Slashdot Log In
The Next Computer Interface
Posted by
timothy
on Tue Nov 20, 2001 05:44 AM
from the pick-an-interface-any-interface dept.
from the pick-an-interface-any-interface dept.
BoarderPhreak was among the several readers who pointed out "an interesting article on the various alternatives to storing your files using a 'desktop' metaphor" at TechReview.com. "New styles like time-indexing, 3D sphere ala SGI's file manager, and even a 3D virtual 'task gallery' from Microsoft. Screenshots available in the article." All of these have been floating around for a while; hopefully soon some radically different interfaces will actually gain widespread acceptance.
This discussion has been archived.
No new comments can be posted.
The Next Computer Interface
|
Log In/Create an Account
| Top
| 336 comments
(Spill at 50!) | Index Only
| Search Discussion
The Fine Print: The following comments are owned by whoever posted them. We are not responsible for them in any way.
(1)
|
2
(1)
|
2
Great site for this stuff (Score:5, Interesting)
--LP
The users control the interface. (Score:3, Interesting)
What I Use For General Navigation Stuff (Score:4, Interesting)
Re:What I Use For General Navigation Stuff (Score:4, Informative)
After looking through the site, it reminds me a bit of Ted Nelson's ZigZag [xanadu.com], only with a much prettier user interface.
ZigZag basically lets you set up arbitrary "axes" of meaning and drop nodes on them. Any node can contain anything, and be a member of any number of axes. All axes are orthogonal to all other axes. The user interface lets you move along any axis from any node. Thus, information is locally coherent but, if you step back, it's a rat's nest.
For example, for organizing things on your computer, you might create an axis named "Games," and link Quake, Starcraft, and Solitaire to it. Solitaire is published by Micros~1, so you might also set up a Micros~1 axis, which contains Solitaire, IE, Word, Excel, Outlook, etc. Solitaire would be a member of both "Games" and "Micros~1", but not of the "Network-aware" axis, which would contain Quake, Starcraft, IE, Outlook, etc.
ZigZag is very primitive right now, but the concept is very intriguing. Written in Perl and runs under Linux. Check it out.
Schwab
3D, voice and why its NOT a good idea... (Score:5, Insightful)
All of these gimmicks tend to miss out on the fact that a simple linear system is much better for _people_ than the fancy gimmicks which developers think are cool. Voice interaction is a classic example of something that can be thought of as "cool" until you have an open plan office with 30 people talking at their computers.
3D is another dead end. IBM's Home project found that people would "lose" things in a 3D environment and in fact the visual cues of the 2D desktop were better suited to the task.
At the end of the day the mantra should be KISS. These break that mantra and add very little except cool graphics. It looks nice but doesn't function well. An everyday example of why simple is better are the icons used to denote things like "radiation", "poison" etc etc they don't actually represent the thing themselves but provide a simple shorthand for the thing. This simplification makes them much better at describing and classifying than attempting a "realistic" presentation.
Good examples of 2D simple interfaces are things like Google. Why would 3D make Google better ? It wouldn't.
Pretty != better. More Gimmicks != simpler
KISS
VI (Score:5, Interesting)
I agree about offices, the technology to make voice interfaces work is here today, but the applications is not, however but Voice Interfaces offers a lot of potential for much more personal environments, like the car home & SOHO.
VI offers a number of advantages over conventional interfaces, biomentric security, easy of use & accessability, even for your technophobic mother/granny.
Imagine a home entertainment gateway accessed by voice, no worries about little Johnny snooping your adult PIN. The inherent Biometric security, will make no difference, if he overhears your PIN.
Imagine re-tuning you IP Radio Alarm, by voice, without opening your eyes.
Imagine switching off your security alarm, by saying 'Hello House', and then following up it up with the query "Messages?" without having to log in and remember your password.
Or change channels without having to figure out which of those six seperate remote you need to use, simply by saying 'TV, select channel 4', or 'TV, News' or any number of other scenarios.
I think the killer application for VI is Home Automation.
Voice recognition in the office (Score:4, Insightful)
I don't understand slashdotters sometimes.
Have you ever been in a callcentre? Okay, they're generally not exactly open plan, but they don't give the drones huge walls, and there's way more than 30 of them in there.
What they do give the drones are headset mics.
This isn't rocket science, folks. Kate Bush, not exactly a hardcore techie, came up with wireless mics in the '70s (well okay she forced her engineer boyfriend to come up with them, but you know how it goes :)
Wireless headsets. That's how you do voice recognition in an office.
the thing is...... (Score:3, Insightful)
a) a tried and tested system which works, and is already fairly* well established in the minds of billions of well-organised people, and was evolved over hundreds of years of trial and error by people who actually NEEDED to organise stuff
b) outmoded and ready for the trash heap.
take your pick.
