Why are we stuck with the old desktop metaphor?
HTML-код
- Опубликовано: 24 июл 2024
- Get the free Cybersecurity Mistakes ebook here: tuxcare.com/downloadables/top...
Grab a brand new laptop or desktop running Linux: www.tuxedocomputers.com/en#
👏 SUPPORT THE CHANNEL:
Get access to:
- a Daily Linux News show
- a weekly patroncast for more personal thoughts
- polls on the next topics I cover,
- your name in the credits
RUclips: / @thelinuxexp
Patreon: / thelinuxexperiment
Or, you can donate whatever you want:
paypal.me/thelinuxexp
Liberapay: liberapay.com/TheLinuxExperim...
👕 GET TLE MERCH
Support the channel AND get cool new gear: the-linux-experiment.creator-...
🎙️ LINUX AND OPEN SOURCE NEWS PODCAST:
Listen to the latest Linux and open source news, with more in depth coverage, and ad-free! podcast.thelinuxexp.com
🏆 FOLLOW ME ELSEWHERE:
Website: thelinuxexp.com
Mastodon: mastodon.social/web/@thelinuxEXP
Pixelfed: pixelfed.social/TLENick
PeerTube: tilvids.com/c/thelinuxexperim...
Discord: / discord
Timecodes
0:00 Intro
0:57 Sponsor: TuxCare
01:54 The Desktop
04:33 Launchers, app grids and start menus
08:03 The System Tray
10:06 Menubars and ribbons
12:54 Windows and titlebars
14:47 Parting Thoughts
15:31 Sponsor: Tuxedo Computers
16:25 Support the channel
If you think about productivity, you can't help but think that having the default state of your computer being an image with a few icons on it is less than stellar. For opening files, it will never be tidy enough to give you access to all you need, you need a launcher or a folder structure, meaning the desktop is bad at this. For opening apps, having visual shortcuts on the desktop is a duplicate of whatever panel or launcher you have.
Next are launchers: start menus, app grids, app lists, whatever else. If you think about it, these things aren't really efficient: you have to either scroll through a list of categories, click one, then click the app you want, or you have to scroll an even longer list of programs sorted by name, which makes no sense since most app names have no relation to what they do.
Most of these menus or grids thus added favourites, so you can place your preferred apps in a special place, but in a small menu, you'll never have enough room. This inefficiency is also why most app grids and menus implement a text field that's active by default, so you can type the name of the app you're looking for. Which also is not that great, because it requires you to know the name of the app.
Next is the good old system tray: this little area is by far the thing I despise the most in any operating system. Not because it's useless, because it's not, but because it's completely inefficient. You start apps from another place, and some of them will stay in the tray, some won't. Sometimes the icon disappears when you close the app's window, sometimes it won't. Sometimes the icon is coloured, sometimes it's not. If you have too many things in there, it tends to fold so you have an extra click to view what's running. Click targets are small and imprecise. It mixes apps and system features. Some apps will react to a right click, some will open a window with a left click, some will open a context menu.
The old menus keep evolving, and they're not getting better. The traditional menubar isn't great. It's tiny, click targets aren't big enough, keyboard navigation is slow, and the sorting of these menus is random and app dependent, meaning you have to learn where each option is for each program you use. If you place that into a global menu, you save some space in the app's window, and you're making clicking on menus easier, since they're always in the same place, but it doesn't solve the menu sorting problem, and the small click targets. If you place everything in a hamburger menu, you're simplifying the default interface, but making it harder to find advanced features.
Finally, Windows and titlebars. We use app based metaphors, and so we need to have a visual representation of each one. Whether it floats, or it's tiled, we all have windows that we move, resize, maximize, minimize, or close.
Again, this isn't efficient. Resizing windows to fit a task isn't great, titlebars are small and finicky to grab and drag, window buttons are small and are moving targets depending on where you placed the window. If you tile things, then you'll need to adjust the layout multiple times to have tiles that have enough space for what you want to do. 
Наука
Get the free Cybersecurity Mistakes ebook here: tuxcare.com/downloadables/top-ten-cybersecurity-misconfigurations/?The%20Linux%20Experiment&
What you described is almost how email worked in real UNIX, you had filesystem folder called mail and you had one program for downloading mail( mail delivery agent), another program to read mail, etc. and because APIs were stable you could change programs to whichever suits your needs. so it is one step toward what you described.
I have a Windows app and web app which are using a design idea that I got from Microsoft's Azure web (early version) and their failed Phone UI. For lack of better name it's a "Panels" design. It works both on windows big screen and on the Phone without much change. I start with a Left Panel that acts as a menu (This acts as the starting point). Selecting items opens Panels to the right. What the panels contain is open ended. These panels can in turn open panels to the right and so on. The first panel is always available to switch to other panel groups. So now I have all-in-one apps and all-in-one webs. When giving features to the customer, it's just a matter of configuration showing/hiding features.
I remembered the ** brilliant ** 3D desktop (apparently made opensource in 2012) which worked so amazingly with tablets back in 2006 or something.
BUMPTOP
Check out this video: Fko4N-EmzgM
Situation: There are 14 competing standards.
"14? Ridiculous! We need to develop one universal standard that covers everyone's use cases. "
Situation: There are 15 competing standards.
Hahah yep. This is a stupid idea and will never be implemented, but I liked the thought exercise
This is XKCD #927, first published on 2011 and holding true and immutable like the laws of physics
xkcd is great
Source: xkcd #927
@@TheLinuxEXP I like it.
You won't take the system tray even from my cold dead hands.
It sucks though. We don’t have anything better, so it’s sort of needed
Of course not, because you can't grasp it in your hands to begin with 😝 😝
Damn straight
@@TheLinuxEXP That's subjective, I don't think it sucks. I understand the concern for how applications utilize it, seeing as they are not forced to use it one way or another (making it work differently across apps) but it's kinda like saying roads suck because a tractor can drive on it and cause a traffic jam.
@@TheLinuxEXP nah you missed with this one 😅
I am a UI/UX designer. 99% of all issues revolve around discoverability and efficient activation of features. Computers are the most complex things we're making and there are simply limits to how much you can hide that complexity.
That the traditional desktop has been around for 30 years now is a testament to how efficient it is. The largest companies in the world have been throwing billions of dollars and millions of designers at the problem and not come up with something better.
Apple has spent the last 14 years trying to replace the desktop metaphor on iPad OS - and largely failed.
Your file-based approach, as presented, solves very little. You just pulled the app launcher from the Start Menu into the left bar (Image, Present, Video...).
