Web 2.0 on a Desktop (This 2.0 reference will only work once)
April 15, 2006 9:26 PM   Subscribe

The Symphony OS. Microsoft may have tired something like this before, but it certainly doesn't look as pretty. This beta linux distro is sporting something called the Mezzo Desktop environment. Under its hood resides a http server and a Mozilla renderer which allows you to run any applications coded with HTML, Javascript and PERL. Think small "Web 2.0" applications on your own computer or more Dashboard widgets for tracking your FedEx packages. You might not get your extended family on to the alterative OS bandwagon, but a different conceptualization of the desktop never hurts. Instructions on how to get it running on Ubuntu are here.
posted by phyrewerx (40 comments total)
I don't get it.

Could someone please come and make some informed comments so I can understand the point of this?
posted by pompomtom at 9:57 PM on April 15, 2006

See older post on Swine Penis for clarification.
posted by Mr Bluesky at 10:01 PM on April 15, 2006

"Hey! Now that client-side Javascript is almost sort of reliable, and XML-RPC lets us eliminate one of HTTP's many inefficiencies, let's just code all of our desktop utilities as web-apps, and force the user to run a localhost web server! It'll rock!"


I'm all in favor of putting a server-in-every-closet, but this seems like a head-scratcher.
posted by verb at 10:03 PM on April 15, 2006

Yes, let's totally remove any benefit provided by better languages/compilers/environments, so we can kludge everything into a web browser. Brilliant. Why not write the entire OS in Javascript, with CPU cache requests sent via HTTP?
posted by krunk at 10:09 PM on April 15, 2006

We're still thinking so differently we're going to rip off Apples ad campaign.

As for the concept, I gota go with verb, meh..

This isn't the next big thing.
posted by doctor_negative at 10:10 PM on April 15, 2006

I can't comment with any authority, but as a heavy computer user who uses internet applications for about 90% of all total use, it does seem like there would be room for a lightweight computer which would run a web-only OS. A small laptop or tablet which was used to check email, web applications, upload photos and edit them online, etc. I think a lot of people could use that and in theory get excellent performance with a minimum of processing power.
posted by cell divide at 10:16 PM on April 15, 2006

I think a lot of people could use that and in theory get oexcellent performance with a minimum of processing power.

But, it won't really be lightweight, will it? It's still Javascript + Mozilla + Apache + Perl + X + Linux. Now days, particularly with all the pretty transperancy they've got in their UI, that would be pretty sluggish on a low-end system. I mean, imagine the processing power you would be using to check your email on that technology, compared to a basic native email application.
posted by Jimbob at 10:24 PM on April 15, 2006

Why is Digg blue today?
posted by ninjew at 10:43 PM on April 15, 2006

The new breed of web based applications like Gmail and Flickr, commonly referred to as "Web 2.0" applications are changing the way people manage their files and data. Symphony OS makes it easier than ever to use these new applications in your everyday computer needs while still having programs that are integrated with your computer.

Okay, I'm not a programmer or anything, so correct me if I'm wrong here, but don't you just need a browser to use those apps? How much harder can it be? Also, how is Gmail "2.0?" Admittedly, my definition of that term is rather vague.
posted by brundlefly at 10:47 PM on April 15, 2006

*"easier," not "harder"
posted by brundlefly at 10:48 PM on April 15, 2006

In my defense, I didn't see it on Digg. The 2.0 thing might not be the best angle to talk about the os, but the decision to get rid of context menus and making the os one giant menu is still kind of cool. (albeit a throwback from the whole multitasking idea)
posted by phyrewerx at 10:53 PM on April 15, 2006

Gmail is "2.0" because it uses Asynchronous Javascript And XML. Things happen without you having to load a new page, and generally pretty quickly. Anything that uses AJAX is pretty much "2.0", I guess.

Anyway. All this hubbub and they can't even antialias fonts? It looks ugly and useless. Linux with a weird front end, okay, thanks guys.

