Slashdot Mirror
Designing Linux for the Masses
Kelly McNeill sent in a pointer to this article by Todd Burgess. Here's a quote from it: "The concept of Linux for the Masses is an honorable goal but one that must not be taken too lightly. The purpose of this article is to point out several of the current limitations in Linux and what should be done to create a usable system." You might want to hit the links at the bottom of the article and check out some of Todd's earlier writing on the subject of Linux usability. You may not agree with him, but his opinions are always backed by sound thinking and are well worth reading.
Windows is no better-this whole start menu is awful because you have to deal with 2 different directory representations instead of one
I think one thing thats often overlooked with the Macintosh is that the file system is layed-out so that the user can easily interact with the structure. The MacOS is the the only GUI where you launch programs by actually acting on the excutable file. (The Apple menu is generally only used for small programs such as the Calculator.)
Every other GUI requires you to interact with some shortcut or alias or another alternative representation of your file structure. This adds confusion and just makes the whole GUI that much more fragile.
The MacOS works because the directory structure was layed-out so that users can actually comprehend what is going on. "Netscape Navigator" is in your "Netscape" folder at the root of the hard drive, not in "opt/netscape/communicator/bin" or in "Program Files\Netscape\Program\bin" or whereever. Your hardware drivers are in the "System Folder:Extensions" folder, not in "WINNT\System32\Drivers" and so on. You can actually drag a driver out of the Extensions folder and not break anything.
The UNIX and Windows file systems were designed with system efficency in mind and *not* user efficiency. All of the gee-whiz GUI stuff can't cover this up.
Another key feature in making this all work is the Desktop Database -- Hard coded file paths are NEVER used in the MacOS. For example, A user can actually rename their "Netscape" folder to "My Web Browser", and nothing is going to break. Likewise they can copy the Netscape folder to an entirely different hard drive, and all will still work. (This system applies to Alias links also, so that if the user is using the Apple Menu or another representation, moving a folder won't break the alias.)
The UNIX file system is so established in Linux that it will probably never go away. Still, I'd like to see some of these file system features adapted in a Linux GUI. This stuff is something Microsoft never figured out, and it'd be nice if Linux could beat them to the punch.
--
Business. Numbers. Money. People. Computer World.
The windows system, like the Mac system, is horrible. Having a seperate directory structure for programs and data is the way to go.
It should be noted that modern versions of Windows and MacOS are set up so that the file dialog defaults to a "Documents" folder. (Hard Drive:Documents on the Mac, and \WINNT\Profiles\$USERNAME\Personal on WinNT)
--
Business. Numbers. Money. People. Computer World.
Yeah, the floppy "shadow" was originally a feature back in the days when the system had 1 400K floppy disk and no hard drive. The fact that they system could remember previously mounted floppies made it easier to use multiple programs and avoided "COMMAND.COM not found" errors and the like.
What was a big mistake was to extend this metaphor to things like CD-ROM drives and Zip disks. The MacOS disables the eject button on the front of the drive, forcing you to use the trash can. That's unintitive, and considering all the UI geniuses at Apple, fairly retarded also.
It has been the most cited "Bad Design Item" for what, 12 years?, but no fix yet.
(I should note that Apple has fixed this if you buy a two button mouse. You now can right-click+Eject a removable volume.)
--
Business. Numbers. Money. People. Computer World.
People think the Mac is logical and easy to use because Apple's ad campaigns say it is. Please.
The Win95 shell isn't great, but it's certainly a step above Win3.1. Right-click menus are a good thing. A desktop you can put things on is a good thing. And the taskbar is nice. People complain about "Shutdown" being on the Start menu, but that's just a flaw in the choice of names - if they'd called it the "Main" menu or something, there wouldn't be a problem.
--
What you describe is a workaround for a broken netscape-communicator application. What if you accidentially delete that icon? You might be able to reconstruct it, but would an average user and/or desktop technician?
An intuitive system would do exactly what your script does. For example, Windows' IEXPLORE.EXE does exactly what your Gnome icon does, without the shell voodoo.
--
Business. Numbers. Money. People. Computer World.
I don't know about you all, but to me, this flies in the face of much of the logic about what makes open source great--competition (or "coopetition"), like between GNOME and KDE. So long as the two don't diverge too greatly, I see no problem with having both (or others, if they are well-developed). The key is not consistency, but interoperability--i.e. so long as I can use a GNOME app under KDE and vice versa, I see no real problem.
As for point 2, I agree that at some point as much of the CLI needs to be hidden as possible--with the caveat that it should still be easily available if needed. Linux should remain user-friendly for those "Dilbert T-shirt" types *as well as* for their moms.
cya
Ethelred
Everyone wants to be Ethelred. Even I want to be Ethelred.
