Slashdot Log In
Jakob Nielsen Answers Usability Questions
from the necessary-knowledge dept.
1) /. usability rating? (Score:5, Interesting)
by Col. Klink (retired) (wklink@yahoo.com)
Would you care to comment on the usability of Slashdot? Good? Bad? Ugly? Be sure to read the apache section before answering that last one.
Jakob:
Obviously, Slashdot has great usability for its targeted user base of nerds. The proof is in the pudding, in that they use it so much and keep coming back. There is nothing here but pure user interface: nothing you buy or get, so if people use it, it must be because it is good. This said, many elements of the interface would present too much complexity for more average users. For example, the many different ways of viewing and sorting threaded discussions is quite difficult to understand. How do you really know what you will see if you click on one of the links from the home page?
There are three elements of Slashdot that I particularly like:
- Simplicity in the layout itself: focus on content rather than flash.
- The liberal use of linking - in fact, the site lives off the ability to link to the rest of the Web. Too many other sites forget that hypertext is the foundation of the Web and provide nothing but a closed world.
- The reputation manager effect coming from the moderation system.
2) Short vs long pages (Score:5, Interesting)
by Anonymous Coward
In most of your writings and interviews, you seem to be recommending short pages as always better than long ones. Sometimes you qualify this as applying only to 'navigation pages'-- but you never define that term. Aren't there more complex rules about when it's okay to have a long page? Don't you yourself find it frustrating when you have to load multiple pages, when one longer page could easily have held all the info?
Jakob:
This very page itself is getting to be too long :-)
It depends on how you look at this page. If you think of it as a single interview, then it is best to preserve it as a single page since the users would just skim over those questions that don't interest them. If you think of it as a set of answers, it would be better to have a short summary of the entire collection and then have links to individual questions and answers. Unfortunately, the Web is currently too slow to support this type of hypertext (we need subsecond response times for true freedom of movement). It would be nice to have a more advanced model of Web hypertext that would support alternative views of both atomic information objects and composite information objects.
Talking about navigation pages, brevity does rule. Users need to be able to get an overview of their choices without having to scroll too much. Ideally without having to scroll at all. If you need to scroll while making the choice of where to go next, then you are forced to keep promising options in short-term memory after they scroll out of view. The Slashdot audience may not fully appreciate the problem of having to keep items in short-term memory since you only become a programmer if you are good at doing this. Most people are not. All users will furthermore suffer from the tendency to select among the visible choices: if something looks like a good answer and as the best answer, then users will often follow that link without scrolling down to see if there might be an even better link on the invisible part of the page.
3) Browsers compensating for bad sites (Score:5,
Interesting)
by Ed Avis (epa98@doc.ic.ac.uk)
To what extent will people start using their browser's features to compensate for bad Web sites? For example, your browser might automatically convert frames to tables, or precis long chunks of text, or concatenate lots of bitty pages into one easily-readable page. Since there will always be badly designed sites out there, do you think this is a useful sticking-plaster?
Jakob:
Great idea. The Web has always been based on this notion to some extent. For example, the Back button in the browser (as opposed to relying completely on site-supported navigation) and the ability to make the font bigger or smaller (as opposed to hoping that every site gets it exactly right).
We may have temporarily abandoned some of the user control over the Web in the chase for better-looking pages, and one of the worst sins in using CSS is to specify text in an absolute font size that doesn't change if the user needs bigger or smaller text.
I am hoping that future generations of browsers will finally live up to their names and actually help users browse (or Navigate or Explore, as the case may be). If the big browser vendors won't do it, then that's a potential market for other browsers like Opera and iCab or for various types of browser add-on tools.
4) Patent culture vs Open Source culture (Score:5,
Interesting)
by tbray
You are the holder (or co-holder) of quite a number of patents. Can Open Source software builders who construct, for example, something that "prints a hyperspacial document" or "updates visual bookmarks" expect to be hearing from your attorneys?