Jimbob's completely right — web-only computers can be much, much lighter-weight than this would be, and would probably look better anyway.

"We're still thinking different." Yeah, uh, I'm still thinking different, too. Next.
posted by blacklite at 10:55 PM on April 15, 2006

It actually looks rather ugly to me. Like a myspace page or something. How the hell are you supposed to read black text on dark blue cloud pictures?

I mean really, just looking at the screenshots, it's laughable. A bunch of ugly semitransparent stickies on a background. It's like the Windows 3.1 interface ported to AJAX. Now we can use really slow network agnostic gui technologies and scripting languages to do our whole interface. HTTP+javascript+DHTML. Um ok.

The accomplish the exact same thing as the people who came up with X-windows 20 years ago.

Except now you can take a standard Linux distro, change a few config files, and call it an "OS".

posted by delmoi at 11:00 PM on April 15, 2006

Their laws of UI are a good start, but I think they're increasing superficial usability at the cost of real customizability.

I don't have such a hard time with Windows as every one else in the world seems to. What I want is more stability, less bloat, easy multiboot, and a more intuitive file system.
posted by BlackLeotardFront at 11:32 PM on April 15, 2006

the technology is meh, but i enjoyed the laws of UI design. I didn't much care for his rejection of customizability.
posted by Tryptophan-5ht at 11:42 PM on April 15, 2006

Just installed it on my Debian linux box and it hung and wouldn't start on login... whatever it is needs a bit more refinement I guess.
posted by zog at 11:49 PM on April 15, 2006

the technology is meh, but i enjoyed the laws of UI design. I didn't much care for his rejection of customizability.

So we all get to look at that rediculous cloud pic?

Real usability means it works like you expect it too. People these days expect their computers to work like a windows machine/mac (which pretty much work the same – hell, the doc is just a prettier taskbar).

Basically that means:

1) a start-menu to start programs or other tasks from
2) a finder-style desktop and filesystem browser (i.e. finder/explorer program where the desktop works like a folder)
3) a taskbar/doc where you can switch between running programs.

It's not a "metaphore" for anything anymore, it's more like a Baudrillarden simulacrum. A copy of a copy of a copy.

And really, it works fine. What people don't realize is that you're not writing UIs for people who have never seen a computer before, you're writing UIs for hardened computer users. They don't want to change, or learn anything new; they just want to get stuff done.

Work needs to be done to ease common tasks like using email, writing documents, etc.
posted by delmoi at 12:30 AM on April 16, 2006

I think the concept is very interesting. A Pentium 3 or equivalent should be able to power a reasonably advanced desktop interface built on Linux + httpd + mozilla + javascript. Maybe nothing as elegant as OS X, but comparable to Gnome or KDE, or even Windows XP. However, the Symphony project seems to be a long way from producing anything like any of those. The screenshots are mockups. They're pretty and intriguing, but one would have to see the system in action before taking it too seriously. Also, there are major security concerns with a browser based desktop approach. I don't think Microsoft or Apple need to worry about the threat from Symphony any time soon.

One of the interesting promises of a browser based desktop OS is to run as a thin client with files and settings saved to a remote server rather than the local machine. This is of particular interest to corporate or institutional IT settings where people share computers. If Symphony OS can achieve it's objectives, it could be very interesting. I hope the project does well, but it looks like they have a lot of work to do to before Symphony is stable, secure and polished enough to compete with existing desktops.
posted by Loudmax at 12:43 AM on April 16, 2006

Does anyone know where I can get the desktop pictures this OS comes with, without downloading the entire ISO?
posted by Mikey-San at 1:20 AM on April 16, 2006

Loudmax nails it. It looks to me rather like a kiosk setup; except instead of taking a fullfeatured Desktop Environment (kde, gnome, windows, OSX) and restricting it down, you start with a basic featureset from the get go.