The article has a few points, but goes way too far. The only way to configure a computer to structly meet the presented guidelines is to remove the innards, and fill the case in with epoxy (so that the user can't take the 'wrong' path and complicate his life by having innards installed).
After all, if the box will actually do anything, it is possable for the user to make the wrong choice somewhere (insufficently constrained).
Let's face it, you can't stop the user from using the mouse as a hammer. (or the CDROM tray as a built in cup holder)
Improving GUI configuration tools is a must, and appears to be happening slowly but surely. Software installation could be improved by standardising icon location (for GUI apps), and configuring various app launchers to know about the new app.
Under no circumstance is Linux going to become more usable by removing capability and choice. The user will just have to know their own skill level, and choose the interface that matches. This shouldn't be anything new, car buyers manage to know whether or not they can (or want to) drive a manual transmission, and choose appropriatly.
On the one hand, the article sounds like end user advocasy. On the other it manages to be condescending to the same end users by suggesting they should not be allowed to have any features they could hurt themselves with. Safty scissors and crayons for all! (I am reminded of Stoogemania where they are teaching recovering stoogemaniacs the proper use of a hammer)
In many cases the problem is simply that computers are reletivly new, and so the use of them is not yet cultural knowledge (for many adults anyway). At one time, new car owners needed to consult the owner's manual for starting and driving instructions, but everyone knew how to ride and care for a horse. Icons are NOT intuitive, they are just what the masses have managed to learn to use. Take someone who has never seen a computer before, and they probably will have to be told that moving the mouse moves that arrow, and that pressing and releasing the button is called 'clicking'. Then they will have to learn what the icons mean ("I want to draw a picture, but all it offers is a paintbrush. Is there some way I can use the mouse instead?").
Short conclusion: Like every other tool in existance, proper use of Linux will eventually become cultural knowledge and the common person will laugh when they see the 'ease of use' features their great-grandparents couldn't figure out.
Without wanting to make this into a "me too!" thread, I agree 100%. By all means make the OS able to use little words and pretty pictures so that "average" computer users feel at home, but make it "grow" with experience. This is one of the reasons I like AmigaOS in its later incarnations. You can do virtually everything from the GUI, and most native Amiga apps (i.e. not Unix ports) are graphical. However, power users can drop down into a shell and do things that way. The GUI and shell complement each other perfectly. That's the way it should be.
(since being an Amigan seems to be the best way to get yourself ignored/flamed/ridiculed on Slashdot at the moment, feel free to ignore everything I say :-)
But Linux's design actually can make this easier.
/home/ directory. So not only should WordPerfect automatically assume data files are in that directory, but you could also have a "show me all my WordPerfect files" button on the File Open dialog for WordPerfect and it could quickly look through her subdirectories and give her a list of them.
Where would you look for a user's file on Windows (or MacOS for that matter)? Just about anywhere in the file hierarchy. In contrast, a Linux user's file should be somewhere under the (much smaller)
Note that you should probably also have a list of recently changed files for a given user. A file created five minutes ago is more likely to be the one I want to work on now than one I haven't touched in six months. And the WordPerfect file I saved five minutes ago is *extremely* likely to be the one I want when I start up WordPerfect. It could be that easy, without changing fundamentals of Linux.
BTW, why does Linux need to say the system is halted twice?
Ooh, a sarcasm detector. Oh, that's a real useful invention.
My point is that if your EXE is buried six directories deep, and is one of 100 files in that directory, and has a 8.3 file name, who's going to bother.
--
Business. Numbers. Money. People. Computer World.
Spare me.
If and when you graduate from com-sci and enter the real world try watching how people who don't have computer knowledge interact with computers. Or how abould taking a psych course this summer so you can get an understanding of how people think (are you more likely to remember the word dog or a picture of one?).
For you, the CLI snob, typing arcane commands in a very particular order at 45 wpm is a very good way to interact with a computer. For my 60 year old father (a farmer with a grade 10 education and NO typing ability) it is not. He is better off with a GUI. Not any GUI, but a well designed GUI (take an HCI course). A lot of stuff in Windows is terrible GUI design...some of it is very good and intuitive. So Linux should keep the CLI for those who WANT or NEED to use it, but provided a well designed, intuitive GUI for those who WANT or NEED to use that. Why stop there? How about voice interaction with a computer (Sound User Interface)? Or touch/movement (for the paralyzed etc)? How about OCR or other kinds of scanners/readers (brainwaves?) a user could employ to interface with a computer? The possibilities are endless. The CLI is not buried but face it, except for programmers and a few older people, no one uses it as their primary interface to a computer any more. They use a GUI. Not because it is easier to use nescesarily (although a properly designed one would be) but because it's easier to remember how to use and more intuitive to figure out if you've never used one (or even to figure out where you are or what application is presented to you). It give the user confidence that they can figure it out - more confidence than a blinking [user]$ or c:\> prompt would be.