Jakob:
The literal answer is no, since "my" patents are actually not mine but owned by the company I worked for at the time. I cannot speak for the attorneys of Sun since I don't work there any more. But it is pretty standard for big computer companies to get as many patents as they can for basic reasons of self-defense: if somebody tries to come after you then you can fight back with your own patents. That usually does not mean that the company wants to go after smaller companies unless they attack first.
5) Revolutionary UNIX GUIs (Score:5, Insightful)
by Anonymous Coward
In a Wired article on Eazel posted to Slashdot the other day, you said:
"They need to rethink the entire approach... They're saying let's implement a Mac-like interface so that we can have a nicer Unix. That's a nice thing, I guess, but it's not really revolutionary."
Can you describe some specific ideas and UI elements you would consider if you were designing the "revolutionary" Linux GUI?
Jakob:
I know that Slashdot readers don't want to hear this, but the very first question is whether it is even possible to create a truly good user experience on top of Linux. Many other companies have tried to make Unix easy to use and many very talented designers have worked hard on these projects for several years without very good results.
The only data points we have say that it can't be done.
I tend to believe in an alternative interpretation of the data, which is that the various approaches to designing better Unix interfaces were doomed because they always kept reinventing the same thing again and again. They never did the two things that are necessary for great UI:
- Don't just reimplement something that had a different design center (the Mac which was designed for a small black-and-white screen, 1MB RAM, and a puny 68000 processor)
- Iterate. Your first design will be a flop (say, Xerox Star or Apple Lisa). You gotta keep improving rather than giving up as the Unix vendors have done.
There is already one type of revolutionary UI built on top of Linux: embedded systems in the form of information appliances. Linux inside. You can't tell from the outside, though. A typical information appliance may only have 2-3 commands in the form of simple buttons or knobs.
All respect for info appliances, but we also need a workstation-style interface that can help knowledge workers survive the information flood of modern society. And that's where I think we really need revolutionary designs that go beyond the Mac. For example, ways of managing tens of thousands of documents by a rich set of attributes and content-oriented navigation. Simply showing files as icons in folders doesn't cut it beyond a few hundred.
We also know from many studies that the average user is very bad at hierarchical filing and typically never moves a file once it gets to live in some directory. Even if the file would be better off elsewhere. This problem is magnified several hundred times when it comes to managing email. I am starting to think that the solution is to treat information objects as members of a soup and manage them by attributes rather than by hierarchy and name.
6) Standards Compliance (Score:5, Interesting)
by HerrNewton
What are your views on standards compliance for, baseline, HTML 4.01 and CSS-1? Are we fighting towards a goal which is universally unattainable (due to the embbeded nature of some browsers like WebTV and *cough* IE on Windows), or are we nearing a new age for web developers?
Jakob:
At least WebTV can update its browser when/if they decide to do so. And IE is also getting better, even if it doesn't do everything I would want. But we will soon see a new generation of hardwired browsers inside information appliances. Once a piece of consumer electronics ships, it usually doesn't get upgraded. Thus it will be really important to campaign for full standards compliance from such truly embedded browsers.
I am basically hopeful that we will see more respect for standards on the Web. The concept of proprietary extensions has lost and very few mainstream sites do anything any more that cannot be seen by the vast majority of users. This is one of the true benefits from the boom in e-commerce. No self-respecting salesperson wants to turn away paying customers at the door just because they don't have the latest beta-download of some browser.
7) Non-GUI apps and usability (Score:5, Interesting)
by washort (washort@samford.edu)
Much attention is given to usability in GUIs and Web sites, (such as in your column Novice vs. Expert Users) but what about textmode and primarily keyboard applications such as text editors? Personally, I believe that Emacs have the best user interface of any text editor I've ever used (vi's a close second, calm down people :), but it's geared towards experts. What do you see for the future with regard to synthesizing novice usability and expert usability? the "smart menus" as seen in MS Office 2000 seem to head in that direction, only showing basic options unless an expansion button is pressed at the bottom of the menu. The best touch is that it "remembers" what you last used from the full menu and puts it on the basic menu. How can we smooth the curve?
