Slashdot Mirror
How To Fix the Poor Usability of Free Software
flosofl writes "Matthew Paul Thomas has an entry on his blog called Why Free Software Has Poor Usability, And How To Improve It. While this advice is helpful and may indeed lead to improvements in many open source programs, the guidelines may be much more difficult for smaller projects. From the entry, 'Free Software has a long and healthy tradition of "show me the code." But when someone points out a usability issue, this tradition turns into "patches welcome," which is unhelpful since most designers aren't programmers. And it's not obvious how else usability specialists should help out.'" Thomas has been developing the ideas in this essay for years. The critique is comprehensive, listing 15 challenges in the way software projects, and in particular free software projects, are structured, with suggestions for improving each one.
UNIX programmers (the majority of FOSS developers) design software the way that they would want it to be used. It makes sense to other programmers and it is actually probably the simplest way to operate if you have the base knowledge. Users, on the other hand, focus less on the architecture of the software they are using than on the front end. Windows and Mac OS X systems and applications are easy to use because the front end has been designed to meet all usual purposes, even if it cuts back on the functionality. Linux and most UNIX systems and applications are harder to use because they are built with the architecture of the code in mind. A good UNIX program can easily work with other UNIX programs, and a good UNIX program is made as general as possible to maximize speed and reduce bloat as the program advances. A 'good' Windows program is made for only one cycle, not the entire development lifetime. Firefox is a good example of a program that meets both the UNIX and the Windows definitions of a good program. But Firefox is very rare, and there have been multiple revolts over de-generalizing the code for a single release.
I think application programmers should keep the Firefox success in mind when they develop code, even though it will be much more expensive and time consuming than the UNIX mentality since they will have to keep stopping what they are doing to release and polish versions for users (essentially dead forking every couple of months).
I'm going to take issue with this. Basically you're saying if you can't do X, then your critique is useless. Just because you can't sing doesn't mean you can't critique other people's singing. Just because you don't know how to make a car doesn't mean you can't critique that horrible dashboard layout.
Okay, I know these are unpopular things to say, but I feel they need saying. These are just my opinions.
(1) "Usability" is in the mind of the user. If you learned how to use some other system first and now expect that any other way of doing
things isn't "usable" enough, that's just plain old resistance to change. It says more about you than it does about the usability of the software in question.
It seems you've never used emacs, little one.
And no, it doesn't say more about me than the software if I find that almost all GUIs in Linux are shit and that the command line is more usable.
(2) "Designers" who can't code have absolutely no business "working" in software. If you think you really know how an interface should work and look, then learn to code it. Otherwise, you're just a critic of the kind that the NYT doesn't hire.
"Right."
And neither have "coders" who can't "design" something the "average" "user" "thinks" is "usable" .
Bull crap. An architect doesn't build the building, the carpenter does. An electrical engineer working for the power company doesn't hang the lines, but they layout and design them.
The same thing can go for almost any profession, software included. I don't code and no desire to code (I'm a chemist and have no interest in that part of computing), but I can tell you when the interface sucks and what could be better.
My opinion is, people who won't accept criticism from others because "you don't code," or "you don't know what you're talking about" are nothing but elitist.
Gone!
At the bottom the article links to John Gruber's "Ronco Spray-on Usability" article, which also provides a lot of background on the challenges of good interface design.
In the original article, I think the most important point is number 8 - "Scratching their own itch." I can see how programmers interested in, for example, having a stable and scalable web server would work on Apache. I don't see the same passion coming from a human interface designer to fix, for example, the horrible user interface for joining wireless networks on desktop linux.
In my opinion the only way the user interface will get fixed is if Ubuntu or another distro pays for expert user interface folks to fix UI issues. I don't see the volunteer community being up to the task.
- "When you want something with all your heart, the entire universe conspires to give it to you" -Paulo Coelho
(2) "Designers" who can't code have absolutely no business "working" in software. If you think you really know how an interface should work and look, then learn to code it. Otherwise, you're just a critic of the kind that the NYT doesn't hire.
Why should software be different from any other industry? Just because the barrier to entry is, in your opinion, lower? The people who design houses don't build them. Sure, they need a solid foundation in what can and can't be done, but architects and interior designers work to a mutually beneficial end with builders, carpenters and plumbers, etc.
(1) "Usability" is in the mind of the user. If you learned how to use some other system first and now expect that any other way of doing
things isn't "usable" enough, that's just plain old resistance to change. It says more about you than it does about the usability of the
software in question.
