Fog Creek Software
Discussion Board




Usability v Learnability

Not the real question: Now that the action has kinda shifted to the forum, would it be better to shift the link to Ask Joel to nearer the top of your home page?

I have to scroll to get to Ask Joel.

The real question: I liked the point in your book on "learnability" v "usability". We have a web app that has to capture a lot of information and provide a fair bit of not unsophisticated* analysis.

We have a few learnability issues, and we find that there is a real trade off between this and usability - the more learnable we make it, the more noddy it becomes for regular users (which the average user will become over time): "Yeah, yeah, I get it, now get lost".

Have you had to deal with this much - at what point do you say "You DO have to have some training to use this."

*I learned in Law School that, unlike in programming, a double negative does not necessarily equal the positive.

Patrick FitzGerald
Thursday, March 18, 2004

It depends on the type of users you have and how experienced they are.

If your application is a retirement planning app that the average person will use once every five years, you better go for learnability. If your application provides a form that professional travel agents will use every ten minutes to book flights, go for usability. The more frequently your users use the app, the more you can get away with a learning curve.

In either case it's sometimes possible to create good compromises and get the best of both worlds. For example in a traditional GUI application with pull down menus, the pull down menus provide a canonical, complete list of commands but there are also keyboard shortcuts for a lot of the common ones which people learn over time.

Joel Spolsky
Fog Creek Software
Thursday, March 18, 2004

Patrick,

I asked about possibly moving up the forum links to avoid scrolling but Joel decided not to answer it. I then remembered seeing his mention of a forthcoming redesign of the JoS site, so I'm assuming that this will take care of it in some manner.

That, and Joel doesn't like people questioning his UI choices :) (or asking for the reasoning behind them).

Wait a little while, you might get what you're asking for.

  --Josh

JWA
Thursday, March 18, 2004

Actually, what he implicitly said in his reply was "Get a bigger screen."

Anon-y-mous Cow-ard
Thursday, March 18, 2004

>> "The real question: I liked the point in your book on 'learnability' v 'usability'."

I believe that the 'learnability v usability' issue is a most vile and detrimental false dichotomy.  The extent to which it pervades software engineering astonishes me, and scares me when I think of the sheer magnitude of wasted man-hours expended by users of poorly designed software interfaces.

In mathematics, a considerably more mature discipline, practitioners acknowledge the importance, power and beauty of elegant work.  Elegant proofs are indeed powerful, because, while a poorly constructed proof might consist of pages and pages of impenetrable esoterica, its elegant counterpart might literally be a few lines which even the uninitiated could well understand.  Elegant proofs come from God, according to the Hungarian mathematician Paul Erdos. His theory is that God has a book containing all the best proofs.  Well, right next to this fancy proof book on God's shelf is a book of elegant, learnable and usable software interfaces - and Outlook ain't in it, and neither is any Unix shell. 

If your software is not both learnable and usable, then you haven't worked hard enough.

Dave
Thursday, March 18, 2004

"Actually, what he implicitly said in his reply was "Get a bigger screen." "

Um, what reply?

  --Josh

JWA
Thursday, March 18, 2004

Dave,

I am not sure I 100% agree.

If what you said was true, there would be no need for "X for Dummies" books, nor would there be a need for colleges that train millions of people in how to use specialised software every day.

This issue will come up for people who are designing software for a wide "church" - seasoned pros who want to chug through the process at a rate of knots and complete novices who are frightened using a complex tool, paid for by a third group who have a different perspective on the software and shared with a fourth group, the consumer, who can view and play with the software a little.

Somehow a balance must be struck.

You should put Photoshop in the hands of someone who has finally come to grips with how to send an e-mail. They will struggle, although that person might well master it if properly trained. I don't think it is up to us to tell such people "You are not allowed to use Photoshop - that is only for the gods of the modern age such as myself to use".

As this untrained person struggles with Photoshop , would you turn to its designers and say "You obviously have not worked hard enough"?