Jakob:
There was a good deal of research on the usability of textmode UI back in the late 1970s and early 1980s. Conference proceedings like CHI'83 (first large user interface conference) are filled with papers on issues like command abbreviations and best use of command keys. However, as we all know, interest changed to GUIs after the Mac came out in 1984.
We do need more attention to the productivity of expert users. All the same methods apply for how to study and measure interfaces, no matter what their interaction style, but I admit that there is not much work these days on keyboard interfaces.
The other part of the question is much harder to address. How to smooth the curve from novice to expert. Nobody has found the way yet. Cue cards, boot-up tips, and the little annoying paper clip are all attempts, but nothing works really well. Progressive disclosure is the best tool so far: show people the basics first, and once they understand that, allow them to get to the expert features. But don't show everything all at once or you will only confuse people and they will waste endless time messing with features that they don't need yet. Interestingly, research by Jack Carroll at IBM in the 1980s proved that a "training wheels" approach to computers makes people better at understanding the expert features once they get to them. The reason being that users learn the conceptual structure of the system better when they are presented with the smaller set of features first. Not seeing something during initial use of the system would result in better use of the hidden features later.
8) Education (Score:5, Interesting)
by Duke of URL (iridium@sporkandspam.mauimail.com)
What type of education did you (and others ) have to receive to become a useability expert? Basically what's the best route to get a career in human-computer factors?
Jakob:
The only real way to become a usability expert is to watch lots of users as they perform lots of different types of tasks with lots of different designs and interaction styles. If you have only seen people use a single type of user interface, then you don't have the breadth of understanding of user behavior. I find that I often draw what I learned from the studies we did with IBM mainframe interfaces back in the 1980s, even as I advise on the design of websites. One reason, of course, is that many Web interfaces are as primitive as the old 3270 designs. But another reason is that watching what people do under many different circumstances helps generate insight into what they will do under new circumstances.
There is no single answer in terms of degree. The best people in the field today have degrees in countless topics, including psychology, anthropology, computer science, mathematics, graphic design, and theater. None of these degrees is perfect for becoming a usability expert. The real way to learn usability is to do usability as much as possible.
9) What's Next? (Score:5, Interesting)
by moonboy (armstrong.spamalicious.99@yahoo.com)
What is the next "big thing" in interfaces?
Surely "windowing" can't be the end-all-be-all of interfaces. Is there some paradigm shift around the corner which we can't conceive of right now? Perhaps the same "leap" which occurred going from command line/text to windows.
Jakob:
There are two things I do not think are the next big thing: 3D and speech recognition. Speech suffers from the Star Trek fallacy: it's a great audience interface but not a good user interface in most situations.
I think there are two big paradigm shifts coming: Augmented reality and content-and time-based computing.
Augmented reality is the ability to project a user interface onto the physical world. For example, when repairing an airplane engine, a trainee mechanic can see an animated hand grab exactly in the right spot. And read-outs from various diagnostics will display in the context of the thing they are diagnosing rather than on a separate device. Lots of other ideas in this realm, including wearable computing, smart clothes, etc.
I also believe we need more information-rich interfaces as I was discussing above. I think the current Macintosh-style UI will be turned inside-out and we will start to manage information objects depending on a much larger set of attributes than simply their name and hierarchical placement. In particular, history and other time-based attributes will become more important. When did I last touch this object? What other things were I doing at the time?
Also, the computer will need to become a personal secretary and help the user manage his or her time. The opposite of push technology which was based on constant temptation to procrastinate. In the old days, an operating system was designed to optimize the utilization of the computer's resources. In the future, its main goal will be to optimize the user's time. For example, in terms of protecting you from too much e-mail.
10) Disturbing anecdotes (Score:5, Insightful)
by Anonymous Coward
Jakob,
Your work is chock full of terrifying statistics about what happens when we create slow, hard-to-navigate sites. When I (an information architect) try to convince my project teams to heed those statistics, though, nobody seems to listen. People continue to clamor for images, frames, JavaScript, etc.