Moving the file from app to app becomes more convenient, yes. But when I have a video editor open, my "file" is a video project. If I click "Audio", what is an audio editor supposed to do with my video project?
It also completely fails to address the issues. You state that launchers, menus, etc. are bad, yet your solution has launchers and menus. Only now they're associated to file types and managed by the OS. Which is so much worse than the current situation.
Also, what about activities that don't have files? Browsers, media players, maps/navigation, banking apps... do they still have apps?
Replacing/augmenting traditional menus with search has been a thing for a long time now, it's often called a Command Palette. I first encountered this in Sublime Text in 2009 or so. Most code editors have one now.
They solve the problem of efficient activation of features, but are terrible for discovery, unless you like to scroll through never-ending lists of commands.
The thing that will have the most impact on ALL OF THIS is AI.
AI is a Command Palette on steroids.
💯This. Thanks for writing it down. Much appreciated.
You have nicely summed up my feeling about the video. The proposed solution is just the same as it is now, instead of movable windows you get floating panels - its the same.
I dream of an AI OS. Where I just tell the AI what to do and it does it. Why so much clicking and typing. I just tell the PC what I want and I get ir. That's all. Eventually there would be no PC as we know it. The PC would be everywhere alway on ready to listen to me.
Agree with you. There's also an entire video focused on how you could make VR a better experience and it focuses heavily on the base design of the OS. And while it's focused on VR, it goes in depth as to why the current desktop experience is as it is and why it is EXTREMELY unlikely to change. Video is here: ruclips.net/video/Fhlw88_Beu4/видео.html
Great comment, OP. This is exactly what I'm thinking about. An enhanced conversation metaphor, where you can describe what you want in a "command palette on steroids" and the computer would understand, which is, incidentally, a better terminal in which you can write precise commands (like we already do) or describe what you want to do. We have to face it... we're social animals, so the most natural way to do anything is through a conversation. Up to now, the computer has been a terrible interlocutor, because it is too literal and demands you to speak to it in a precise way. But this can change. And I don't believe that anything short of mind-reading could be as effective and natural for the user than talking.
The appeal of the current desktop metaphor in my opinion is the flexibility. By working with files and separate applications, things stay decoupled and indepedent. You can have separate windows if you're working on two or more things at the same time. If you tie your files with your apps, congratulations you now have something like iOS where you can't really take control of your data anymore, and you can't do two things at a time.
It's not only the flexibility. To a big extent, it's also muscle memory. combined with visual memory. They idea of a single desktop that provides everything as presented in the video, imho, is heavily flawed. A single UI that never really changes much prevents remembering the "typical" look and feel of ones' favourite apps, and it prevents building muscle memory as well, as one can't combine said look and feel with the position of UI-elements of a certain app.
Plus, to have an app that shows me exactly what I need at the moment if would have to be able to read my thoughts, or I would have to describe what I want to do in such a detail that it would take far longer than just using a learned procedure to start exactly the app I want to have.
@@stephanhuebner4931 youre wrong, imagine file is text document, and imagine you have emacs installed...... when you open mail in emacs it shows you actions for email, when you open HTML file it shows you actions towards programming, that is what he means, unfortunately emacs is not multimedia capable.
@@stephanhuebner4931 also "A single UI that never really changes much prevents remembering the "typical" look and feel of ones' favourite apps, and it prevents building muscle " so youre saying that single ui does not provide "one"look for all your work... makes no sense logically. emacs is one ui for all my text,email,programming....
That's not the desktop metaphor. That's the files metaphor. And yes, it's far better because you can use your data independently of a particular app.
I don't think I've ever seen a video with such a high disagreement per minute count lol
I don't think the desktop metaphor is inefficient at all. Just as when you sit on your desk you take out your tools and your objects/files, I really don't see the issue with doing the same on a digital level. your ideas add opacity to the experience, a real lack of clarity on what tools you're using, etc. Many of those things just bring the problems back again - please tell me the difference between your "UI blocks" and "windows" lol - and while I can see the appeal of a more seamless interface I just don't think "predictive" is ever going to be as good as the value you get out of learning to use the tools and organising your workspace as it suits you
I see big value in not having to pick a tool and know which app does what, compared to having to sip in and out of applications all the time.
Of course, it’s just an opinion, as I said in the video :)
@@TheLinuxEXP I know, I know, it's a very good video as always :)
@@TheLinuxEXPThere are different kinds of apps, not all have the same purpose or usage. Some are ment to be in the background and by their purpose, offer some live status and fast interaction and that needs to be visible (VPN status, GPU in use, Remote Desktop app is on/off etc.); the Tray Area is a generic, flexible, good solution. Hiding those icons takes from their apps' power and functionality and adds a stupid extra step (multiply it with hundreds of times you'd click during a day to check them up for no reason at all if they stay visible). Move them in some sort of widget and they are hidden below other windows and again, you have to minimeze all to check them.
@@TheLinuxEXP The question becomes, what do you do when you don't have a "workflow" created for a given task with all the UI Blocks arranged? In that situation, aren't you left with a computer as we have it already?
He just pulls file based workflows out of his ass so he doesn't get the desk metaphor
Your launcher replacement just reinvented Microsoft OLE/COM/ActiveX components. It's been tried many times, so far no one has succeeded. Not saying it won't succeed, just no one truly has yet.
Yeah, it’s likely a stupid idea
@@TheLinuxEXP It might be fun to go back historically to all the other dead branches of the user interface tree of life and see if there's anything worth pillaging.
It used to be that you could replace Windows Explorer in Windows and I think... Norton had a replacement that people liked?
Heck, the first example I ever saw of a tabbed ribbon was in the user interface Hall of shame.
It used to be a cardinal sin to scroll vertically. Now we do it all the time and don't think about it.
@@megan_alnico Why was scrolling vertically an UX sin??!!
@@3333218 it was considered lazy. Your layout should show as much as possible together. If you had too many controls then you needed tabs or a list or a tree to divide stuff into logical sections.
The the web came along and was like ..
"Look were so easy, just scroll vertically" and all the UI people died a little inside.
I was going to bring up OLE as well, but thought I'd check the comments so I don't repeat someone who had the same thought.
The header bar on phones is pretty much the system tray
It's a bit different these days cuz the status bar usually placed next to the selfie cam where content can't really be displayed reasonably anyways.
@@maxxiongnot all phones have hole in screen for camera
just more consistent, ie icons don't look out of place
also they are only icons, not clickable, so it makes more sense then system trays
Nah it's not. There are no random apps that show functionality there in an incoherent manner, it is controlled by the OS and managed in a coherent way to give you easy access to and information about core functionality. Kind of what Gnome is trying to move to.