You are a good example of the "works for me" mentality in the open source software community. "If it works for me, then it must work for anybody else, and if it doesn't, it's their fault".
(2) "Designers" who can't code have absolutely no business "working" in software. If you think you really know how an interface should
work and look, then learn to code it. Otherwise, you're just a critic of the kind that the NYT doesn't hire.
I'm sorry, what? Did you just say that if designers can't code, they shouldn't contribute? I thought this open source stuff was about the fact everybody could contribute to the program if they wanted, but apparently I thought wrong. Apparently only uber-geeks with a masterful command programming knowledge can work to improve open source software.
Design is important. Not everybody wants to learn the idiosyncrasies of software; they want it to "just work". The moment everybody thinks that way is the moment open source software gets better.
That article hit it on the head. There's 1001 programmers in the world, who are excellent coders and whip through the strings like a first chair, but there's very few project designers in freeware. They concentrate so much of function (which, yes, is critical!), but forget about ergonomics and userability (especially *how* end users can and will use their product, and ways to cut out excessive keystrokes or right clicks). The end users winds up getting a proggie that can function well, but such a chore to operate (or even painful, if it not ergonomically friendly). As we more and more get "connected" to computing, it's no longer just being on a keyboard or using a mouse an hour or two a day. Now it's more like 8+ hrs. Programmers need to consider the impact of their software, and beyond how it functions itself, but the whole project. That's where product design is so crucial, and something not just best left up to management to figure.
(2) People who can't cook have no business eating food and complaining when it makes them puke.
Why does my post history abruptly stop? I want to laugh at the stupid things I posted as a kid.
You seem to confuse usable with usability. Just because something can be used, doesn't mean it is the best option. By your logic we would all be using rocks to pound nails because, "It works, you just have to think a little bit." Being intuitive means that people want to use it that is the point. Unless of course you don't want people to use your product and if that is the case, why are you publishing it publicly?
Your second point is rather funny to me. I wonder if you would say, "If someone doesn't know how to design an engine they have no business designing the rest of the car(oblig. car reference)?" People have specialities, that is how society works now.
Finally a red herring is irrelevant, we are talking about Linux not windows. I don't really like windows either but that is another discussion.
1) "Usability" is in the mind of the user.
There's some degree of truth in that, but since most people have essentially similar minds it makes sense to assume that certain types of operation make sense to the vast majority of people. Past experience can lock people into a certain conceptual mindset, there's no denying it. But the holy grail of usability is that which most closely reflects the way in which we have evolved to think.
All computers are essentially tools to enable us to do things we otherwise would not be capable of, or would take much longer without. The best computer systems are the ones which interface as organically as possible with that which we are capable of. It's common sense.
(2) "Designers" who can't code have absolutely no business "working" in software.
That's ridiculous. If it weren't so absurd it would be offensive. Look at the credits for any major software release from the commercial realm. Let's take games, for example - in the entertainment industry, where if something is not 'usable' it is not fun, coders are outnumbered somthing like 20 to 1 by other types of developer, be they writers, artists, testers, etc. Hell, let's take the credits from GTA IV, the video game. They take HALF AN HOUR to scroll, when you complete the game. What percentage of those HUNDREDS of people do you think can code? Even to a reasonable level? Five to fifteen percent at the outside.
When you need a job doing, you hire the person with the skills to do that job. You DON'T hire someone with the skills to do that job AND the skills to code. Sometimes that's not even desirable, and at best it's overkill.
What? An Apple is more usable than Windows or Linux. BeOS was superior to Windows and Linux. AmigaOS was superior to Windows and Linux, as well.
That's your opinion. I personally find Mac OS harder to use than either Windows or Linux, simply because of what I'm used to. Which is exactly the point of the grandparent post.
Make the GUI simple to navigate so a 3 year old can handle it and powerful enough so a geek can enjoy it.
Good luck with that. It often happens that if something is easy for beginners, it's not so convenient in the long run for experienced users. For example training wheels in bicycles. A good UI lets you remove the extra wheels and tinker under the hood, though.
Escher was the first MC and Giger invented the HR department.
Humans have overall tendencies in what they find easier and harder to use. For example most people find it much harder to memorize lines of commands than to look through a set of icons, hence one of the reasons why GUIs are popular. There's real empirical research that has been done on this, some by psychologists, some by companies like Apple. While you may fall outside the norms for various reasons, that doesn't mean they don't exist.