Assuming you've got a low-spec piece of hardware, you can already run a very light windowmanager + X, but it's not the most pretty or friendly user experience. Since this is browser based, you could easly farm out the app heavy lifting to a server without needing the complexity and cost or platform-specificness of the various Terminal Server options out there - and still have a friendly easy-to-use collection of web-based apps.

Smaller network load, smaller server load, smaller client load. Nowhere near as flexible as compiled apps running on a fully-featured desktop environment, but for say an internet cafe or generic corporate desktop running internal network based apps, could be quite interesting.
posted by ArkhanJG at 4:25 AM on April 16, 2006

I gave the desktop environment a spin (on Ubuntu), and will be promptly removing it. It's okay, I guess. But it's really slow compared with Gnome. And my computer is old. Slow is bad.
posted by TheFarSeid at 5:14 AM on April 16, 2006

But does it run RoR (Ruby on Rails)?

If the kids from 37signals aren't involved, it can't be that cool.
posted by Mick at 5:27 AM on April 16, 2006

I looked at it, thought, "Bill Gates' desktop!" and then my brain shut off.
posted by thekilgore at 5:56 AM on April 16, 2006

I actually like the idea of this, but the parts I like are more or less what Apple is trying to do with the Dashboard and Applets. I haven't dug into that yet, but my understanding was that they had worked with Mozilla, Operasoft and the KDE folks to define the tagsets and standards in such a way that you could make applets portable across OSs. Agreed, it adds a layer of structure and it's not efficient; but it's got a mush shallower learning curve than traditional tools.

But these "laws of interface design" are arrogant and ill-considered at best, and smack of narcissistic, word-check ideas of usability. It's just not very well thought through, though it's certainly passionate. Let's go case by case.

Nested menus are evil. A good user interface will eliminate nested menus since humans have a hard time targeting menus in the first place, let alone panning up, then scrubbing to the right or left in a 20 pixel wide corridor.

False. Usability experiments going back to at least the early 80s (I read the papers in the early 90s) have demonstrated that "wide and shallow" decision trees tend to significantly decrease usability. As do "deep and narrow" decision trees.

And anytime that anyone puts "x [are/is] evil", they lose me. (For that matter, "evil": Could we just put a moratorium on the use of the term "evil" in technical or technically-oriented discourse? "Evil" is irrelevant to utility.)

Scrolling sucks. A good user interface will minimize scrolling, and encourage the user to create volumes of information that do not promote scrolling.

Mezzo Solution: Auto resize icons and information. .... When more items are visible, they shrink in size, and/or offer less information. I coupled this with the desktop Target Menus themselves replacing much of the need to launch a file manager or other 'scoll-happy' interface driven program.

Meh. Probably not a good idea -- people tend to get confused when the interface changes in front of them without action on their part. And it also sounds (to these 41-year-old eyes) like a recipe for eyestrain. This would give me a headache. Users should, at the very least, be able to turn this behavior off.

As for the bias against "file managers", that's another interesting tell. Sounds like he's an advocate of the ever-elusive "document-centric" paradigm. But that's a whole 'nother discussion...

Drill down interfaces are evil. ....

Mezzo Solution: The entire screen is used to present the Target Menus under each type of target.

Microsoft Excel for Windows introduced this in the early '90s, and it's a feature in every modern operating environment I'm aware of. Unless he's talking about soemthing qualitatively different from a context menu.

Configuration gluttony must be stopped. Thank god for open source. There comes a point at when the UI is hidden, drilled, stacked and nested so much, just because there is a knob or button for everything. This has to stop. UI is about making decisions to help the user, not about weaving a rope to hang themselves with, or smoke when they get frustrated.

[What does this even mean, other than being a rephrasing of previous statements? But I digress....]

