Moore's Law, and the CPU time is no longer an issue.
Error messages scare people. So why show them when you can print out a friendly error message, preferably one that ignores the original issue and bases on guesswork? I'm reminded of the fortune cookie:
A reminder: Ken was one of the original authors of UNIX. How times have changed!
The environment doesn't really understand networking. Microsoft goes to the extent of separating local and global networking, and referring to the World-Wide Web as “Internet”.
The lack of understanding of networking means that there are few opportunities to spread processing around a network. Both Microsoft and Apple have “remote desktop” access, where an entire display (“Desktop”) is relocated elsewhere. But they don't really address the issue of displaying individual windows on a different screen. And many X clients are castrated to do a similar thing: firefox will only allow a single process to display on any single X display. If you try to start a new one, it will hand off the request to the one already running—even if it's running on a different machine! This is a particularly annoying thing in my environment, where I'm trying compare the behaviour of browsers running on different machines. This “there can only be one” approach has other annoying issues as well.
Incomplete mumblings from here on
| Greg's home page | Greg's diary | Greg's photos | Copyright |