I applaud the effort to find something better, but really, i think "natural selection" would have found a better real-world parallel if it existed.
relationships between files in the structure is a brilliant idea, but that's just metadata and cross-referencing.
- says the man with the cluttered desk - at least my machines don't have virtual beer bottles leaving ringmarks on my HTML documents.
*irony
Getting seasick? (Score:3, Interesting)
Some ideas (Score:3, Interesting)
One of the main problems today is that the OS doesn't make the best use of the information available to it. The OS can know the file type, the application(s) associated to those files, when the file was created etc. but in general, it doesn't do much with that information. Sure, if you double-click on your file, it can find & launch the application associated to the file type, but you're still left with the problem of finding that file.
My own proposal would be to make better use of the file information that the OS has available to it. Its theory, but basically you place a database layer over the filesystem. We should also make use of MIME types for each file, and create a hierachical directory structure, one for each MIME type inside the users "home" directory. As a simple example, you may have something like:
user1/
files/
image/
jp eg/
pn g/
x- bmp/
audio/
x- mp3/
Now we have this, we can put the information we have to good use. Whenever a file is created, rather than asking the user for a directory and a filename, we ask them for a description of the file. Create the file in whichever directory suits the files MIME type, with a system generated filename, and add an entry for the file into the database which is layered above the FS. The record should include the users long description, type, creation date etc.
Now when the user wants to find & open a file, they can easily find their file by e.g the decription, using a wildcard if they wish. Or the creation date, using a range is they want to. The major advantage is that they don't need to navigate through a heirachical directory structure, nor do they need to remember what type of file their looking for, as the OS can present all of the files that the user can open as a flat list in the dialog.
O.K, it's a clumsy description, but the basic premise is that a) The OS can handle placing the file on the FS instead of forcing the user to decide & b) We know have a flat list of files to manage, instead of a possibly very complex heirachical tree. We do retain the advantages of the hierachical tree for the filesystem, however.
Re:Some ideas (Score:4, Interesting)
1. Computers organise things hierachically because we like it that way. We haven't been conditioned into it by computers. Remember the game where you pick a number between 1 and 100 and somebody tells you higher or lower? Remember how much easier it was when you learned to start in the middle? It's the nature of things.
2. 98% of the time, you don't need to "search", if you organise things properly. No matter how great the search is, it's always going to be more efficient if you stick things in relevant folders (/documents/biz/2000-2001/invoice110.doc)
Personally, I think the biggest problem with the wimp interface is modal windows (this should not happen ever ever ever amen), and a decent way to keep track of more than 10 open windows/programs.
This is dumb. (Score:3, Insightful)
These people should really learn that the desktop is the best way to handle a two dimensional computing environment. The only possible ways to make the current computer interface better is to either add new hardware to interact with the computer (3D goggles, hand sensors, microphone, what have you) or to tweak the current desktop interface to make it just a tiny bit better.
These people are trying to reinvent the wheel by making it a square or a triangle. My wheels are fine the way they are, thank you very much.
Not another one! (Score:4, Interesting)
And yet none of them have taken off. Why's that? Maybe (heretical thought!) it's because the current model actually works quite well for most people.
I don't want a system where the computer organises things for me. I can organise them better myself. (Occasionally I might lose something, but probably less often than if the computer was filing stuff for me. Anyway, we have good 'find' tools on Windows and Unix.)
I don't want a 3-D interface. It's much harder to visualise and navigate than a 2-D one. (A set of 2-D interfaces, as in Mozilla's tabbed browsing or many window managers' virtual desktops, is good. This is perhaps one of the real UI advances in recent years. Windows could do with virtual desktops.)
The article says: "Conceivably, an inference engine can be made so intelligent that [...] machines would automatically present information to you as you need it." Well, maybe when that's true I'll change my mind.
wrong concepts (Score:3, Interesting)
My point is that these attempts at deciding the future of GUI are pathetic as they don't even take the current GUI's limitation in consideration:
(Note: if you don't agree with the following then you had to adapt yourself to these. Take anybody who doesn't use computers and just observe him.)
Future GUI concepts should take the problem the reverse way:
I am actually working on a GUI concept which will *not* be 3D. This will be Open Sourced.