BTW, if all of this "Ease of use bullshit" is a "fucking lie", how come most home OSes (Win/Mac making up well over 95% of that market) are GUI based or driven? If your arguement was true, we wouldn't be having this debate - people would want Linux BECAUSE it was solely CLI (and thus by your arguement easier to use)not DESPITE that. Why are so many people asking for a better Linux GUI? Huh?
I like the CLI. I think GUI is better for most things. Niether is the end all be all of computer interface. Use both of these and a lot more (see above) to make the computer truely easy to use...and drop the elitist, snobby attitude. It makes you seem like a hot blooded fool.
Never by hatred has hatred been appeased, only by kindness - the Buddha
The assumption that programmers don't matter and shouldn't be cattered to ignores how the whole open-source thing works.
Programmers see they have a need, and they write a program to fill that need. Then they share the results with the world.
Under this model, the programmer is not going to say "Hmmmm, I wonder what a 45 year old house wife from Denmark would want this icon to look like." (Unless said programmer is a 45 year old house wife from denmark) They design things the way they want them. There is little motive for them to do otherwise. (Unless you are paying them to write the program with 45 year old house wives from Denmark in mind)
"In the Linux for the masses the command line interface (CLI) is dead. What may of worked for line printers and dumb terminals has been made obsolete by PCs with graphics cards capable of supporting a GUI."
I think i have heard this one before like 100 times, why is everyone so scared with command line interfaces. Why can't there be a GUI and a CLI? You could hide it from the user maybe, but why kill it? CLI's are very powerful and i couldnt ever see myself using an OS without one. Reason being that with a CLI you dont need a customized "wizard" to guide you through setup, etc. Doesnt it seem pretty silly to have a different window pop up for every little task that you want to do in an OS? I think CLI's are the ultimate in customization for any OS, after all you are only limited by your own imagination with what you can do with it, you can combine commands, redirect output/input, pipe output/input. Lets see you do that with a GUI. I think it would get very messy. This guy should just stick to windoze from the sounds of it, or maybe better yet he should buy a calculator...
Sigh. No matter how well thought out and well written this article is, it is still the same self-negating, point-missing argument that should have been put to rest many years ago. If implementation is irrelevant to users then, by definition, choice of operating system is also irrelevant. Computers and operating systems are tools, people. Accountants don't use operating systems - they use spreadsheets and accounting systems. Secretaries don't use operating systems - they use word processors, PBX systems etc. No "average user" can be expected to use the tools they are given without three important things; training, training and training. Users have jobs to do and these jobs generally have nothing to do with computers or operating systems. Users can be trained to perform their jobs efficiently whatever platform, operating system or application is available. For example, someone hands me a screwdriver saying "This is the most advanced screwdriver in existence - it has been painstakingly tested and developed to the be easiest, most consistent screwdriver on the market. Now, go fix my car!" Gee, where do I start? However, train me how to fix a car and I will probably be able to do it with a sturdy knife rather than the super-screwdriver. (BTW, I am not a mechanic and it shows :-) So, my point. All you talented people out there hacking Linux software - keep on doing what you do best. Sure, try to design good user interfaces and help users as much as possible but keep in mind that your software exists to solve problems. Users who need your solution can be trained to use it. (I really need to cut down my coffee intake :-)
Platitudes about user-friendliness, stability of Linux vs BSD, or whatever, are great, but I really miss the technical details. Want to make Linux user-friendly? Ok, let's walk through the list of daemons and other services that a typical distro starts up and decide whether they belong on a desktop OS. Want to compare the TCP/IP stacks of Linux and BSD? Ok, give me some technical details about what is different about them. I don't much care about when and where they were written.
Maybe I'm not going to find this kind of information from Slashdot. Where does one find real technical discussions these days?