You also need to be careful in thinking that because you've taken the time to learn something, that means that it is easy. For example I drive a manual transmission vehicle. I have no problems doing so, it is second nature to me, no more problematic than an automatic. However, I am not going to say it is as easy. I remember a rather frustrating learning period when I was getting used to the feel of a clutch and how to shift. It is easy now, but only because I have a lot of experience. It is NOT an easier way of doing things.
I find this kind of thing fairly prevalent in the OSS world. You'll need to do something and someone will say "It's easy!" and then give a couple commands to run. To them, it does seem easy, because they've trained themselves on it, however that isn't. To a new user that's very hard and uninviting.
Ignoring this issue doesn't help free software. Telling people "It is just as easy you just have to learn it," when it isn't doesn't help, they'll just ignore you. The author is trying to say that OSS people need to stop with this attitude that it is just as usable, or that usability doesn't matter to popularity. If you want OSS to start to become the dominant way of doing software, you have to make it as easy to use for the non-technical masses as possible.
I assume he was forced to respond to them, because otherwise he wouldn't have been paid.
Those using pirated Tinysoft signatures(TM) are a real threat to society and should all be thrown in jail.
There seems to be a whole movement who's against the "patches welcome" statement. I fail to understand this.
I'm an open source developer. Look at it from my point of view. I've written software that people find useful. It's not perfect, but it's useful. Then, one day, someone criticizes my software:
Person X: [...] this and this sucks [...]
Me: patches are welcome
Person X: what? what an unhelpful response! no wonder open source sucks, and you suck too!
Now, tell me. I have a job. I maintain this software in my free time. Why should I devote that time to you, for free, instead of, say, hanging out with friends or seeing a movie? You're not paying me for this software. You probably would go away if I ask you to hire me. What exactly do I owe you? I already made the source code available. Why do you criticize me for not working for you for free? Why don't you do it yourself, or hire someone to do it for you? If you can't do either of those, why don't you contribute documentation, mockups, or something else that's not technical but is still useful? Do you expect a baker to give bread to you for free when you criticize his breads for not being tasty enough?
In my experience, most of the people who complain about these sorts of things do so because the new app lacks some key feature that makes their workflow better. Dismissing someone's opinion simply because they have actually used similar commercial software is stupidity, as it means you are unable to gain wisdom from the study of designs that have come before yours. In fact, it is precisely this mindset that causes user interfaces in free software to almost invariably suck---the whole "this is the way we do things, and you'll just have to get used to it" mentality does not lend itself to learning. I'll probably get modded down for saying it, but it's worth burning the karma if it gets people's attention.
Couldn't disagree more. I would argue that most UI designers have no business coding any more than most coders have any business designing user interfaces. The two skill sets are very different, with very little overlap either in knowledge or in the mental structures used in performing the two tasks. Good UI design is more closely related to software architecture, not coding. You have to figure out how best to lay out a user interface based on careful study of what the user is likely to want to do, careful thought about what things a user is likely to group together as similar operations (menu layout), and careful study of actual users to watch their workflow followed by probing interviews to find out what worked well and what didn't. It is basically scientific in nature. Software architecture is the same way. You're looking ahead to what the coders might want to add, what features the users might want, etc. and designing the overall architecture to accommodate those future needs.
Coding is a different skill set entirely. It focuses on knowledge of what types of methodical steps you have to achieve to reach a goal. Software architects and UI designers are the architects where coders are the construction workers. Expecting UI designers to code is no less silly than expecting your architect to be proficient at hammering nails into boards to build a wall.
Further, a strong case for the separation of those elements is the Mac design model. You have a separate Interface Builder application (well, it's part of Xcode now, IIRC) that gives people the ability to design the UI without writing a single line of code. Then, coders can come in and hook that functionality up to the underlying code. By doing that, it not only allows people to focus on the things that they do best, but also makes it much easier to modify the UI because you can create it visually instead of messing around with programmatic drawing that constantly puts UI elements in the wrong place. I get the impression that some Linux developers do the same sort of thing with Glade, but I've never used it, so I can't draw any real comparisons.
Trust me, getting UI designers to write code is the last thing you want to do....
Check out my sci-fi/humor trilogy at PatriotsBooks.
I'm not sure I agree with this, but there is a definite confusion in most people's minds between easy to use and easy to learn.
Vim is incredibly difficult to learn, but is actually very easy to use if you can learn it. A lot of programmers using an editor for extended amounts of time have found this to be the case.
The problem with Microsoft stuff is that it's pretty easy to pick up and use. But once you've learnt how to do it, you often want command line tools to start scripting and batching work. Generally with MS, and to a lesser extent Apple, you find you can't do that.