I dunno (Score:3, Insightful)
Anyways, I think the problem is this. We all have a shitload of files - on my 100gb drive, I have 85,120 files taking up about 80GB. This is my "extra storage" drive, I got 45,920 files on my C drive. Ok, perhaps I'm extreme, I have a shitload (about 40,000) of text files, books in pdf, etc etc..
Now, how the hell are you going to make it easy for me, or anyone, to access a good 120,000 files, preferably within less than 5 user interactions (clicks, speaking something, etc..)
OK, a new gui, cool, but if it is going to succeed it will essentially be a sytem based on "Organisational units that hold things" i.e. Folders / directories / objects.
A Chronological system won't work for a situation like this, it'd take too much of a mental effort -answer this - what did you have for dinner 1 week ago? I think that would be a perfect question, because things on the computer tend to be "routine" - did I work on this or this? It doesn't take a psyc major to tell you that humans suck at remembering what happened in the past.
With a hierarchical structure, it is painfully easy, and it scales well.
i.e.
e:\asdf (ok, ok, its easy to type and a throw-back to my 286 days, wee didnt have no stinkeen gooey)
e:\asdf\music
e:\asdf\music\Rock
e:\asdf\music\Rock\Prodigy - minefields.mp3
Moreover you can actually communicate the location to someone else, not "a file from sometime last week", or "the file in the Blue gallery on the right wall in the clipboard". Ever try to explain over the phone how to get to your house? Ever get lost becuase the directions sucked?
I can't argue with the article too much though - clippy the annoying mother fucker gets bashed on
Imagine this... (Score:4, Insightful)
Incidently, I am this }{ close to losing the GUI alltogethor. With the fantastic (but slightly unwiedly)mplayer [mplayerhq.hu], and Q3 now working from the CLI, I see little purpose (personally) for those quaint little GUIs.
A mouse is what you play Quake with.
Any ideas for a better 'Clippy' (Score:4, Interesting)
The article mentions how annoying Clippy is, but says that MS researchers still think a 'helpful' interface is a good idea if done properly. Can anybody think of a good way to do this without it becoming annoying?
One thing that I really hate about those little characters is that they get in the way and take control of the computer away from me. But what if a little box on the task bar showed the three 'most likely' things you wanted to do and you could activate them (complete with little wizards if the task is complex) by clicking?
Just one catch... (Score:3, Informative)
Although the idea of chronological storage is intriguing, and will work well for small groups of files, it breaks down completely once the user starts manipulating many files. This can be mitigated by storing them hierarchically, but this then ceases to offer much advantage over a desktop list view that's sorted reverse-chronologically by date.
3-D has two main disadvantages, both stemming from this notion of "space" as a way of managing files, as opposed to a flat "surface". The first is speed. Because there is more area (or, to be more accurate, volume) to navigate, the user has to spend more time looking for stuff. Second, as the article points out, things become easier to lose in 3-D space. You can alleviate some of this if you add the notion of "hallways" and "rooms" in which to organize things, but if you do, then you're still thinking in hierarchical terms, and that puts you right back on the Desktop.
Then, there's that funky sphere idea. Somewhat less of a problem than true 3-D, because you're still dealing with only one surface rather than a space Less easy to lose things. However, with all the spinning and zooming that you'd need to do, you lose speed, big time.
Microsoft's task-oriented stuff just doesn't work out. It's well-suited to carrying out actions, but not for organizing files. You just get dumped onto the Desktop.
It's true that the desktop metaphor has its flaws. In fact, truth be told, it's pretty bad. But it's like democracy in that regard: the only thing worse is everything else we've come up with so far.
The concept of "files" is the problem... (Score:3, Interesting)
When I write something in a notebook I don't have to "save" it, or give a special name etc.
For example, I've used a wordprocessor, called "YeahWrite", that does away with files. You simply open new pages and write. Everything is automatically saved and you pages are arranged in time order. This works great for people who are not computer expersts and are not interested in learning about computers.
In "The Humane Interface" Jef Raskin describes an interface that's based on plain text. There are no documents, just one big text stream that contains separators. The user interface just manipulates this text.
Finally, do these usibility experts actually watch people work? One of the most useful UI features is the idea of "Virtual Screens" (as implemented by Unix window managers). Each virtual screen keeps the context of a particular task and makes it easy for me to switch between them. Why hasn't this become a standard feature of Windows is beyond me!?
User interfaces can only go so far... (Score:4, Insightful)
This is especially obvious to anyone who's worked in teams of more than, ooh, one person who have had to share a single file structure. What one person perceives as a logical structure (/docs/reports/outgoing/date) another would view as being totally redundant (/docs/date/out/reports). You end up with a compromise that suits neither party, and by the time you move up to >100 people sharing a file structure you're in real trouble...