The thing is, there are many, *many* more people like your mother out there than there are people who are comfortable with the complexities of linux (as it is today). I realize that there are a lot of people here who believe that there's no reason to change linux to make it more 'user friendly' - that it is, and should be, an OS for the technological 'elite', for lack of a better word. This is a perfectly valid opinion---but it's mutually exclusive with another common (and again, perfectly valid) idea I've seen in the linux community, the 'take over the world' mentality. I do run linux at home, and would classify myself as a linux newbie at best - but I'm still far more knowledgable about computers than at least 95% of my peers - and these are all highly educated, intelligent people, people who wouldn't consider linux even if they knew anything more than the name. Most people simply don't care, as long as they can use computers to get their work done, use the web, or play the occasional computer game. Endless configurability and myriad choices of interfaces just aren't an issue. And I won't even get into the uselessness of having source code available to the average user (I'm not disparaging the concept, or the obvious benefits - just stating that the average user isn't, and never will be able to take advantage of it personally). The needs and desires of the traditional linux user are so different from that of the majority of computer users, and future computer users, that the likeliest common ground I can see is simple binary compatibility. Let the hard core linux user use the CLI, customize to their little heart's desire, get things running _exactly_ perfectly. Give the average user a simple, stable, consistent interface that they can learn quickly (and only once) - and let them use the same applications so that both groups can work and play compatibly. This is assuming that the average user can be convinced that there's a reason to use anything other than windows and office... anyway, my point is this - there simply isn't any way to please both groups, and without the pool of average users, linux won't become mainstream, ever. Linux has the potential flexibility to serve both camps, but there's no reason to even try to make everyone happy with one distribution/interface set.
Let me explain.
I work for a fairly large corporation. We have lots of unix machines for various engineers who do lots of CAD and stuff like that.
The environment, however, looks the same across platforms. Whether you run an Alpha or a Sun, everything is in the same place as your home machine. Likewise, it doesn't matter if gimp is in /usr/bin(DistroA) or /usr/local/bin(DistroB) So long as the Gimp Icon is available from a toolbar or menu.
This is also what we see on windows and Mac installations. People don't precisely know where important system utilities are when they sit down, but they know to look around in the start or apple menu.
Also, being limited to one window manager is silly. We have three supported window managers here: OpenLook, CDE, and Mwm. You have a file in your home directory that your init script uses to decide which one to start up -- and it's the same wm across machines and across platforms. If you want to change it, it's a menu option. Having the same window manager shouldn't be manditory across platforms -- but perhaps there should be a standard list of window managers that should be available. And it should be easy for users to switch to their window manager of choice when they need to.
I certainly agree that for a LinuxForTheMasses distribution that the command line and administration knowledge should not be necessary. I could not live without my command line -- it's the perfect file manager, program launcher, and with vim, word processor. But my mother has enough problems remembering where to double-click, remembering the syntax for various commands isn't something she's interested in.
I think we've started to see some of this in the server market -- the Cobalt Qube is configurable with buttons on the front and web pages -- no knowledge of Unix administration is necessary. And I think that Gnome and KDE are coming up with a good base on which to add the features we need for LFTM. Just give it a few years to mature. It will come.
-- Erich
Slashdot reader since 1997
I'm glad that articles like this are being written; the issue of usability is one that Linux geeks have ignored for too long and are still mostly ignorant about.
An operating system is good if it helps you get a job done with as few problems as possible. Stability and efficiency count, and Linux shines at this. However, usability is just as important, that's why KDE and GNOME have been created, and they both still have a long way to go. There's little point in stability and efficiency if the user can't get to it because they're first required to learn 100 different arcane interfaces and commands.
The author is right with his assertion that interface design shouldn't be left to the programmers of applications, as they tend to design from the inside looking out as opposed to the user's point of view, from the outside looking in. Not only that, but interfaces need to be tested, and this is done by exposure to users unfamiliar with the program. If the users have trouble understanding the interface and can't use the program properly, 99% of the time it's the fault of the interface, not the users.
There's a horrible attitude that's quite pervasive in the threads above and it's one of pointless elitism: that Linux should be for geeks only, if you make it easier to use then you get more and stupider users and you lose configurability and the ability to do the complex things you can do now. And it's all utter crap:
a) If you really want to use an OS that hardly anyone else is using, there are loads out there: Plan 9, RiscOS, OS/2 etc. They all have lots of good points. Just don't expect much of an application or support base and don't expect much progress. If you want to be part of a minority, you pay the price.
b) Everyone reading this had to learn Linux at some point. Would you rather spend more or less time learning how to do something? (Personally, I'd rather spend less time learning and more time doing)
c) Unix has already progressed in terms of usability from when it started; there are applications in common use that greatly simplify necessary Unix tasks, and they purely exist because of usability needs. If you don't think usability is a major issue, try replacing your favourite text editor with ed or pico. It's just as powerful, but it's a hell of a lot less usable. The fact is that most of the "we don't need usability" idiots depend far more on usability improvements than they think.
d) Decent computing power should be available to everyone. We believe that Linux has that power. If it's held back by bad usability, there's no point railing against Microsoft, because we're not providing a usable alternative. (And do you want the less tech-able of your family using Microsoft forever? I bet you get pretty sick of the support calls...)
I disagree with the author's assertion that all feedback from programmers should be rejected; programmers are users too, and you're not going to get feedback on a C++ IDE's usability from your average secretary. All feedback should be counted and considered.