If Ronald Reagan's speeches proved one thing to us, it's this: a well-chosen anecdote can drown out innumerable (and true) statistics. I was wondering whether you might have any good terrifying anecdotes that might scare people who are about to make an unusable Web site into doing the right thing.
Jakob:
Boo.com is one good anecdote. They wasted millions of dollars on fancy design which they had to retract shortly after the launch because nobody could use it. Even on a fashion site, people care more about the products than about the bleeding edge design.
Also, the Web itself is one big anecdote. What do all the big sites have in common? Minimalist design. I made a very simple analysis of the usability of the ten sites with the most traffic compared to the sites from the ten biggest companies (which would have had an inherent advantage if they had been more usable). The result was very clear: The ten biggest sites had much better usability scores than the sites built by huge corporations. For example, the download time for the home page was eight seconds for the big sites and 19 seconds for the big companies.
What happens is very simple: the good sites win. If the pages download fast, people return. If they can find the products, then they can buy the products. If people understand the site, they use it.
Re:WOW! But where does the stupidity come from? (Score:3)
Having worked for years in the web development industry, I can tell you really simply why sites are like this. It's because clients are DUMB! That's right. If you show them a usable site, they won't actually use it themselves, it's their clients who will use it. They don't understand usability. However, they DO understand big flashy 3D Flash Java sound applets! Especially when the demo site is loaded on a laptop's harddrive which can achieve a megabit/s throughput ...
Sure, you could educate the bozos. I've tried to do that in technical meetings. They just look at you funny and eventually they buy the competitor's Java-bloated project. Duh.
Re:Minimalist design (Score:3)
Nielson's observations boil down to two axioms:
- The longer your site takes to load, the greater the chance that someone's going to give up.
- The more useful your site is to someone, the more likely they will come back.
Even on my cable modem or at work on a T1, I sometimes find myself switchingSure, users want to experience attractive websites, but it's much lower on the hierarchy of needs. If a pretty picture is keeping content from me (and is not, itself, the sought-after content), I'm going somewhere else.
"Nielsen's advice, while applicable to all web designers, tends to encourage to creation of look-alike e-commerce websites. The needs of a user buying stuff at E-Bay are different than a person checking out an online art gallery."
Remember that he is giving general advice; the most usable website for an amalgamated average of all users of the Web is probably a generic e-commerce site. (Since usability is directly proportional to profit for these sites, that's not surprising.)
Your point is well-taken, though -- different audiences will prefer different UI's, and sometimes the general rules don't apply. Ditto for different user tasks.
It would be instructive to see some field- and task-specific usability analyses on useit [useit.com] at some point.
"What the big sites have in common is that most of them got started on the web at an early stage."
Lots of entrepreneurs "got started on the web at an early stage" -- and aren't around today. The really big sites stuck around longer because they were more usable.
"The "big" sites maintain this advantage now through massive amounts of advertising. That's what big sites have in common these days. Yep, they all look like portals, but that's a fad thing, not a design consideration."
I actually don't like portals very much (I usually ignore 80% of the page once I can find the "search" blank), but that is just one page out of the site -- albeit a big one. Where e-commerce sites deliver is on their catalog and ordering pages. Making someone feel secure with your credit card number is no mean feat.
Advertising actually is something that counteracts usability to an extent. However, remember the old lesson of Maapo cereal -- advertising gets the first visit, but not all the rest.
phil
Re:Essence of goo UI (Score:3)
The fact is, a GUI can be user-optimized and not waste your system's resources. Slashdot itself is an excellent example of this -- you can select a number of different ways of viewing the page, and the code responds accordingly. Try turning off icons with articles, for example: /. saves the bandwidth required to send those images, the processor time to locate the images, etc., and the user doesn't have to wait for the images to load.