Our software is automating simple processes containing complex analysis in a specific industry that has a lot of 40-55 year olds who know their stuff but are not all that tech-savvy. However, they have to get into some heavy-shifting detail, and over time their numbers will dwindle as the younger, more tech savvy take their place.

I was just wishing to draw out some comments from the audience to see how other people coped with the dilemma.

Thanks for your views, however - I feel like I got my first post on Slashdot.

;)

Patrick FitzGerald
Thursday, March 18, 2004

>> "This issue will come up for people who are designing software for a wide "church" - seasoned pros who want to chug through the process at a rate of knots and complete novices who are frightened using a complex tool..."

Interesting point.  My response would be that a sophisticated designer comes to terms with the fact that a particular interface can't be all things to all people.  Going back to my earlier analogy of proof in mathematics, one can't put forth a elegant proof that the set of primes is infinite if one has an audience that spans preschoolers to post-docs - and a good mathematician wouldn't try.  A savvy GUI designer, similarly, must define a manageable target user class.  If your goal is to create an advanced CAD program, and part of your target user class doesn't grasp pointing and clicking, then you're doomed.

Now, this is not to say that you can't have one software package that meets the needs of novices and experts alike.  Perhaps you have separate 'elementary', 'intermediate' and 'advanced' interfaces into the same software.  Perhaps you've some other clever solution like an easily configurable interface that - with a fews clicks here and there - can suit everyone's needs.  The point is, with enough inspiration, craftsmanship, and hard work, there is an elegant solution. 

Dave
Thursday, March 18, 2004

I'm afraid Dave I subscribe to the school that sees elegance comfined to fashion, methematics and other fields that have little to do with reality.

Software is more like language; it's a mess. And it's a mess often because it evolves. The main thing that prevents version 2 being elegant is version 1. When you add new features you either have to bolt them on to the existing ones and create a kludge that becomes more kludgy with each version or force all version 1 users to start all over again. It's pretty clear which users prefer.

Word is an example of this. I haven't looked at a book on Word since I started with computers round about the time of Word 95. Yet it would only take me an hour or two to become a power user of Word 2003 which I haven't even seen yet. Yet nobody would dream of calling the Word interface elegant.

Stephen Jones
Thursday, March 18, 2004

>> "The main thing that prevents version 2 being elegant is version 1. When you add new features you either have to bolt them on to the existing ones and create a kludge that becomes more kludgy with each version or force all version 1 users to start all over again. It's pretty clear which users prefer."

Hmmm...  Stephen, you claim it's pretty clear that users would prefer consistency in a user interface to a new, more elegant, powerful version.  This certainly sounds plausible.  But think for a moment about a pretty popular user interface of thirteen or fourteen years ago - command line DOS.  If your assertion were correct, MS users would not have made the switch to Windows 3.1.  Neither would people have switched from 3.1 to 95.  I think this illustrates an interesting point:  Users will be more than willing to switch interfaces if the newer interface offers them enough.  And who's job is it to create interfaces that offer the user something?  Ours, of course. 

It should have been Microsoft's job to look at that crappy Word interface and create a new, elegant and usable interface that offered the users so much that they couldn't possibly refuse.  That Word looks pretty much the same as it did in the mid-nineties is not evidence of users' unwillingness to accept change, it's an indictment of the software community for chronic intellectual and creative stagnation.

P.S.
By the way, I personally not only switched interfaces, I switched applications from IE to Mozilla because of Microsoft's inertia.

Dave
Thursday, March 18, 2004

Dave, we have seprated the "views" of the system. Unfortunately, the main view of the system has to be shared by a wide range of people. These users must go through the main steps - we cannot hide things from them or simplify things for compliance reasons.

It is not so diverse as a CAD program being used by people with a point-and-click problem.

Our usability sessions for this one group of users have been interesting. Some come out and say "too simplistic" others say "too complicated".

The interesting thing is that they are all saying that different aspects were difficult. So, one group found the graphing/modelling page easy-to-use (unlike the last user) but found the analysis page a bit more challenging to come to grips with (which the last user found a breeze).