There are things you can't configure in the Mezzo interface. Like the position of the close window button, or the window décor. If that makes you unhappy, then there's other alternatives that can be easily created or downloaded and you can move the window decore until you are thoroughly confused to your heart's delight. ... Users should not be burdened with over-configurability. Instead their computer should be like their TV and just work, with a few small adjustments if necessary.

Actually, this one I don't have a problem with, except that again, I get this whiff of scripture from some of the phrasing. Again, usability research demonstrated back in the 80s that users wanted more configurability as they improved their mastery of the software tool. There have been more or less interesting attempts to work to that. "More" including things as diverse as the idea of having a common file format across applications with widely varying capabilities and complexity; for example, you might use a lightweight piece of software like AbiWrite to write and output to RTF, and something more complex to assemble and format the parts. Those tasks might be done by more than one person. On the less-interesting side, consider Microsoft's randomly reshuffled dynamically prioritized menus, which do nothing but slow people down.[1]

On the other hand, some of his language here betrays a contempt for users and a basic misunderstanding of the difference between computers and a TV.

[1] This is also a good example of slavish devotion to the instrumentally measurable metric. The dynamically-priortized menus in MS apps put the most-used choices at the top of the menu; by definition, then, those were the ones that could be reached most quickly. It does not follow that they were the easiest target to acquire. Since the position of elements on that menu is effectively random (from the perspective of the user), there's an extra cognitive load to the acquisition step. Because we think so fast, it's entirely possible (and in the context, I would say likely) that the extra cognitive load will not register on the instrumental metrics.

Consistency is worth more than multiple placement. People are trained monkeys, but they also make associations that help them learn faster. If you group like-oriented activities in a single place, the user can trust the interface to find similar stuff in that place. However, if, like Windows, you put access to a function in several disparate locations in an effort to hopefully “be there” when the user is looking, the UI defeats itself. You confuse the user more by giving them no logical location-association for a given type of action. The cost out weighs the benefit.

Here he fights against himself. He seems to want to make tings easier for users, but again he has contempt for them -- we're "trained monkeys", we don't think, we only associate. We're conditioned. We don't think. (He does, of course, though -- he thinks for us. He's here to save us from ourselves, I guess.)

He's also got some weird ideas about how Windows works. I'm trying to think of cases like what he describes (putting 'access to a function in several disparate locations in an effort to hopefully “be there” when the user is looking') and I'm hard put to come up with one. Not to say they don't exist; but I can think of an alternate explanation for something that might look like what he describes. For example, I might put access to the "print" functionality in a preview pane and in a menu; I might put it in a right-click context menu inside a document, as well as in a right-click context menu in a file manager. To me, that looks like putting the function where it will be wanted. If we take this manifesto at its word, that's a bad thing, because we're not 'grouping like functions'.

The drag-and-drop desktop and its icons are the junk-drawer of the modern computer and should be eliminated. ... [D]esktops quickly become cluttered without user intervention. Making it harder on the user, every file-dialog encourages saving to the desktop by offering the confusing choice of “Desktop” along with Documents, or Pictures. If we had a nickle for every time a user saved something to their desktop and wondered where it went, we'd be rich people. If users need to save a picture of the web or grab a URL, then the browser has wonderful facilities for doing that; Bookmarks and Save Image As. Popping things up onto the desktop like CDs or Hard Drives or copying program shortcuts there leaves the door open to the user a)missing what happened, or b) losing the UI elements they've come to depend on in a see of surprise clutter.

Mezzo Solution: No traditional desktop. There is no DnD desktop, no magically appearing devices, and no copying to it by rabid installers. Instead the desktop has useful information (Desklets) that stays there until specifically moved by the user. We have an advantage because most of the world instinctively mouses to the lower-left corner for their programs menu. The absence of desktop icons should take half a second to overcome for most people. [emph added]

Again, actually, I agree with much of this. But he doesn't say what he replaces it with. Users need a way to store and access their documents. I'm not seeing him having a philosophically consistent way of dealing with that. He seems to be handwaving on this, implying that somehow you get your documents by mousing over the lower-left Fitt-target (er, I mean, corner); since he seems to think that wide and shallow decision trees are the holy grail of UID, I'm not optimistic.