@@no_name4796 I find sys tray very important for me, yet I never click on it. Like everything in my status bar, they are used only as a indicator
I disagree with pretty much everything in this video. There are so many assumptions that does not hold ground in the real world. All of the "solutions" sounds like a usability nightmare.
Yes and yes ❤
I was looking for a comment who would call out.
GNOME already tried to change the desktop / window paradigm with gnome 3, shell, etc and it took 5-10+ years to get wide adoption.
Someone should have mentioned Emacs by this point
So like Emacs but with a sane, useful GUI.
Came here bacause of this. When I heared him talking about the os doing everything and changing depending on the file my first thought was "emacs". Funny thing that a lot of times the "future" is in the "past".
Came here to say Emacs! Or even something like Medley InterLisp.
Sure! And use it on Arch BTW :D
This is a great lesson of how UX and "efficiency" are not the same for a machine as they are for a human. Redundancy can make sense. A taskbar typically shows open and commonly opened apps, an app picker's job is to give an overview of everything that's available to launch stand-alone. "You never really worry about ..." yeaaaah... no. I don't even know where to begin with telling you that there are MANY use cases where a file doesn't imply the same UI/workflow. Wanna play the video file or edit it? Wait, you don't want to load up a full-on editing suite just for playing it? Wait, the UI is too cluttered for editing hi-res video because your UI needs to be able to adapt to various use cases? And auto-hide isn't doing it for you, because you may only sometimes want that stuff to get out of the way? A text file isn't a text file, but rather a chat history you intended to view in a dedicated app? Oh, wait it isn't a text file, but a settings file and you accidentally opened it, but due to some funky extension it opens in an unexpected app...
Only that you won't know, because the UI isn't app-based, but intent-based. You won't know if the modules loaded treat the file in the way you'd expect.
And the critique for skeuomorphism really doesn't make sense to me.... Do we want to get rid of it, just because it has an IRL equivalent?
Also, whenever you focus on simple UIs you'll be overloaded with demands for edge use cases that become increasingly hard to embed.
We talk a lot about tech debt, my last point shall be how there's also user ability debt. If you break the habits of people too much you'll come up with an unmanagable mess of new tech iliteracy.
I feel that forcing a unified way of working is not appropriate for efficient workflows. For example, when building a house, the framer, finishing carpenter, plasterer, electrician, and plumber, all have different paradigms for getting their work done even though they are all working on the same house. That is, different aspects of a project require particular methodologies that are optimal for that task. Computer workflows are no different. By the way, of all the distros that I have tried, I find that Tuxedo OS (i.e., Kubuntu KDE Plasma) is exactly what I want for computer interaction, and I would not change a single thing.
This is so true. Reminds me of the endless terminal vs gui debate. In my opinion, if you need to do a task, and can be broken down into repetition and/or moving data between several programs, a terminal is often going to be superior. If your task requires selecting a few options from a list of hundreds, a search is great, and a CLI can be an equally good solution assuming that you know about the options that you need. If your task involves manipulating/consuming visual media in a visual or necessarily manual way, keeping a ton of context in mind, or being straightforward for most people a GUI is the way to go (I'm sure there are other reasons too).
I sympathize with my fellow terminal enjoyers though. Sometimes a terminal interface is far and away the best way to do something. Making a one-off and highly specific change to your system configuration? You bet the terminal is the only way to go. I'm only going to know about a setting in control center if I've just read a help article. Might as well just set the setting using a single command that I can grab straight from the article.
I think terminal evangelists are largely butthurt that people won't spend the time it takes to learn how to use the terminal, which is ultimately sad but pretty understandable, most people don't know that using a terminal could bring them value.
There have been so many times that I've been able to *NOT* install new software because I learn a new command that I already have installed, and each time is amazing.
The idea around the omni file handled by the OS instead of going to an app first is really undercooked. First, it's a programming nightmare. Second, it describes and fits only a specific work type and workflow.
I do agree that rethinking those concepts is worth it tho
"I wanted a python script not a Matlab live script damn it!!!111!!!"
Me trying to omni-file my homework.
This is what I was thinking. You'll have to design and program every workflow possible, which in itself is an impossible task.
Yeah, his example(s) clearly showed from what limited workflow he was coming at.
I don't think it's as much of a nightmare as all that. Programs would have to present interfaces for importing and exporting documents in whatever format is supported and have information regarding which formats lost what kind of data, but really it's basically just pipes in the GUI. The main difference would be the negotiation of what file format to use rather that just text (although with pipes you can transmit any data, that relies on the user being able to convert the data as necessary).
I think the key to the concept that makes it a worthwhile idea is that users would be able to install task-based packages to suit their individual needs that would plug into the APIs of the general-purpose UI, much like you can install extensions to a desktop environment or packages of terminal commands. Use .xml for GIS mapping? You install the mapping package. Use .xml for music publishing? Install the music score package. You would then do the GUI equivalent of the command required to make a new .xml and then the command to edit it as a music score. The music package would include export functions so you could make, for example, a PDF. You would then have separate packages to handle PDFs (e.g. an editing package, a publishing package, etc.). Programmers would not be required to make monolith software but would instead build packages of tools to meet a certain use case. An analogous paradigm shift is happening in mobile right now with the shift from apps to widgets.
The desktop metaphor feels so 1980s to me.
In the 1970s we had the command line interface, often with menu screens.
Before that there were just switches and lights and your binary command understanding.
Your document-based idea sounds EXACTLY like Apple's OpenDoc system that was developed in the late 1990s and then ditched when they decided to go with Mac OS X.
I guess if it’s a concept Apple discarded, it has to be a bad one 😂
My first interaction with computers was with punched cards.
I got it, I got it!!! A virtual stack of punch cards you can drag and drop into the desired order, drop into virtual card readers, and so forth. To create new data, virtually load a blank card and punch in EBCDIC on a virtual teletype keyboard!
@@michaeldeloatch7461 You should also simulate the shock when you accidentally drop your card deck on the floor and you suddenly realize that you forgot to draw a diagonal line across the card edges to help you reassemble the deck in the correct order.
Ahh I wondered if someone would mention opendoc. System 7.5.x was my intro to computing and the main OS I used growing up yet I had no idea opendoc existed until long after it was dropped. I've often wondered what a small-composable-component-based GUI environment would be like to use. I don't know the history of opendoc but Wikipedia makes it sound like the main reason it was abandoned was that Steve Jobs didn't like it.