Excel is the only Microsoft software I've used where I found it did everything I wanted as a power user. I think it's one of the very few examples of something both easy to learn and easy to use.
I also find that most people don't really know what they want. It's usually after a fairly long period of using something, when people start to understand what's specifically wrong with the application. By that time they're used to its idiosyncracies, and it won't help them to make the application easier for beginners.
Have you heard of usability testing?
Believe it or not, there are people who make a living figuring out that kind of thing. Obviously people can disagree about what's best, but you and the parent post seem be saying that usability is a lost cause so there's no reason to even try.
This is a case where the perception is the reality. If the audience for your app has a hard time figuring out how to use your it, it has poor usability, whether you accept it or not.
Your high horse is showing. :) If you don't like my GPL'd interface, you are free to design one of your own and hire someone else to code it for you, if necessary. FOSS does not mean, "anything you want, gratis!". It means if you don't like something you have the freedom to change it. I have little respect for those who use the word "free" as an excuse to try to hold up developers with their limitless opinion-based suggestions. Some projects are very open to user suggestions, others are very tied to following a well-defined vision. Neither model is "incorrect". No-one is forcing anyone to use any GPL software "as-is", that's the whole point. It's not so you can get your way while contributing nothing but opinions. Though occasionally you may get lucky like that. :)
Caveat Utilitor
The command line in Unix is more "usable" because it is older, more mature,
has more features and flexibility, and is more easily extensible. I can
augment a GUI with a few lines of shell scripting with considerably less
effort than it would take to "fix" the given GUI application.
I can make some shell script both easier and simpler than just about
any GUI application and tailor it to my own needs so that it most
closely meets my needs (and thus meets my own personal notion of
"intuitive").
That said, there are still Unix/Linux GUI applications that are
more functional, simpler and easier than their Windows/Mac
counterparts that we're supposed to be cloning. The "chasing
tail lights" criticism is a very good one in this respect because
sometimes neither of the allegedly "better" platforms do it right
or better.
At this point, it's no longer a given that the Mac or Windows version
of some GUI for some sort of app is any better.
Improvement of Linux GUI's should not be limited by what Mac or Windows
applications do or what design dogmas they follow.
A Pirate and a Puritan look the same on a balance sheet.
I wouldn't go so far to say that a critique is useless, but there is truth in his statement. A lot of usability problems are not just where you place the buttons, but they are much more deeply down in the code structure. Take Gimp for example, one of my issues with it is its lackluster animation support, you of course can try to polish the GUI a bit, but the real problem are much deeper down into the code and need to be fixed there first before you can even start to think about a proper GUI, same is true for a lot of other issues.
Usability in the free software world is simply not just an issue of GUI, but goes down through all layers of the code, since after all, they have to be usable too, both by developers and users. Tackling the GUI is all nice and good, but it really is only a small part of the whole picture.
The command line in Unix is more "usable" because it is older, more mature, has more features and flexibility, and is more easily extensible.
And also because all these Computer Science / Physics / Engineering graduates that argue how easy is to get things done with a shell so conveniently forget the amount of time they dedicated to learn how to use a shell.
I am personally (much) more comfortable using zsh than any GUI for manipulating files, or other tasks. But what most people tend to forget is that there is a much larger pool of users that are not willing to memorize 200 commands, and read man pages for days and days in order to just use their computer.
"Command line for experts, GUIs for casual users" is a false dilemma anyway. Which is why people don't generally use command lines to draw images, compose music or construct cars and airplanes on a computer, no matter what their skill level is. Even the most skilled "Engineering graduates" use, among other software, CAD programs (which have GUIs).
EXACTLY! I can learn a programming language and write a program that does precisely what I want it to do. But that takes a vast investment of time to have a tool which does instantly what I need done. Perfect for people like programmers who might do one thing over and over and over and over... but not so hot for someone who just wants to do something once every 3 months.
The bane of the command line program is this: ">"
I sit there and stare at the little arrow. I want to copy a file. But all I can see is ">" What does ">" mean? Maybe I should type in what I want to do "CopyFile"... "command unknown". Great... well I'm out of ideas. Yes the interface might be faster... if you don't have to open a MAN file every 10 minutes because you forgot if it was cpyfile or copyfile or cpyfle or copy_file or copy-file or... whatever. The point is. Command lines are like code. They will do exactly what you want them to do once you know the magic incantantation with the right pronunciation. But if you know what you want to do but have never used the software before you're SOL. This is why I love node based tools. When I'm learning a new programming language I have no idea what classes and functions exist. I might even know the exact tool I need "I need to find the position in a string where "hello" is found". But translating that into a class name is a game witchcraft and endless help document scimming until you happen to find an arbitrary function name which does it.