And again, he's fighting himself. On the one hand, he has contempt for the intelligence of users -- or at least, for the idea that there's conscious cognitive load involved in using the UI -- and yet here, he's expressing great faith that users "should" adapt (maybe it's a moral judgement?) in abotu "half a second". I guess it's all instrumental condtioning...

Pop up dialogs and ballons are a horrible interface tool. Popup dialogs demand attention, often at the most inconvenient time, no one reads them, which is dangerous, and having to read them for important information or to take a desired action interrupts the users flow of work, and should be minimized if not discarded all together. Thank heaven for Firefox and the thoughtful integration of the Find Dialog, tabs and “above-content” messages embedded in the UI. Balloons are simply popup devils that are smaller and more area specific. They are evil too, just in a different way, since balloon disappear entirely if not touched and vanish, leaving the user to wonder where the important information that needed their attention so badly that it had to be popped up at them went.

Mezzo Solution: The desklets and Target menu desktop interfaces serve to provide the user with information that they can observe and are constantly updated with relevant and new information. For example, if someone send me an instant message, the desklet can show the message and even change color, but it won't popup in your face, demanding attention and a mouse click.

Again: What's with the "evil"?

Seriously, this one is just dumb. As described, it's just not an issue anymore, except with badly-designed applications. And he's just wrong about "balloons", at least if they're used in a considered manner. And his solutions don't address some of the problems he sees with pop-ups (e.g., "no one reads them" -- they're even less likely to get that information now, dude).

Additionally: Translucent information elements are antagonistic to usability. The extraneous data displayed behind the image is distracting. And the slavish attendance to Fittism means they are likely to focus on usability changes that are instrumentally measurable, with a further bias toward those which are easily measurable -- e.g., measuring time to acquire a "target" at the edge of the screen -- at the expense of changes that would have less easily-measurable effects.

Which is not to say that putting menu targets in the corners of the screen is a bad idea. It's probably a good idea. But when they invoke "Fitt's Law" as a way to validate that design decision, I start to suspect that they're practicing usability more as a faith (and maybe as an art) than as a science. They're buzzword-checking; they're not really thinking it through, and certainly aren't performing valid tests.
posted by lodurr at 6:29 AM on April 16, 2006

shut up, please.

thank you.
posted by quonsar at 7:24 AM on April 16, 2006

I think it's somehow appropriate that his slideshow is hosted on mac.com - who's thinking different now?
posted by OverlappingElvis at 7:36 AM on April 16, 2006

I don't understand his hatred of nested menus. Since the file system is organized pretty much the same way it makes sense to organize the UI that way. It makes sense to me, anyway, and that's all that counts, right?
posted by mr_crash_davis at 7:53 AM on April 16, 2006

Awww, did I make you scroll? So sorry. I forgot that scrolling was evil.
posted by lodurr at 7:55 AM on April 16, 2006

I think his thing about nested menus may be a lapse in communication. Because as I re-examined his "desktop-wide menus", they're realy not menus at all -- they're directory listings. What he's really doing is replacing cascading menus with buckets of like things.

That works alright if you don't have a lot of things you need to do. But I think he'll find it breaks down as he has to map it to more complex applications.

Though it might be interesting to see it applied to Photoshop....
posted by lodurr at 8:00 AM on April 16, 2006

no Digg.
posted by MarkO at 8:11 AM on April 16, 2006

mod parent up
posted by jepler at 10:21 AM on April 16, 2006

Assuming you've got a low-spec piece of hardware, you can already run a very light windowmanager + X, but it's not the most pretty or friendly user experience. Since this is browser based, you could easly farm out the app heavy lifting to a server without needing the complexity and cost or platform-specificness of the various Terminal Server options out there - and still have a friendly easy-to-use collection of web-based apps.