I don't think graphical interfaces suck. They sure as hell beat having to use the terminal, which kept people from using Linux for years.
Of course they’re better, they’re just not optimal and we could probably do better
I prefer the terminal for many tasks
terminals are either the best or the wost depending on the person you ask.
i am in the 1st group btw
Depends on how complicated the task to he done is. Making many apps interact together is far easier in a terminal, but for "I don't even know what my options are, let alone how to spell them" graphical apps are great. And naturally they tend to excel for graphic-focused tasks.
@@TheLinuxEXP The problem is "optimal" for one person is a horrible for another. The middle of the road option works for everyone, because it doesn't cause strong emotions.
The fact that we *haven't* come up with something better (or at least not something that works for everyone in every situation) is a big hint that there may not *be* a universal solution. If there were such a thing, it would spread like wildfire, but many of the innovations from the last few years (I'm looking specifically at the hamburger menu compared to a global menu bar) are worse than what we had before, in terms of access-speed. It *may* be better if you have a huge hamburger button on a touch device, but for regular mouse and keyboard interaction, it's a bad idea.
People have different workflows and some need to use a mouse to access their apps via some launcher, while others (me as well), use a command-based launcher, which can be invoked via a keystroke and can get an app running in a second or two. But it's not even about speed necessarily, but about what people like to use, and everybody has a different taste. As is being said in the video, what is "better" is often highly subjective.
"It may be better if you have a huge hamburger button on a touch device, but for regular mouse and keyboard interaction, it's a bad idea."
This I agree with. A mouse can hit a much smaller target, and there is a lot more space on a desktop monitor. Burying everything into a single "hamburger" menu adds an extra click to everything you do. Ideally, a UI should be able to morph to whatever device you are using. There's a principle called "responsive design" in web programming that does that. I've developed web pages that are a full menu when the window is wide that morphs into a hamburger menu when the window is narrow. It's a bit more work, but tools can be offered to make it easier for developers.
@@logicalfundy Exactly! There is no single solution that fits every use case. I think I've seen toolkits that provide possibilities such as the one you described. Ideally, that shouldn't be something a developer has to think about, the toolkit itself should do most of the work.
Disagree. In reality, learning curve and UI backwards compatibility is just way way more important than the 10 % efficiency we could get out of a redesign.
There are so many things in design (esp. computer UX design) that have no other reasoning than a historical one. If you really think about it and look at the examples, I can't imagine you'll still believe that user interfaces are designed from scratch to be the theoretically most efficient one.
Just think about it: If a company tries to make you _pay_ for _having to learn_ something, they'll be out of market ASAP.
@@aheendwhz1 I don't need to believe anything, because these design principles didn't come out of nowhere. They're no accident but often based on decades old studies that showed that they work. It's not "history" as such that makes them valid, it's that they do work and are efficient. Which doesn't necessarily mean "fastest", because ease of use plays a big part in it. It rather seems to be that some of the later, trendy design principles are the ones that aren't based on research but just wild ideas that sounded good in theory but do not work so well in practice.
I don't understand your last sentence, at all. You don't *need* to pay to learn how to use a computer, as the companies always had some kind of manual included that explains the most fundamental ways of using it.
@@aheendwhz1 I do agree that the learning curve is important, which is another strike against trying to shoehorn mobile UI conventions (such as hamburger menus) into an area where different conventions are the norm.
The conventions that arose around computers aren't arbitrary or "have no reasoning." They're built around the way we interact with them (keyboard / monitor / mouse). Which is why adapting to and respecting the platforms they are designed for is important.
Your idea would require the whole world of software and apps to change for it to be achievable, since everything is made with this current metaphor in mind - the biggest revolution in operating systems in 30 years..
Yeah, that’s why I said at the beginning it’s probably not even a good idea!
It would also require a huge amount of openness between developers and corporations not trying to create moats for things to be interoperable.
The document rather than application paradigm is all well and good... until you want to play a video game. Do you create a "Quake game" or "play solitaire" document?
And when he tries to open that omni-document 1-2 years down the road I'm sure it will work flawlessly still as all the apps will still all be there and support all those little features in all those sub-objects.
yeah, there should probably still be individual applications in this paradigm, and this would be mostly used for applications working with files
Exactly. This "document" idea is why we're stuck with horrible broken web: HTML (document), then Javascript plastered on top of it to make it app-like.
Everything should be application-based. Applications can do documents, documents cannot do applications.
That reminds me of something from an old version of Wikipedia's article about Apple's Lisa computer (from 1983): " _Apple presents tasks, with Lisa, in the form of stationery. Rather than opening LisaWrite, for instance, to begin to do word processing, users initially "tear off stationery", visually, that represents the task of word processing. Either that, or they open an existing LisaWrite document that resembles that stationery._ [...] _A drawback of task-oriented design, when presented in document-oriented form, is that the naturalness of the process can be lacking. The most frequently cited example with Lisa is the use of LisaTerminal, in which a person tears off "terminal stationery" - a broken metaphor._ "
Windows 95 had the embedding concept you are talking about, it was called OLE (object linking and embedding). You could have a spreadsheet in a text document, and you would even get a mini editor interface in the host application. It wasn't very discoverable, which is probably why it never took off.
I came here to say THIS
and it was limited in its ability, for real stuff one had to break out the big tool anyway not to talk about incompatibility issues if ANY of this data has to be accessed 1,2,3,5,10 years down the road.
OLE is still a feature in Windows 11, the only problem is that almost no programs, 1st or 3rd party support the feature and the ones which do are obsolete and are usually hidden behind Window's optional features now like the Windows Wordpad which is the one program left which still supports OLE natively.
Never took off? You've obviously never worked in the corporate world. I come across lots of Excel spreadsheets embedded in Word docs or PowerPoint presentations (either put there intentionally or as a result of copy/pasting!)
@@asystole_oh the nightmare of not being sure if the embedded Excel file in the monthly steering committee PowerPoint was really saved 😵💫
your idea of a task-based workflow is interesting, but I think the level of collaboration between developers would have to be incredibly high for everyone to write task "blocks" that all plug into the same huge framework. not to mention such an underlying framework might get really complicated and messy, as it would essentially be a superapp that must be capable of every conceivable function.
one other point, yes it may be somewhat inefficient to have a blank desktop with a pretty picture when you log into your computer or once your apps are closed, instead of some sort of widget array or big launcher that takes up your screen. however, using a computer isn't only about "completing tasks", it is also about "imagining possibilities" and sometimes having a nice artwork to stare at is better for your imagination than your computer "intelligently suggesting" a bunch of potential tasks to you.