However, the author makes a very good point about hardware - why the hell does the average need to know what hardware is in the machine? Why do I need to mount drives manually or know which graphics card I have? This is all stuff that the computer should be detecting and taking care of for me. It just gets in the way.
Usability is vital if Linux is to prosper. Fortunately, there are more and more projects happening that will contribute to Linux's usability and friendliness. Let's all assist where we can and ditch the childish elitism.
Let me tell you a story. I was recently at a pretty young woman's house, um, helping her with her computer which she'd managed to foul up again. While I was there, I decided to reinstall some drivers. So I opened her CD-ROM drive and what should I find but the Internet Startup Kit that her ISP had given her. No big deal, but she was terrified. "Don't I need to play that CD to use the Internet?"
Yeah yeah, she's not stupid, just ignorant. But this is the type of user most people are. They know as little about computers as I know about the inner workings of an internal combustion engine.
You want a user-friendly computer? Make a Palm Pilot with a big screen. Press a button and you're in 'email' mode. Press another button and you're in 'web' mode. Anything more than the minimum function set necessary to do the job at hand (e.g. email/web) is wasted on the average user.
Believe me, I deal with average users every day. They may not be stupid, but they definitely do not have the time or inclination to learn how to use these admittedly esoteric gadgets. Today's desktop interfaces are an expression of contempt towards the end user.
It's a well structured, thoughtful article.
Many of Todd's points would be true if Linux was a proprietary OS, trying to break into the market, like BeOS.
But it isn't. Linux is a Vast Internet Thing that doesn't care anymore about competition, or markets. If MS disappeared tomorrow it wouldn't make any difference (although IBM might actually admit they make OS/2 Warp 4)
As for the flexibility thing - Linux's flexibility is the reason it is where it is.
#ifdef OFFTOPIC
Rule 1 of HCI. Microsoft Windows and its applications are, generally, how NOT to do it. Windows lost the HCI plot when it moved to the 95 explorer shell. Windows 3 was nice, everything was consistent, users liked it. The 95 shell is broken from a UI point of view. It's inconsistent, illogical and confusing.
If you want an example of a really easy interface, you still have to go to the Mac.
#endif
Linux is easy to use *now*. Installation doesn't count. The people who would barf on a Linux install would equally barf on a Windows install.
Compare like with like. Windows-only boxen should *only* be compared with Linux-only boxen.
There'll always be a place for character displays (ask the accounts department and goods inward) too.
Command lines are useful as well. AutoCAD 2000 keeps the command line and for good reason - it's often the fastest way to do what you want.
Still, a good article and food for thought.
Peter.
Peter
First let me say that I agree wholeheartedly with the author's underlying goal of creating a "system your mother could use". I do, however, have a lot of concerns about the approach being advocated.
My most important concern is that, while the author seems to know there are many different groups of user, he proceeds to outline a scheme that caters only to those who are most technically naive.
He seems to be advocating eliminating all the diversity and choice than exists in Linux at the moment, in order to simplify the lives of the one group that doesn't use Linux, hasn't heard of Linux, and doesn't care what OS they use. In the process following his scheme (getting rid of all but one distribution, all but one window manager, all but one toolkit, and every single command line shell) everone else, all the people currently using Linux, all the people who actually care and keep the community alive, would be alienated. Even if it were possible to do this, which obviously it isn't, would it really be worth it ?
My point it this: linux is not designed poorly - it is designed for technical people, and only technical people are going to care enough to keep it alive.
I have no problem at all with a "Linux for the Masses" like the author proposes existing (although I don't really see who wants a warmed over Windows clone). Its the idea that all other Linux systems must be sacrificed and the entire existing user community alienated, to attract the people who care least , that I find disturbing.
On a somewhat more minor point, 'usability' means a lot more than UI design, and is not that easy to disentangle from implementation issues. Some of the worst systems I've seem had their UI developed separately by a 'user centered' individual who unfortunately had not understanding of the application domain. Users are not all the same - probably the only person who can do UI design well is someone who understands the application from the user's perspective, not someone who just thinks they understand 'users'. Oddly enough, that person is sometimes the programmer.
Now for some pickiness. Windows is really a pretty poor example of how to do UI design. Most Mac applications, and lots of NeXT ones are much better. The point of window managers is to manage windows, not to read mail, therefore to demonstarte a window manger you show a picture of lots of windows. Seems kind of obvious - even the author's typical user (who seems much stupider than anyone I've ever met, if he gets information overload from looking at 10 pictures on the same screen) should be able to deal with that.