This is an issue Raymond Chen has written about: http://weblogs.asp.net/oldnewthing/archive/2003/07/28/54583.aspx

One approach we quickly rejected was "direct it squarely at the main body of the customer base".

Not only was it too difficult to work exactly what skill level they had in whatever the middle was, but we also found that people like to have something that is just a short notch above their skill level (so they feel that it is a serious tool).

Notch measurement tools required!

Patrick FitzGerald
Friday, March 19, 2004

The point about GUI's versus command line interfaces is that the former allow you to have a whole batch of new users, because the GUI works on the principle of restricted choice.

And you can still do  hell of a lot through the command line, even with XP!

Stephen Jones
Friday, March 19, 2004

Patrick FitzGerald,

Wow!  Seems like a pretty grim situation.  It sounds like your team has put a great deal of thought into the problem and there's still no apparent solution.  Perhaps you've actually got a specification which has no elegant implementation.  I have to admit that, in reality, there are some dreadfully ugly mathematical proofs out there - the proof to the four-color problem springs to mind.

But, when Erdos talked about God's book of elegant proofs, I don't think that he meant that, in the most literal, quantifiable sense, every proof can be transformed to an equivalent proof which is 'elegant'.  What is 'elegant' anyway?  An eye of the beholder kinda thing at best.  And I have a feeling that some logician can prove that, in any given formal system, there exist statements that can't be proved in less than five-hundred pages - and I don't imagine anyone would consider a five-hundred page proof elegant.  I think he simply meant to impress the audience with the fact that in Mathematics, a premium is placed on attributes akin to brevity, readability, and learnability etc. 

Similarly, I don't mean my comments to be taken in their most literal form.  There may well be some specification - maybe your specification - that won't lend itself to an 'elegant' interface.  But consider this:  When we work with graphical user interfaces, we are not bound by cumbersome formal systems like poor Erdos.  We're literally bound only by our imaginations.  Graphics are sometimes worth a thousand words, and if the designer is inspired, they might be worth ten-thousand.

Good luck!!! 

888 

Dave
Friday, March 19, 2004

Dave

I wouldn't call it grim. It is challenging, for sure, but a lot of fun on the other hand. We are toying with some new-ish ideas on the system-user conversation and are about to take them to the user groups.

I think most of them will be rejected, but some will be accepted. See how we go!

P

Patrick FitzGerald
Friday, March 19, 2004

I used to see UI design as exactly a question of who you support : naive or expert users.

Ideally I wanted to do both. But couldn't quite see how.

Now I have a new way of thinking : The UI should support naive users *becoming* expert.

What does this mean? It should support them learning to move from thinking about the problem in rather concrete, literal way (ie direct manipulation of objects with the GUI), to thinking about it in a more general, abstract way (manipulating higher level objects such as groups, templates etc).

There's a discussion of this here : http://www.nooranch.com/synaesmedia/wiki/wiki.cgi?AnEasyInterface

phil jones
Sunday, March 21, 2004

By the way, I'm not sure what's so terribly wrong with Word's interface. Please enlighten me.

Frederik Slijkerman
Sunday, March 21, 2004

the main problem with Word's interface is finding things.

You have some formatting settings in Tools|Options others in Tools|Autocorrect and others in Format|style and so on.

The bulleting and numbering is a nightmare to control. You have to use the icon to decrease indent to change the level of outline numbering,as using tab has no effect.

To insert text in a header and footer you go to the View menu, but to insert a page number you go to the insert menu.

This is Word 2000; I am sure it hasn't got any better with word 2003.

There are real problems with knowing what exaclty Word is doing automatically and what it isn't.

And of course there is the question dealt with in Windows annoyances of the default choice of icons for the taskbar, which seems to have been dictated by Marketing circa 1996 rather than by any study of real usage.

Stephen Jones
Sunday, March 21, 2004

Phil

Great link - thanks.

P

Patrick FitzGerald
Monday, March 22, 2004

I 99% agree with Dave!

MX
Wednesday, April 21, 2004

*  Recent Topics

*  Fog Creek Home