Emacs does that, doesn't it?
Cool beans Nick but I think it would be overwhelming and I would just shut the system after 2 minutes
Hahah it’s just an idea, a probably stupid one!
Haven’t seen someone use cool bean in looong while
I like my good old desktop.
And that’s totally fine! My idea is probably dumb
@@TheLinuxEXP Not dumb, just not for me ;)
@@zolaarczakle same
Microsoft spent a lot of resources on UI research during win95 development, there is a reason why they offer same ui.
That’s was 20 years ago though
@@TheLinuxEXPyou forgot about Windows 8. That was a failure.
@@TheLinuxEXP 1995 was 30 years ago...
@@TheLinuxEXPJust because it’s old doesn’t mean it’s wrong. That’s a logical fallacy. Like saying theory of relativity is wrong because it’s old.
The offer the same UI because everyone is used to the old one and gets mad whenever they change anything. See Windows 8 or 11.
The concept of windows is great because the *segmentation* that individual windows provide allow for as many different workflows as there are people.
Most of what you are describing aligns with Apple and IBMs discontinued efforts with Copland and Taligent en.wikipedia.org/wiki/Copland_(operating_system)
and particularly the OpenDoc - en.wikipedia.org/wiki/OpenDoc
Project. Generally OS/2 Warp and its use of SOM was far closer to being capable of getting to a "task based" end than anything I've seen in GNOME.
I think your project requires Linux people dusting off the work of SOM version 3 getting IBM (who know own Redhat) and structuring a clever project implementing Firefox, LibreOffice & OpenOffice (a case of OpenOffice with work done in LibreOffice), QT, SOM and some other odds and ends ...
one word: KParts
You are describing an OS set for a very specific user case, yours. But a computer is designed to be used by any kind of people, with all kinds of software and processes, any kind of needs. That is why we have the flexible design that can accommodate everyone, with or without some changes.
Just as you say, I will also add my subjective opinion:
- the Desktop - we are humans, we don't need only functionality, as if that was the case, we wouldn't have created art. We are not robots, so I couldn't care less about just clicking once to start working, I can do 2 clicks if I enjoy more my experience. Plus, the simple existence for decades with nothing better to replace it (there were tries), shows that until now we don't have a truly better replacement.
- System Tray - one of the best inventions for some background tasks. I can see if the VPN is connected and interact with it in one click, I can see the battery level and set the power level (used to be like that), I can tell Shutter to make a screenshot without opening it, I can see if my Remote Desktop app is online and working fine, I can see which GPU I am using (Intel or Nvidia), in the past (on Windows) I could see and do things with the antivirus etc. And no, hiding those things isnnot better, I want to know some of those work without clicking 2000 times a day to open some kind of window. Phones have this, too, of course, adapted to a small touch-screen interface; you can see the battery level, if you are on wifi or mobile data, if bluetooth is on, the time, if there are alarms set, if the VPN is on etc.
It is the flexibilty of these that makes the computer functional for everyone. Imposing some niche needs over the needs of all is not the way forward, it will never work.
I'm not sure efficency is the best goal, especially for a work machine. Flexibility and adaptability have always been much more important to me. The idea of an empty desk space that you then load up with the appropriate tools for a given task is just more flexible. See a neat new tool? You can just add it, and it's not going to affect any of your other tools. Want to switch from programming to painting? Sure, just put one set of tools away, and pull out the other.
KDE does this.
@@turtlefrog-tn3ek Obviously. I was talking about design ideas, not looking for help.
This. I'm so sick and tired of "efficiency obsession".
i feel it's the same type of idea that led humanity to Juicero. Is it really better to have everything in a single workspace? What if i want play some music while i work, check if my server is not down, reply in 3 different messengers and still remember what i was working on?
Multiple workspaces (with the xfce workspace switcher on a panel) is my favorite Linux desktop feature. I tend to use many programs full-screen, and workspaces give me so much more room to work and multitask. The xfce implementation is far better IMO than the Windows version, or other desktops I've tried.
@@javaman4584Why keep some windows open without needing that? I don't need the Remote Desktop app visible, I just need to know it is connected to the server. Same for others. It is also a spare of resources, as they need to be kept in graphical memory.
The computer will know what to launch (or what to close) by reading the expression on your face. Good use case for shoving some more AI-based spyware into our computers.
Haha yeah, we do NOT want that
No no no... not your computer, directly to your brain. I don't see how else.
Trying to reinvent the wheel is something else. It has to be a genius breakthrough, or it will fall on deaf ears. Take the qwerty keyboard for example. Other keyboard arrangements have been proposed, but only few people are willing to learn a new system. It was hard enough to learn it before it became muscle memory. Qwerty gets the job done, so it will probably never change.
I don't have a problem with the desktop metaphor. I do agree that implementations can be done better.
As far as your "idea" of a document-centered interface instead of the application-centered norm, this has been discussed, prototyped, etc. many times. If anyone could succeed, it would be Apple or Microsoft, but they haven't. The incentive of the lock-in is great, but they haven't done it. There's a lesson there.
I like it that you question the basic interface model assumptions but so far I haven't seen things that work better for me. I don't like the workflow-based idea because it means I can only do the workflows that the system designer has thought about and implemented. For some special purpose devices or systems, that's fine; but for my general purpose I want a toolbox with readily available tools that I can use in ways that suit me. A mix of the current desktop and CLI work well for me. Auto-completion, voice recognition, and AI do not; I find myself constantly fighting against their wrong guesses about what I mean or what I want. The better the AI gets, the harder I have to work to find the complex and subtle ways I'm being led in the wrong direction. I'm open to trying new things and do so quite often, but so far in the systems I use, I see mostly misguided gimmicks.
The problem is, nobody actually cares, neither developers or users. They just want their things done.
What if what they think gets their things done actually makes them not get their things done or makes them take double the time?
Great response. I didn't know how much time the terminal was costing me until I learned to do the same jobs with a GUI.
Not sponsored by Gnome?? 🤨😅
No, why?
@@TheLinuxEXP they seem to be the only project trying to change "the desktop" :D
@@Beryesa.And they better stay the heck away from my phone, tablets, ereaders and streaming devices.
@@imacmill hey, it fits well to the tablet form factor tho
@@Beryesa. Ya, nothing says usability like a big, empty screen.
We improved on the WIMP concept backwards. Old OSes were so much better at that (not Windows, but Mac and Amiga, for example). And OF COURSE we had something way better in the olden days for (almost) everything you mentioned.