To me, the first half of the article was meaningless nonsense. The second half argues, in essence, that Windows is the current "OS for the Masses",a fact that retroactively validates the Windows philosophy. And since the philosophy is valid -so the argument goes- Linux must buy into it in order to compete.
IMHO, this "Linux for the Masses" argument boils down to copy-catting the philosophy of an operating environment that assumes the users are dumb, and produces more dumb users as a consequence.
One important distinction: what you are saying is certainly true of a multi-user, shared system like Unix or Win NT. However, for Win95 or the Mac, where there's usually only one user, it is a different story.
/temp?
Both the Mac and Windows have a fundamental flaw that affects single and multi-user: they make the physical drive (partition) structure visible. On Windows, you see C:, D:, etc. On the Mac, you see named drives, slightly more elegant and immune to letter changes as drives are added, but still the same basic idea. How many hard drives does a Linux system have? It's not nearly so important to the user -- and for non-removable drives, there's no reason for it to be.
I've never seen a well-organized Mac or Windows system. Windows newly-added user documents placement puts them as a subdirectory of the system -- making it all the harder to back Windows up separately from the user data. And Windows apps generally ignore it, other than some of the latest ones from Microsoft. And the Mac advocates here have advocated *both* putting all apps under an Applications directory (presumably one per drive, btw), and not doing so in order to shorten the paths to find them.
The Linux file permissions mean most users *can't* put stuff everywhere and anywhere. Unless you have root, where are you going to put your stuff,
Ooh, a sarcasm detector. Oh, that's a real useful invention.
There's a horrible attitude that's quite pervasive in the threads above and it's one of pointless elitism: that Linux should be for geeks only, if you make it easier to use then you get more and stupider users and you lose configurability and the ability to do the complex things you can do now.
I haven't actually seen much elitism on this thread. There is a serious problem with the approach the author is advocating though, and perhaps you are mistaking this for elitism. The point is that Linux has succeeded because it is a good operating system for geeks. If we do what the author wants, and ditch the open source thing, and all the choice in toolkits/wms/shells etc., to pander to the lowest possible common denominator (and let me point out here that I have a higher opinion of the competence of most users than the author does) we will lose the very people who keep Linux going. The people who actually care what OS they use.
This is not only a reason why the approach being advocated is a bad idea. Its also exactly the reason why it won't happen. Linux developers are going to keep developing things they like, not things for 'average users'
--
I think that post should be moderated down. Not because I disagree with the statement, but because of the way he/she chose to articulate his/her opinion. f*ck is not really proper here.
But you talk as though Linux is a corporation with your
until Linux brings in more money and then free-software developers will become history once they can afford to hire programmers and turn everything into propriatory
statement. It sounds as though there's some mysterious Linux Industry out there feeding off of the Free Software Foundation, and as soon as it gets its money, it will hire programmers and through away FSF.
1) Linux is under GNU GPL. GNU GPL is basically FSF. Can't become propriatory(sic).
2) Linux is an OS. As long as people are out there maintaining it (as Free), it will not be affected by any corporation.
It is the ones who are tired of programming in Windows that wants Linux as a Corporate OS. That way we can start enjoying our jobs because we can control it, and not worry about someone else's closed source bugs.
-- Most open source patches are workarounds for closed source bugs --
Steven Rostedt
-- Nevermind
(...well, not literally, you know what I mean)
This is a good article, but I think that it represents a throwback to the 'bad old days' when operating systems competed head to head (amiga vs mac vs Win vs OS/2 vs...). The users of one OS looked at another and said, "we like parts of that, but we'd really rather other parts of it look like this." In this case, Windows users crave both the standardization of Windows with the stability of Linux. The proposed solution? Try to shame the Linux developers into making a 'standard' Linux.
This will hopefully never happen. 'Linux For EveryUser' would not be a linux I would use. Say what you want about them, but CLI commands are usually smaller, lighter, faster, more flexible, can be more easilly run in batch (when you need to do the same thing or similar things on large groups of input) and can be more easilly run remotely (through telnet) than virtually _any_ GUI app.
I also reject the argument that multiple distributions are inheirantly bad. Multiple distributions means that if I have different needs, I can select a different distro that meets those needs better. Microsoft has brilliantly demonstrated that the "good enough for most users" solution isn't "best" for virtually anything.
Similarly, the multiple UI argument is both bogus and already satisfied. Firstly, I want to be able to select my own interface. (I'm a sick puppy, I _like_ olvwm -- I don't like all the bells and whistles of E or afterStep, and CDE -er- KDE leaves me cold.) But that is my choice. Locking me into an interface that 'you' have decided is 'standard' (which usually means, it meets _your_ needs adequately) does nothing for my support of the system and little for my productivity.