GUIs show me my choices. Yes they show me my choices every single time. But I like to know what it is I can do. A BAD gui doesn't show me my opportunities. It buries it in a menu which is as bad as burying it in a command line command somewhere. But even a GUI is better than a command line while learning because when you finally do find the command you want to do... you can act on it instantly! No closing the MAN file reciting "'copyfile 'filename' -f -s-t-u #silent *underhanded" in your head hoping you don't forget it before you type it in.
Again many developer tools can operate on command line because the functionality barely changes from version to version. In all of the software I use every release adds hundreds of new features I need to learn and I need real time feedback on changes I make... I can't wait to execute a command to see if it's right.
Take SQL as a perfect example of a command line tool. SQL has barely changed in its command structure in 15 years. Of course you're super fast at writing complex SQL scripts that deliver exactly the information you need! You've been practicing those exact same commands for 15 years! What if SQL radically changed every few months? And what if instead of only a few dozen commands chained together it was 10,000 tools and settings in a single application? You cany very quickly get to the point in most software where it is simply impossible to use it as a command line tool and expect a user to use it without a MAN file permanantly open next to it.
I've spent 3 years learning MaxScript for instance for 3DsMax. Which is its command line toolset. It can take me a week to create a script which I can do in 20 seconds using the GUI. Once I write the script it might take an artist half a second but even I completely forget the name of a function I wrote earlier in the morning "What was it again? GetObjectIDFromDatabase or was it getObjectfromDatabase... what was the order of the function calls again? What was the name of that function which... I wrote it myself! And I still have to refer to my own help file to remember what the name of 90% of my commands are. It's unreasonable to expect someone to remember the command names and possible flags for 200 functions... and that's a primitive script. And even if I did expect someone to memorize all 200 commands... they would just have to memorize another 200 the next month when another tool is added to the toolset... and another 200 the next month after that... and another 200 the month after that.
Have you actually used AutoCAD, say?
Yes I have. (I'm an avid Emacs user too) You're right, the command line in Acad is great, but it's the combination of the command line (and its scriptability using AutoLisp) with the UI that makes it truly useful (e.g., as you say, the fact that each time a point is required, it may be entered numerically as well as interactively using e.g. snapping). Acad was written by judicious and wise people. If it had been written by pure click'n'drool advocates, it would only have a GUI. If it had been written by die-hard "GUIs are for losers" Unix freaks, it would only have a command line.
It absolutely amazes me how many people like yourself get modded +5, Insightful in every single discussion like this, when you're all dead wrong. It amazes me how you think you're being logical and can go to incredibly great lengths to justify the continuing elitism of "if you don't know how to build the car you have no business driving it" kind of attitudes, which is exactly what you're espousing. Did you build your own car? Do you know how every single piece of it works? Can you recreate it from scratch? If not, shut your stupid pie hole.
This idea that you have to be the world's foremost expert in a particular field in order to be allowed to open your mouth is just so unbelievably ridiculous that I can't understand how it keeps being perpetuated so strongly even on a site that is supposed to be full of relatively intelligent people.
You're all constantly using obscure corner cases to try and demonstrate that any end user who ever criticizes the work of a "coder" is an idiot and wrong in all cases simply because they don't understand the precious code. Sure, users often misunderstand what the actual issue is with a program that's not working for them, so what? The fact remains that there is an issue, and blaming the user every single time solves nothing.
The longer attitudes like this get perpetuated, the longer most open source software will remain the underdog that most non-coders won't touch with a ten foot pole because it's so baffling or aggravating to use. What all you elite coders need to come to grips with is that you are too close to the code. You understand the code too well and it blinds you to the real life usability problems. You don't realize how much your knowledge of the code warps the way you see the interface. You have to learn to forget what you know about the guts of the software and look at it with fresh eyes, like any new user does.
The real question here: If you aren't making the software for people who don't know how to code, what are you doing putting a GUI on it in the first place? That is after all the entire purpose of a GUI in most cases, is it not? To make software accessible to and usable by people who don't know how to write the code themselves?
Jeebus cripes, folks. Get over yourselves. Stick to writing command-line stuff if you don't want to respond to interface criticism or suggestions like a reasonable human being.
Oh, and stop using Firefox as if it shows how great all open source code is. It's one application that has been worked on by teams of very talented people for several years and is supported by a business. Of course it's one of the best, but 99% of the open source software world falls a few miles short of that. Face up to that fact and you'll be better for it.