(1) On the Amiga, the desktop was a place for things (files, apps, everything) that you currently worked with. You could put anything on the desktop without moving them away from their original folders (*drawers) and you could "put them back" to their original places with a single click. It was perfect. The desktop was exactly what it was supposed to be.
(2) the launchers are an additional layer of complexity that is only needed because the file and folder paradigm is not working as it should in modern computers. The problem is that files that you care about as a human (the executables to launch your software, your documents, the relevant configuration files) are intermingled with system files you have no business interacting with. And these are spread all over the system, in hidden folders, system folders, user folders, pre-configured folders, without you having a say about where they should go. Of course you will need additional software to make this huge mess user friendly.
On the Amiga, the important files had an icon, and this is what you saw. The launchers, your documents, help files and guides, configuration files. Everything else was hidden by default. You could keep your files and your software wherever you wanted. It was easy to organize your computer by just having a folder (*drawer) dedicated to an activity, and it had all the software, documents, documentation, everything that you needed nicely organized. If you wanted to move it you only have to move the entire folder and everything would go with it.
By just having this small paradigm of having an icon for important stuff only, the Amiga replaced the launcher, the Application Manager, the activity manager, the "My Documents" folder, the registry, the /usr/bin folder (and others), the AppData folder, the Library folder, Preferences folder, and so on and so on that are only layers of unnecessary complexity on modern OSes to compensate for the failures of their original abstractions. And it was BETTER. You could actually be organized.
(3) With the Amiga OS, you could already organize your workflow as you described by only using folders (*drawers) and icons.
(4) A better paradigm for the tray was already implemented... you guessed it, in the Amiga. They were called Commodities, you would start them from a startup folder (WBStartup) from which you could just remove them if you did not want them to be launched automatically, and there was a utility you could start with a key combo where you could start them and stop them and access their configuration interface.
Otherwise I entirely agree that we should have a workflow based interface rather than an app and document based one :) But every attempt so far on Linux has failed. I tried KDE "Activities" last, and my impression was that the developers did not communicate well, or just did not agree on what it's supposed to do, and today it's a complete mess, totally unusable, with one interface implementing half of one paradigm, and the other interface half of another. Absolute mess. I miss the old days when computer interfaces were developed by professionals for professionals, and not by autistic Trekkies for absolute morons.
Fully agree.
The issue with this system is that the OS is deciding what app you need. Aside from the inevitable "we made a deal with this app so we're pushing it by default" that would pop up, how does it know what I want to do? I copy an image from a website, what am I doing with it? Viewing it? Saving it? Editing it? Sending it to someone? If I'm sending it to someone is it e-mail, text, discord, etc?
My biggest complaint with current UIs is title bars. I often wish I could just get rid of them and have another way to close the window. This is especially prevalent in games, since a lot of them refuse to let you go into ultrawide to match your monitor (looking at YOU Fromsoft). But if you put it in a window now the stupid ass titlebar is pushing the actual window down and cutting off the bottom, unless you either put it in too small of a window or use a 3rd party app/tool to move the titlebar off screen.
I find it funny that we both had the same idea but coming from completely different positions. Many years ago when I thought of what you call this "canvas" idea, I thought of it exactly BECAUSE of the desktop metaphor. When we, in the physical world, bring something to work on our desks, we can interact with it using whatever tool we want, a knife or hammer or highlighter etc - so making every relevant tool available to a file as needed, like you show, is actually closer to the real life top of a desk than what we have now.
BTW, there even is a lot of software already doing it, freeCAD with their "workbenches" for one example, DaVinci Resolve for another (where you can change the whole workflow and layout and focus through those "tabs" at the bottom, going to the VFX panel, Audio, Colour Correcting, etc).
Sill, of course we both know it is absolutely impossible to have a system like you described here - some mime types are simply incompatible. While on my actual desk if I want to draw using rulers and protractors and take note of every line length and curve and describe it with a single "formula", but also put a sticker or photo of someone, no change is needed to the underlying "file" - while on the PC of course raster and vector are completely different beasts, and that can be repeated for literally every single file format. If you annotate a PDF should it be saved in a sidecar file or merged into the image of the pdf? If you copy and paste do you want a jpg to be taken inside a graphics editing or do you want to copy the text into a text editor? Or maybe even the nodes of each letter from the typeface to be imported into some layouting/publishing or vector software. And so on.
I kept watching and thinking "is the twist going to be that he's reinventing the terminal" into like the 8 minute mark
👍
I always found the current iteration of modern start menus (minus the ones riddled with ads) fairly efficient if you're using them properly. For apps I'm familiar with, I'll hit the Windows key, start typing the app name, then hit enter to launch it when it appears. If I'm using something I dont open as often I'm willing to take the extra second or two to click through the GUI categories to get to the app I'm looking for. With things I'm using often, Im essentially bypassing the GUI with some fast typing on the keyboard which is very efficient, and for ones I may not remember the name of the GUI categories will guide me to them with little hassle.
An idea i have is to create work sessions on startup, kind of like user profiles: you select a user, then select what session you want (work, drawing, gaming, web bruising) that can be pre configured and customizable. You launch your session and all the necesaary apps start in preset locationson the screen. You can even create an assistant to help set up the session with prioritary apps and others hidden (accessible via exact naming on search). Further down the rabbit hole, a custom right click actions list (super right click for example) per window or app. Could be more easily implemented i think
KDE Plasma already has that. You can do it in other DEs with some scripting and workspaces.
But that's not really how most people use their computers anyway.
People don't lock themselves into "one type of session".
They do some work, then in the middle do some gaming / browsing / chatting, etc.
Humans are complicated creatures.
Your new desktop idea is very interesting and i think it could be beneficial for a lot of people, but it would require a whole new integration ecosystem. So if someone really wanted to start that project, it would compete with all the other desktop environments. I'd love to see that. :)
The only time I see the desktop screen is the first boot. Any other days I just open the lid of my laptop and continue from where I stop. That's also a reason why I stopped putting widgets on the desktop or caring about wallpaper.
For the launchers, I use Raycast (which is the FREE replacement for Spotlight and one of the main reasons why I can't switch to Linux from MacOS). The big bunch of stuff (from quick opening apps by searching its name to see recent meetings, calculating stuff, convert currency or quickly translate the word) I do with it without even switching to another app's screen (since Raycast is just a pop-up)
The global menu thingy is sucks (in my opinion) when you daily use a small 13/14 inch display of your laptop. All this dosens of menu items just can't find enough space and even shifts menu bar items I care about (like VPN/WiFi/Battery/Headphones connection status)
About menus, they still should be there, but I agree, that they mostly need a search bar as well. Maybe it is worth having a "Quick Panel" as it exists in Obsidian or VS Code.