Secondly, linux already has a standard user interface that is common across all distributions: the dreaded CLI. I can sit on a RedHat or Slackware or SuSE installation, type 'ls' or 'cd' or 'vi', 'find' or 'awk' or 'perl', 'ps' or 'kill' or 'cat', and get the expected result every single time. More to the point, I can use 90% of those same commands on virtually any Sun, HP-UX, OSF (er sorry Tru64), AIX, Irix, or BSD machine and get the expected results! And best of all, I can do it through telnet (or rsh in closed shops) from the comfort of the other side of the office/building/planet! Try running even gmc across an internet VPN.
The point is that choice is good. You can select the right tool for the job. When the only tool you have is a hammer, every problem tends to look like a nail. I guess you could similarly say "when all you have is Microsoft Windows, every problem tends to look like either a labour-intensive impossibility or a GPF waiting to happen."
:)
Some things are not simple. Nor can they be made simple. Sometimes you need clever people to do difficult things.
--
you should read everything on the internet as if it had "but I'm probably talking out of my ass" appended to it.
There's a common misunderstanding that making software easy-to-use means making it less powerful. Many people talk about having to "dumb-down" linux in order to make it usable by the masses. This simply isn't true.
In the HCI field, there's a strategy for designing user interfaces known as "progressive disclosure." This design strategy presents a simple and limited set of options and actions to the user in the first "layer" of the user interface. As the user becomes more familiar and comfortable with the software, more features and options reveal themselves (possibly including a command-line). A really well-designed UI can support both the novice user and the expert. Progressive disclosure works much like a good teacher does; the basics first, then the intermediate to hard stuff when the student's ready to handle it. Designing UIs isn't about making pretty icons; it's about constructing a meaningful dialogue between the computer and the user.
If we truly want to raise the technological intelligence of the world population, then I believe this is the best approach. On the other hand, if we want to keep powerful technology in the hands of the techno-elite, then we can continue to toss new users into a CLI ocean of device drivers and kernal patches while smugly watching them drown.
Dustin Beltram
usrdzign@netscape.net
...people, for the (Dilbert T-shirt, jean wearing)
people.
First, make it work, then make it right, then make it fast, then, make it bloated!
I wondered why we hadn't gone up to v3 already...
-- This and all my posts are in the public domain. I am a lawyer. I am not your lawyer, and this is not legal advice.
Designing User Interfaces, be it graphical or not, is not easy, and certainly not technical. The design process itself should not involve technical problem setting at all, which is what most people designing for the various OSS projects are doing. The UI implementation phase is different, but that should come only after designing the UI.
UI design is a world of it own, comprising of processes and thinking models that most people are not very familiar with. Good UI designers are good with people socially, as they're good at figuring out the way people process information in their heads. Some people can't ever become good UI designers because they are too closed to various sources for ideas!
One of the problems I see in the GUI projects related to Linux is that in order to design a good UI novadays, you have to look at Windows users to figure out behaviour patters that people have established and design those in mind, as most people do use Windows. Anything they're not familiar with, ie, not Windows-alike, will make using the UI harder for them. This doesn't mean everything should look like Windows, but you can't go too far from it without losing usability either.
So anybody who absolutely hates Windows (or a Mac) to the point of not being able see it as a viable platform for UI design ideas is never going to be able to make a very good UI. I hope people will relax in this sense a little more in the future..
"Although it is not true that all conservatives are stupid, it is true that most stupid people are conservative."
Please don't embarrass us again. I'm literally terrified that this poor guy
will get thousands of nasty e-mails from a bunch of techno-bigots (his term).
He's absolutely correct in every way. I fell into a trap a few weeks ago,
where I was praising Linux, up and down, to everyone. One of those people was
my mother.
She asked me if I would install Linux on her computer, because she's tired of
windows crashing all the time. She's also tired of "gackling:" a term she
uses to describe windows poor memory management, which results in that funny
hard drive sound. Much to my surprise, I told her that she wouldn't like
Linux. This, after I had spent so many hours praising Linux to my little
masses. The fact is, although she is extremely intelligent, she probably
couldn't make Linux go. All she wants to do is send e-mail, surf the web,
make posters, scan things, and make business cards. These are all things I can
do in Linux. I'm not convinced she could. At least, not without a
never-ending stream of panicky phone calls.
I talked my roommate into running Linux as well. He informed me last month,
that if it wasn't for having me around to ask questions, there isn't a chance
in hell that he'd run Linux. See, unless you want to spend several hundred
hours reading HOWTOs, Linux just won't work when you need to get things done.
Currently, my productivity is well over 10x my windows productivity. This may
be because I love Linux, and am able to do almost anything in it. The poor
interfaces don't bother me at all, because I can just design my own from the
things I d/l off the net. Could my roommate? No. He uses fvwm. *shrug* His
machine... Could my mother? I'd have to set up gnome for her or something.