The key is designing the system to have what is called "graceful degradation" -- which isn't necessarily that hard to do programmatically -- it just easier if it has been thought out well in advance. The key is, when things degrade (performance is bad, an item is turned off in the browser, a user isn't advanced yet, etc.), how and in what order do I gracefully remove the non-essentials without disturbing the user's ability to have a successful interaction with the system?
Keep these things in mind and it's a win-win for everybody.
Re:Attributed Filesystems (Score:3)
Basically, its much easier and more likely to succeed to devise products around the way people work than the way people think.
What I'd like to see is a number of things that make common work tasks like refining your work and sharing it easier.
First off, why don't operating systems version files for you? VMS did. It's so obviously useful, if you give the user control over it. I'd also like to go back to a prior version of a file and fork it. I'd like to be able to check out documents and check them back in. I'd like full fledged calendaring and project management to be a standard desktop feature.
In many ways, the Lotus Notes addresses many of these kinds of concerns: versioning, routing, commentig, indexing etc; but the UI is laughably bad and the integration with other components cheezy.
Re:#8: Types of education... (Score:3)
This is a classic problem. At every company that has a project massively behind schedule, you'll find copies of _The Mythical Man Month_ and _Understanding the Software Development Process_. Heck, Microsoft Press publishes some of the most respected books about software development, and look at the messes their parent company produces.
It's easy to chalk this up to the "That doesn't apply to me" syndrome, but I think it's simpler than that: people know what all the experts say, and might even quote them in interviews, but they don't really believe it deep down inside. Someone wanting to start up a news site, for example, may read all the books about site design, but goes with a graphic-laden site because (1) that's what other sites do, and (2) he or she wants to look as good as the pros and not so homemade. We *know* that you shouldn't judge a book by its cover, but then we still want the fancy packaging. I've known people who have bitterly complained about software packages that come in jewel cases or have black and white manuals. They'll just shove the case in the closet and never read the manual, but they want all the useless frills anyway.
One time I used 100% recycled paper for a customer mailing. I thought using recycled paper was a good thing, and I was glad to support the maker of the paper. But in the end I think that people looked at the gray paper as cheap and very unpolished and would have preferred something on bleached white, virgin paper. If you asked all those people if they supported recycling, though, they all would have said yes.
The bottom line is that it's easy to convince people that usability and simplicity are more important to web site design than gobs of professionally done graphics, but that doesn't mean they'll listen. Because they won't.
#8: Types of education... (Score:3)
Many large companies, like the one I work for, have usability departments with Intranet pages linked to Jacob's pages - and yet the company's pages still have way too many images all in the name of marketing.
Nielsen is a typical Slashdot Lefty (Score:4)
The answer is simple: There is no praise because Nielsen, like Slashdot itself, is a mindless pawn of the liberal media. As each day passes, the number of objective sources where true, red-blooded Americans can get real news about the state of their country dwindles. Soon, all we'll have will be CNN (Clinton News Network) and the three major networks, all of which are owned by noted Communists. There are already plans in Washington to have Fox News Channel shut down. Let's remain vigilant, friends. We can beat these fucking lefties.
WYSIWYG is your enemy, mod_include is your friend (Score:4)
I've been hacking around with mod_include directives, and have settled on the following scheme. Every page requires three files: the content file, the template file and the configuration file. (actually four if you count the style sheet).
The content file has the text of what you want to say with only basic markup. It is just an html snippet. This can be done in an HTML editor and the body snipped out, but I find it easier to write the content as a text file and put the markup in manually after.
The template file is shared by all or many of the pages in the site. It includes the navigational elements (actually included from another file) and decoration. The HTML start and end tags are in the file, and somehwere inside this file there is an include directive to include the content file. My own template file uses a table and the content is included in a table cell.
The config file is the one that is the target of the URL; it sets some variables and then includes the template file, which in turn knows what content file to include because of a variable I set here. Here is an example:
<!--#set var="readfromfile" value="hometext.shtml" -->
<!--#set var="title" value="Advanced Computer Resources" -->
<!--#include file="frame.shtml" -->
This may seem like a lot of rigamarole, but in fact the benefit is that every page can conform perfectly to one or two different templates. Combine this with style sheets and one of the biggest problems with making a site usable is simplified: making sure the guidelines are followed everywhere.