All the things you described in Raycast sound like KDE's KRunner, I don't know Raycast at all so maybe it's different, but you might want to check it out in a VM or something
Edit: I looked it up, it does look like it integrates with a LOT more things than KRunner, which is already super good IMO
raycast sounds a lot like rofi or dmenu to me
I always love watching your videos and hearing your thoughts on things. You always make some interesting points on on equally interesting topics!
That was some fun food for thought. IMO Tiling Window Managers + keyboard shortcuts + app launcher is our best solution currently for a productivity focused "desktop" experience. Something I'd like to see come with a WM is to add some sort of universal/standard configs (system wide, not just barebones settings like it is now with WMs) that can be edited via text editor or GUI to change a plethora of things like your 'default' apps like what we have now, the ability to change/add/remove new keyboard shortcuts, ect.
From my experience, TMWs don't have a good OOTB experience and you need to plug away hours/days to setup what you'd like from the ground up, so it'd be nice to see one that's catered more towards newer users and not just "go to X's github and use their dotfile configs" but still offer great customization options if you want to do down that rabbit hole.
Mate, at what point the computer uses itself, what I do like about current systems is that you can pretty much personalize them to your liking, it forces you to be creative in order to be productive and the one size fits all isn't a real thing for example I do like windows style UI elements from the tasks bar to the start menu as long as they're customizable what I don't like and I can't get used to is macos style or Ubuntu style desktops, now you don't have to like what I do and that's the beauty of it, for instance on android the launchers are mental from purely search based ones to simple lists to tiles and even desktop styles but the beauty of it is in customization something I feel android has more of than linux and certainly more than iPhones or macs.
Starting the job with a blank file that becomes something at the end raises problems with compatibility and if the user is asked to define the file type it's a new can of worms.
Also the "new file" ui has some issues with specialized stuff like programming and arduino or diagnostic a car or some hardware, it's just extremely an extremely limited os
With the advent of OLED monitors, I've always thought it'd be cool to have a desktop environment that was controlled via pie menus (or I guess "marking menus" as Fly-Pie calls them) rather than static bars and stuff that always stay in the same spot and are easy to missclick. Even right click context menu items would all be in a pie menu. The whole thing would require more than two mouse buttons though (and middle click isn't a real button, kinda like how L3 and R3 on Playstation controllers aren't real buttons, ew, clunky), and you can't just expect everyone to have a pointing device with good forward/back buttons (or to even have them at all), and this whole paradigm I'm picturing would be not nearly as good if it required keyboard combinations, which I hate; I like pie menus because they're a way to effectively turn one button into multiple buttons without having to hold down a modifier.
Also, I haven't actually watched the video yet, I just instantly wanted to ramble about this UX idea that I wish was a thing. Is anyone working on a Fly-Pie alternative for Plasma or a desktop-agnostic solution, by any chance?
This is why I went for a MiniLED display instead of OLED. OLED's also too dark during daylight hours. I'd give a transflective LCD a try if they were available.
@@mmstick I've never heard that term before, what's it mean?
@@HontoNeet TLCDs are about to make big comeback. They are LCD panels which emit images by reflecting light from your environment, rather than by emitting it from a LED. TLCDs have an optional backlight that can be turned on if it gets dark, but otherwise they only need natural light. They're much easier on the eyes and use 70% less energy. So they're like looking at paper, and will be competing with e-ink.
@@mmstick Whaaaaaaaatttt? You're telling me this whole time there was a reflective type of LCD that looks like epaper/e-ink? And it can use either ambient light or an actual backlight backlight? That sounds crazy, having normal refresh rates on a display that looks like that. For me though, I just wanna see microLED become a thing for real, overcome whatever manufacturing yield issues they have so the prices can come down and actual buyable products can be developed. MicroLED apparently can have even better pixel response times than OLED, and I really hate non-infinite contrast ratios, so the idea of a display that's basically OLED but better is very attractive to me
@@HontoNeet See the Daylight DC-1's Live Paper; and Radiant Eazeye. Neither are currently available to buy yet.
For context, I was a tiling window manager user and recently switched to KDE.
Desktop:
Personally I don't use a classical desktop; no icons and no background image. Even on KDE.
Launchers:
The start menu is useful to find apps when you don't know or forget name of the application. My preferred way is searching instead using the menu, but having a list of apps categorized into genres and alphabetically sorted is something I do not want to miss. At least in KDE a short description with easy to understand keywords can be included below the application name. I personally dislike what Gnome does and replaces each app name by generic terms.
Having said this, I really appreciate and like your idea presented at 6:50, which is file or task oriented workflow. The idea basically is to have a single application window, which will get filled by menu items from other programs functionality in a dynamic way, like a plugin. That's an interesting design philosophy. But this does not need to replace entire desktops, it is enough to be "just" another program which other programs can support by providing an extension/plugin. It just needs an API and service that runs all the time to hotplug functionality.
System Tray:
It's actually useful and I like it. No need to open a window or place something on my visible workarea, no need to switch window, especially with maximized windows. And even more important when you use multiple virtual desktops. The system tray will always show me relevant information in a very small area of the screen, regardless what window I have visible and which desktop I am in. I argue, the systray is even one of the most useful features of a PC system. Even smartphones have it.
Gnome did it way worse. No longer are tray icons visible, which would show me some information or state of an app. It's hidden in mulitple clicks away.
Menus:
Menus itself aren't a problem, its only a problem if devs put too much into it. The hamburger menu made it even worse for me, as the menu items are behind another click.
Ubuntus retired Unity desktop had this neat global menu system, which only displayed items relative to the current active window. Kind of similar to your previous discussed idea of task based workflow.
Titlebars & Windows:
Now my time has come. Automatic tiling of windows should be the norm, as manually managing windows is terrible. Off course with exceptions. Title bar is mostly needed to move a window and should only be displayed when having floating windows. But one can still resize and move windows by holding a key and moving mouse too, or with shortcuts.
Overall I also use a lot of terminal, which has its place. Every PC user should learn how to use the terminal, to get the most out of the computer. And I mean everyone, even non technical users. Because I am convinced that every non technical user would have a benefit of learning some basic terminal functionality going forward in their life using a pc. I'm actually surprised you didn't "touch" the subject of touch. I was expecting a touch screen friendly and focused user interface from you.
Wow nick, this is such a cool experiment. I think this shows some really cool concepts that I think would be cool to see realized. As a first draft it raises some interesting possibilities, and Im curious where a more iterated version would go.