She still couldn't get the scanner to work.
Linux needs a mom's interface before I would ever dare install it on her
machine. Linux does not need to become windows. It only needs a simple and
consistent window manager. Then I could install it for her and she could
happily run her programs without ever rebooting. I'd still have to install
because of the install process. It seems simple to me now. I could do it in
30 minutes if that machine was fast enough. To my roommate though, it's about
200 very difficult questions.
Jet (jettero@.nospam.yakko.cs.wmich.edu)
I often have to provide tech support to Windows 95/98 users. Describing icons and menus to the remote user is a distraction and a time-waster. I'd estimate that 95% of the time, the first thing I have the user do is open a DOS box so I can tell him exactly what to do to get his problem fixed ("Now, enter D-I-R, that's delta india romeo...now press the spacebar..."). Isolating and fixing the problem often takes a handful of commands and a little reading on the part of the user ("OK, it says 'volume in drive C has no label...'") but this saves me and the user a lot of time and frustration in the long run.
It would be expensive and frustrating (a nightmare?) to try to support an OS over the phone without the ability to dictate concise commands that allow no room for interpretation, and usually return only the information required. Any Linux distributor who tried to eliminate the CLI in order to produce a "distro for the braindead," would either be forced to charge unrealistic prices in order to cover the additional support costs, or go broke.
slashdot broke my sig
I agree with many of the points of the article, but disagree with some others. As other posts have pointed out, the author's previous articles take an opposite viewpoint from this one, so perhaps he is just painting the two extremes that the Linux community has available.
Users will never need to know Unix to use Linux for the masses. I agree that it is important that a casual user not have to understand Unix to use a Linux system. There has to be at least one bulletproof point'n'click interface which maps fairly closely to the established GUIs that new users will be familiar with: Win 95/98 and/or Mac. Note that this is for the casual user - a CLI should still be available for the user if they want to explore more of the capabilities of their system, or if a more advanced user needs to use the system for something.
Common system tasks will be automated. Automating common system tasks is a good idea too - the problem is that a network OS has a lot of system tasks that have to be running for 'net connectivity, and knowing which ones to run is non-trivial. People criticize Redhat for setting up too many daemons to run at installation. In a certain sense, they have automated these system tasks, so that the user doesn't have to set up inetd manually. The tradeoff is that this may make the box less secure. I'm not sure how much of the administration you can safely take out of the user's hands without making dangerous assumptions about their planned use of the system. Of course, the casual user probably won't be installing Linux on their own anyway, so maybe this isn't as much of a problem. As far as common system tasks like mounting a floppy, etc., these should be automated from the point of view of the casual user.
No multiple distributions or window managers. I don't know about this one. I think it would be safe to say that within one large installation (corporate office, university computer lab, etc.) the distro and WM choice should be consistent, so that users don't have a learning curve to use each other's machines. I don't see why we have to restrict choice when providing Linux to the individual consumers, though. Right now home users are running DOS, Win 3.1, 95, and 98, and MacOS. They already have a learning curve to use each other's machines. What will probably happen is that most Linux users will end up installing Redhat and using the default WM, and that will end up being the consistent look and feel for casual Linux users. I don't think it's necessary to remove the choice of other distros or WMs for this to happen. I politely disagree that flexibility is a fault; forcing too much flexibility on a casual user is a problem, but not providing flexibility for the advanced user is a bigger mistake.
My biggest complaint with this article was this: it isn't a mistake for the window manager screenshots to show multiple applications running at once - that's the whole point of a WM, isn't it? If the user is only supposed to run one thing at a time, then they might as well be running DOS. Sure, most users still only run one or two things at a time, but that's exactly the reason that WM screenshots should show many things running. A new user can look at that and realize that their computer is more powerful than they had expected, even if they don't recognize everything that is running. This is a powerful feature even for casual users, not something that should remain the province of Linux gurus.
Your right to not believe: Americans United for Separation of Church and
A GUI is good for simplifying common, well-defined tasks. If you need to do anything out of the ordinary that the GUI designers didn't think of, then forget it.
In short, a GUI makes easy things easy, and hard things impossible. You never want to totally throw it away.
Of all the comments I've ever posted, this is definately one of them
What people are objecting to is the "one size fits all" solution advocated by the author. He's saying that to make Linux easy to use, we have to dump all the diversity power users love.
No matter what he says, it's not going to happen; in order to make it so, there would have to be only one distribution, only one window manager. Since we don't have a secret police to send Enlightenment, Gnome and AfterStep users into the gulag, well, we'll always have choice. Sorry.
D
----