Progressive Disclosure & Experts (Score:4)
A long time ago, Kai Krause (or someone at then-HSC Software) tried this in KPT Convolver. There was a 'stars' system where after you performed certain types of actions enough times you got a new star and a new tool (or toolset) which assisted in what you were doing and broadened the possibilities. It was much-hyped, at least by HSC.
As a general-computer power user, it bothered me. A lot. Shortly I found a 'secret' document which detailed what you had to do in order to get each star, and tediously repeated the tasks until I had all the stars. (Or all but one...I think the document was from HSC and didn't detail the last star, but just hinted.)
Part of the problem was that there was no simple way to enable all the options, you had to "earn" them. If you decide to embrace progressive disclosure in your next project, please consider the power users and provide a way to easily switch to Full disclosure.
Minimalist design (Score:4)
I'm a bit disappointed in the questions that Jakob Nielsen answered. None of them are that critical of Nielsen's ideas or challenged some of his assumptions.
Let me mention that I'm a professional webmaster who has been at it for 5 years. I first experienced the web in 1993 at a demo that Marc Andreeson gave at NCSA for a group of librarians. I've also been through one day-long human factors workshop with Ben Schneiderman. Yes, I've read Tufte too. So, I constantly strive for web designs that are user friendly, logical, and attractive.
I'm in the middle of Nielsen's new book, which I'm quite enjoying. He has plenty of sage advice, which I wish that more web designers would follow. However, I think some of his advice is rather fascist, if you really look at it. I think he really downplays the importance of graphics on websites, which is the main reason why the web is popular in the first place. Sure, content is king, but surfers want to experience attractive websites. I understand the importance of lean pages, but I think that Nielsen overemphasizes this too much--this design need will become less important as bandwidth increases.
Nielsen's advice, while applicable to all web designers, tends to encourage to creation of look-alike e-commerce websites. The needs of a user buying stuff at E-Bay are different than a person checking out an online art gallery.
We should also face the fact that the web inherently encourages the DIY (do-it-yourself) ethic--I often compare the web to the zine world translated online. Yes, it would be nice if every website put their logo in the upper left hand corner, but at what point does this become elitism? Ok, user studies show that people prefer this, but are these simply consumers, or an audience used to a different language. You can put together a nicely done magazine, but you have no right to complain about a free form punk zine that was assembled with rubber cement and scissors. Punks are used to a certain vocabulary; slick DTP don't cut it with them. The same goes for web design--the expectations for a particular website isn't the same for each audience.
Finally, I want to address one of Nielsen's statements.
"What do all the big sites have in common? Minimalist design."
This is an illogical argument. What the big sites have in common is that most of them got started on the web at an early stage. Think of Yahoo or any of the search engines. E-Bay was one of the first auction sites. They had time to build name recognition as popular websites. In Yahoo's case, it benefited from incredible word of mouth promotion, because nobody had bothered to do a large directory of the web (the librarians missed a big opportunity here).
The "big" sites maintain this advantage now through massive amounts of advertising. That's what big sites have in common these days. Yep, they all look like portals, but that's a fad thing, not a design consideration. In fact, since so many of these sites look alike, new users have a hard time telling them apart. They've lost their personality, because they've become big, bloated with IPO money for advertising, and are designed by marketing departments. Why is Slashdot popular? Because it still has people behind the website, like Hemos and Commander Taco.