Bleh, people didn't seem to have any problem with X-windows back in the day, on much lower powered stuff. And X-windows isn't platform spesific, either.
posted by delmoi at 10:25 AM on April 16, 2006

From the "Laws of UI design" page:

I actually couldn't find the 'laws of UI design' on the website.


2.Nested menus are evil. A good user interface will eliminate nested menus since humans have a hard time targeting menus in the first place, let alone panning up, then scrubbing to the right or left in a 20 pixel wide corridor.

Who are these "humans" he's talking about? People are very good at sophistocated hand motions. I don't have any trouble with menus, nor do I know anyone who's ever complained about them in the past.

I've probably clicked hundreds of thousands of menus in my life, missing maybe a few tenths of a percent or less.
posted by delmoi at 10:45 AM on April 16, 2006

I'm... having... a... flashback...
posted by smallerdemon at 10:48 AM on April 16, 2006

Bleh, people didn't seem to have any problem with X-windows back in the day, on much lower powered stuff. And X-windows isn't platform spesific, either.

I cut my teeth on Solaris back in the day, and linux is my platform of choice now, but there's a much bigger pool of people using computers these days. Computers are not just for the techies any more. Yes, some people still prefer something like blackbox, but to be blunt, your average user wouldn't. They want an interface that's simple, pretty and quick, since their area of expertise is somewhere else entirely, and more importantly, they don't want to learn about their computer. It's just a tool, not a learning experience in its own right, unlike 'back in the day'.

X11 isn't platform specific, but the apps that use it often are, and many apps don't run on X at all. Java is one way round that problem, as are AJAX browser apps. This idea goes the AJAX route. Suitable for everything? of course not. Is this project going to be a wonderful implementation? I have my doubts. Still an interesting approach though.

Who are these "humans" he's talking about? People are very good at sophistocated hand motions. I don't have any trouble with menus, nor do I know anyone who's ever complained about them in the past.

You obviously don't work in tech support. I help people daily that get lost in MS Office's menus, and have trouble with the concept of 'right-click'. The internet/browser/local app distinction is equally a lost cause. It's not just the hand-eye co-ordination, it's finding what you're looking for. As I've said, most users these days don't know much about computers, and don't want to know. We can either ignore them, or try to provide solutions that work for them as well as the geeks. Apple seem to be doing quite well building systems that aren't as painful to use as the competition, perhaps there's something in it?
posted by ArkhanJG at 11:33 AM on April 16, 2006

Given that it took me about 4 clicks on a thumbnail just to finally see what the full-sized mockup photo looked like - and given that each click opened another damn browser popup window - I'm going to just start by assuming this guy knows fuck-all about designing a web-based interface and go from there. Meh indeed.
posted by caution live frogs at 11:53 AM on April 16, 2006

I would like to use this to help my Grandma use the internet.
posted by Baby_Balrog at 12:01 PM on April 16, 2006

I would like to be excited about it but that is probably because I'm having a such a bastard of a time with Gentoo/HAL/Gnome* at the moment that I'm considering putting XP back on the home desktop. I think this idea could be useful in a limited, web appliance-like environment but cannot scale to handle the way I, for one, use a desktop machine. Shallow menus would not work for me. A faceted file manager/system might be the approach that would facilitate something like Mezzo but it seems they're more about skinning Debian than about the heavy lifting required to put something like that together.

*I fully realize that this is due in no small part to me playing outside of my depth but it is still irritating nonetheless.
posted by Fezboy! at 1:50 PM on April 16, 2006

shut up, please.

That's not even amusing, q. One hopes you can do better than that!
posted by stavrosthewonderchicken at 4:19 PM on April 16, 2006

« Older what a bunch of dicks   |   Germany 2006: The ones to watch? Newer »

This thread has been archived and is closed to new comments