You also get into real trouble when a document has to exist in more than one place within the heirarchy. F'rinstance documents that need to be organised by Date or by Customer or by Author or by Cost code etc etc.
Shortcuts and/or logical links can help some of these problems, but they're both pretty messy solutions.
I have seen, and worked with, several database driven document management systems which show a lot of promise. Whether this is the way forward is a debatable point, certainly having to host a database complicates the implementation for the average desktop user.
Until some form of document management can be incorporated into the operating system all that a new GUI can do is to further obscure the core organisation.
What I want is a document management system which allows me to look at my files in the way that I choose, allows my co-workers to look at the same files in the way that they choose and hides the files completely from people who have no interest in them. The organisation of the files on disk shouldn't be something that I (as a user) have to even care about - slap them in a flat structure for all I care.
Fer [insert deity here] sake, if we were designing a file system from the ground up we wouldn't seriously contemplate a heirarchical model for more than five minutes. There must be a better way!
Cheers
Chris
Why? (Score:3, Insightful)
XYZ is dead (Score:4, Insightful)
Whenever I hear someone* declare something to be dead, it's a good indication that it'll be around for another hundred years or so. Yeah, the desktop metaphor is dead, just like paper is dead.
* Someone refers to the researcher who's inevitably researching what he thinks will supercede what he's declaring dead.
Allright, i'll weigh in on this one. (Score:3, Interesting)
1) What difference does it make how you represent a heirarchical filesystem and its contents? Zippo. Infact, organizing them by document and folder is probably the most condusive way to go, since most people arent like us. The rest of the world thinks in real world terms. Only programmers are accustomed to thinking about such things in highly abstract terms. Grandma shouldn't need to develop a mental picture of a binary search tree in order to find her cookie recipe. The desktop metaphor is boring in most implementations, yes, but its certainly not dead
2) 3D interfaces are rediculous. Take the screenshot that accompanies the article -- Three desktops are presented to the user in the form of a room, with a screen on each wall. What the hell difference does it make if they're on the walls? Youre STILL USING a flat, two dimensional surface to interact with! And so long as you're still using a flat, two dimensional space to interact with, representing them in 3D is pointless. Workspaces need to follow a design similar to channels on a television. You'll notice that your living room has one TV in it, capable of displaying hundreds of different workspaces. You don't have hundreds of TVs mounted all over your walls, each tuned to a different channel. 3D workspaces may have a future, but as a modus to display was essentially amounts to a 2D workspace floating in a 3D scene, they are beyond pointless. They're ridiculous. As in, its ridiculous to improve the design of UIs by "pulling a CueCat." You're inventing a tool to solve a problem that doesn't exist by pushing "2D in 3D" interfaces.
3) The 2D GUI isnt dead. It just needs refinement and rationality in its design. Speaking of irrational and unrefined ideas, take your common everyday scrollbar. You have a device (a mouse) capable of smoothly vectoring along a curved path, and communicating that movement to the computer. However, your damn UI still wants to alter your view of a workspace or document according to explicit X or Y axes. You can scroll up and down, OR, you can scroll left and right.. But never both, an act which would be far more intuitive to the common user. It takes fine adjustment of two separate widgets (a vertical and horizontal scrollbar) to accomplish a task that could be easilly encompassed within one...while wasting a disproportionately large amount of screen real estate in the process. So, rather than whine about it, I decided to do something about it a few years back.
Scrollbars are dead, and we killed them. Been working with someone for the past week or so on (finally) delivering a proof of concept model for the infamous "scrollball" whitepaper I released 3 years ago after InSight collapsed. The model looks fantastic so far (hi Dibos!) and will probably be dumped on Savannah or Freshmeat in a week or so once we fumigate the code to drive the last of the bugs out.
Cheers,
Information Overload (Score:5, Insightful)
No one has proposed eliminating my car radio in any meaningful way. In fact, during the dot com rush, the radio was supposed to be replaced by a satellite fed computer that would do essentially the same thing - stream content. Why change what something that already worked fine *without* a satellite?
I was also supposed to tank my televison for a computer that would play mp3s, surf the web, stream video, and cook my dinner. Why change that interface when all I want to do is watch "6 Feet Under" or "The Sopranos"?
I like the systems the way they operate now. If the researchers were to study how people conduct their daily lives, they might learn that humans use a variety of interfaces to gather information. To use the metaphor of Gelernter, these people seem to be armed with a hammer and view every information problem as a nail.