With that being said, I really recommend that every web designer have their boss buy Nielsen's book for them. Despite my criticisms, this book is very helpful.
makhnolives
http://www.infoshop.org/
Re:Minimalist design (Score:5)
One problem is that developers think that bandwidth will increase at the rate of Moore's Law, when it has generally taken about twice as long. In terms of consumer modems, these are rough numbers of bits/sec (please criticize my dates) of acceptance by about 25% of the population:
- 1982: 1200
- 1986: 2400
- 1991: 9600
- 1992: 14400
- 1996: 28800
- 1998: 50000 (nobody gets true 56k)
- 2000: still 50000 baud
If Moore's law were in effect, by my back-of-the-envelope calculations, we'd all be at three times T1 speeds by now. And I haven't even gotten into the requirements for the backbones, or compression, or latency. And yes, I know 5% of the audience out there is at highly incredible cable speeds. Meanwhile, outlying areas are going to get nothing better than 144K DSL for years to come.Essence of goo UI (Score:5)
In the old days, an operating system was designed to optimize the utilization of the computer's resources. In the future, its main goal will be to optimize the user's time.
Appliance, software, and web designers need to write J.K.'s qoute down and put it up on their monitor, their fridge, and their bathroom mirror. Write it on your hands. Do something, but keep this in mind all the time whenver taking useability issues in hand.
Usability: It's a Good Thing (Score:5)
Wow, Rob, that "overhype just like a real local newscaster" class is really paying off. What's up for next week? "What THEY Don't Want You To Know About HTML"?
"...so if people use it, it must be because it is good."
Let's not confuse "good UI" from "monopoly stranglehold". Macs and Win95 perform essentially the same function, so the only real difference is the UI. Which has the better UI and which has the marketshare? Slashdot may or may not have a good UI, but it certainly enjoys a monopoly position. I'm sure I'm going to hear a bunch of whining about this claim. Tell me, which 3 "Geek news" sites have print ads in glossy mags? 1) Slashdot 2) Nobody 3) Nobody else
"Not seeing something during initial use of the system would result in better use of the hidden features later.
I agree 100% with this statement, but I implore everyone to read it carefully. Note the phrase "initial use". That means that you SHOULD be able to use advanced features LATER. Windows gets this wrong by (correctly) hiding the advanced stuff, but then failing to reveal it later. "Raw" Linux gets this wrong by failing to hide in the first place. GNOME/KDE get this wrong by failing to provide a transition from "beginner" to "advanced".
--
Here is the result of your Slashdot Purity Test.
Attributed Filesystems (Score:5)
5) Revolutionary UNIX GUIs
All respect for info appliances, but we also need a workstation-style interface that can help knowledge workers survive the information flood of modern society. And that's where I think we really need revolutionary designs that go beyond the Mac. For example, ways of managing tens of thousands of documents by a rich set of attributes and content-oriented navigation. Simply showing files as icons in folders doesn't cut it beyond a few hundred.
We also know from many studies that the average user is very bad at hierarchical filing and typically never moves a file once it gets to live in some directory. Even if the file would be better off elsewhere. This problem is magnified several hundred times when it comes to managing email. I am starting to think that the solution is to treat information objects as members of a soup and manage them by attributes rather than by hierarchy and name.
With regards to the above answer, how does everyone here feel about something like the additional (and user-defined) attributes in the BeOS [be.com] file system, where all the additional e-mail info (subject, header, etc. is contained in attributes attached to the main document.
Alternatively, what about a product like The Brain [natrificial.com] from Natrificial, which creates a linked 'web' of parent/child attributes to all documents in the FS? Do you find these products more or less usable than others, and are they the right road to be traveling down, or is another direction needed?
Hidden features and hierarchies (Score:5)
This is an interesting point. Most people seem to be very bad at forming a mental model of how software works. Of course, this is partly because there is a lot of software which does not have any consistent model, but still a lot of users will only do a few linear operations they have learnt, and have to be taught further operations, rather than grokking the whole program. And then you end up with horrible, horrible "wizards" as the only way of getting things done.
Starting with a simplified system would perhaps alleviate this. But how can this be done? Simply taking out half the menu options seems a bit crude. :-)
What would, I think, help, would be to take away the majority of those blasted button bars. Most Word users still have half their screens taken up with rows upon rows of toolbar buttons, most of which they never use, and which are only marginally faster than going to the menu. All these unexplained options must be terribly intimidating to new users.
I'm a great fan of hierarchies. Many users are indeed bad at managing them, but I think this is as much to do with bad interfaces onto them as anything else. New users are always confused as to where files are, not because they don't understand the concept of a hierarchy, but because their applications give them many views of the same hierarchy that all look completely different.
We've got the filer (with its view options that Windows by default changes pseudo-randomly), common file dialogues (which only show files of the type their application is associated with, thus losing any 'sense-of-place' the user may acquire), Explorer (which put directories and files in different panes, again destroying the comfortable feeling of seeing a directory and knowing exactly where you are), and uncommon file dialogues of various shapes and unpleasant flavours. Consistency: zero. Plus of course Windows and Unix both take over the root directory and fill it with stuff one doesn't understand, and applications happily open up unlikely directories like root, a directory hidden inside the application's own domain, or - God help us - C:\WINDOWS\ as the default place to save/open a file, thus making the user totally lost.
Give the user one filer application for doing everything to do with hierarchies and lists - not just files, but, using a VFS, all hierarchical datatypes in applications - and I predict the world will be a happier place.
I'm quite interested in Jacob's idea of an information soup, because I'm currently working on an interactive web log analyser, and its data is inherently both multiple-hierarchical (eg. file request: /dir/dir/file, client host: /com/altavista/spider653, and so on) and loaded with other properties, which one may wish to sort and view in many different ways. The interface is still quite hierarchical in nature, because I really can't think of a better way to structure the data. An infosoup is a great idea but you can't expect the user to issue what amounts to a database request every time they want to edit a document or something.
Anyone got any good examples of interfaces for property-rich data without a strict single hierarchy?
--
This comment was brought to you by And Clover.
Know *when* to use graphics, please (Score:5)
That said, judicious use of graphics is a good thing. I'm shopping around for a PC case, and it is so fucking frustrating to see one little .jpg of the (closed) outside of a case. I want to see shots of the outside of the case, the inside of the case, the stuff inside it coming off, and everything else! Even line art would be okay.
That said, don't make them all huge bloated SOBs that take forever to load. (Yes, I want pictures, but I'm on dialup, too.) Use thumbnails to link to larger pictures. The "$50/month ecommerce" pages at store.yahoo.com are particularly egregious in this respect. The graphics are 40-50k graphics that load nice and fast but are usually just one shot of the outside. They are links, but all they link to is the exact same picture off of the yimg.yahoo.com server.
That being said, use some damn text, too! Too many times, the description reads something like "$BRANDNAME computer case, $75". WTF good does that do me? I can see it's a computer case! I want to see something like the following:
$BRANDNAME mid-tower case
When will these people learn? That last thing we want (especially if they're selling cases, the tech-savvy crowd is who is going to buy this stuff, mostly) is fluff! We want to know what we're getting, not just one crappy jpeg and a one-line text description. Sure, you can always email for more info, but personally, I don't want to wait for that. I want to find out everything now and not wait on someone else's schedule. Besides, I don't want to give my email address to you, a business I've never heard of; if I want to get "product updates" from you, I'll give you my email address with my explicit permission for that.
Yes, a picture is worth a thousand words, or so it is said, but the person who came up with that never shopped online.
(/rant mode)
OT: Does anyone know any sites where I might actually find cases such as this? Sites that sell cases are few and far between. I've also tried to find just thumbscrews, but the only place I've found so far is pcpowercooling.com [pcpowercooling.com]. Even better if there are any Philadelphia geeks out there would be a local store that sells them. (I thought that CompUSA did, but they were just the dinky, crappy little plastic kind.) My email address is bj.XYZ@ZY.netaxs.com, and you should be able figure out what to do with it. If not... :)
Time to submit this, I think. I've probably gone grossly over quota on this posting, :) but I hope this helps some budding web designer. Either that, or I'm preaching to the choir.