One of the main issues with the way we interact with a computer isn't just the software, but the hardware as well.
The keyboard and mouse are inefficient devices, that slow down input, and haven't changed for years. There have been attempts, but none have really moved the physical interface forward. We're moving towards 3d scanning being good enough to see and translate body movements in to inputs, it's good but not quite there.
The future of computing needs a holistic look at both sofware and hardware to move forward.
A great deal of what goes on on the system tray could very easily be displayed/interacted with through hardware as well (physical volume knobs, LEDs next to USB and other ports, etc).
@@hindigente yep, I'm slowly trying to find new methods, as I'm starting to get limited function in my hands.
@@foznothI've had to drastically change how I use my computer thanks to wrist pain, so I know where you're coming from and I'm sorry to hear you're experiencing something similar. For me the game changer was switching to voice commands for most stuff I do, so just wanted to suggest having a look into that. In particular I'm using Talon Voice which has basically saved my career as a developer. Best of luck to you!
@@CS-ep3ku I remember testing the first edition of Dragon Dictate. It would only understand American accents. I may to revisit that. The software, not my bad attempt at an American accent.
I'm hoping for medical intervention as well, so we'll see.
Thank you for your thoughts.
@@foznothI'm American so can't speak to the accent issue (no pun intended), but yeah it's worth a shot, especially if you're tech savvy because you can customize it like crazy (add custom commands etc). I've heard that the voice stuff built into newer OSes is pretty good these days too.
Oh, also might be worth looking into eye trackers (ex: Tobii 5), it seems like a lot of folks have had success with those.
Of course if you do opt for surgery I hope it helps!
"We are stuck with the desktop" -- 😂 laughs in tilng window manager
I feel that lol, on swayfx right now and am loving it so far! Tiling WM's are def not for everyone though cause they require way more configuration
Power toys in windows. It should be bundled with the OS.
They are same in this context
For a beginner Linux user, i love the Tiling from Pop os.
@@joshplaysdrums2143 just for the future tho, it wouldn't be too hard for a settings gui for tilling wms/compositors to be set up
The taskbar and system tray IMO can probably be safely merged into one. But that's different from getting rid of the system tray in terms of still having tray items in some place.
I think the gnome overview effect would be phenomenal if applied to a VR desktop type environment
I always felt the effect was akin to me backing away from my desk to see the whole thing then sitting back down with a smaller section of specific things directly in front of me
I'm dyslexic and visualizations really help me organize my thoughts
There is actually a project that *kind* of goes in the direction that you are talking about.
Essence OS: It is a small desktop operating system built from scratch and has an idea of tabs and windows that are program independent, so you can just have a window with multiple tabs that contain completely different programs.
That's BeOS' design from the 90s. You can join apps together by dragging their titles over another, creating a single tabbed window.
Open source version is available called HaikuOS.
gnome should have a clipboard manager like kde, that can support text, images and files etc.
And a screen brightness GUI/tray icon! And a helluva lot more customization ability.
Desktop metaphore - I am so used to Windows that I won't switch to anything with wildly different layout. I tried to use tiling window managers, I tried gnome (that was a speedrun... look, logout, login to KDE, uninstall Gnome), I tried MATE (this is quite decent). But for now - KDE and LXQT are my favourites. And please keep desktop icons intact. I do use the desktop as... well... my desk top. I keep the most recent files that still wait to be categorized there. From time to time I have to clean it up. But this is much more convenient than no icons at all.
I would like if the icons were widgets that show status of given program. For example - number of unread emails, face of Doom Marine from the last save etc. I know this would require some part of the programs to run in background... so I could either set the update frequency or even update manually. Evolution over revolution.
Dashboard with widgets is also quite good idea. But with, not instead of icons. Even Android uses icons to represent programs.
Launchers - easy and quick way of categorising the programs is nice (like old Windows XP). I usually resort to text search.
7:42 That would be something nice. However it would require some "uber-formats" that employ possibility of adding multiple data types.
11:52 typing to find any option from current application is great. VS Code and IntelliJ Idea have them and I use that feature every day. I want every single app to have this.
KDE is awesome because it recognizes nothing beats the UX design principles Windows pioneered back in 95-2005. Not on PC. Instead of reinventing the wheel badly like GNOME, they just kept refining it.
After many years using emacs, the fact that pretty much everything there is a buffer is super powerful.
Just think about it - in, say, a desktop spreadsheet program you have a completely different interface for your data, for the settings menu, for saving files, for the million little dialog boxes... it's actually kind of terrible. In emacs, everything is a text buffer - all of your usual methods for navigating, searching, switching, etc. all work everywhere, whether you're editing a file, customising things, browsing files, answering a yes/no question, whatever.
Difficult to see how this could apply to anything other than a text-based interface, sure, but it's like GUIs have never even tried to reach that level of consistency. Modal dialogs must be destroyed, though.
Yes it works great for text-based stuff. Not for other stuff unfortunately.
They literally have GUI / UX engineers. I remember countless projects in corporate environments where they would spend months / years obsessing over every detail, color way, icon, etc.
All 100% form of function types, holding up multi-million dollar deployments.
Unfortunately human beings are shallow creatures who only care about appearances.
So, form over function is what sells the "shiny new products".
We need to stop buying form over function.
But tell that to billions of people who are obsessed with how thin their laptops are.
At the end of the day, everything is rooted in human nature.
For me desktop metaphor works greate, your idead is a bit how blender works.
And also your main point that is hard to firgure out which app opens what - I don't get it was always easy for me,
as I find app I need once and then I autmaticly remeber it and also system will connect this app with proper file types.
You need diffrent apps, as you don't want to load all tools for edting file when you want only to view it.
I love system tray - I don't know way.
I hate hambuger menus, I want all apps to have "Command Palet"(like Unity's HUD) a searchable menu, some apps that have it:
VSCode/Godot (Shift+Ctrl+P), Blender(F3), Gimp(/), Inkscape(Shfit+/), LibreOffice(Shift+E)
The system tray is awesome because it conveys so much information in a very compact, always visible way.
I wasn't able to find a bar for Hyprland that I liked, so I tried doing without and it has been totally fine. All the usual bar stuff is relegated to a "special workspace" which is an overlay that can be toggled on and off, like macOS Mission Control or whatever it's called. No title bars, no tiny buttons for minimizing or maximizing, no fiddly mouse controls for changing window size. Having a (search-based) launcher I'm totally fine with, by far the most effecient way to open what you want. Also, no notification daemon, I consider them an anti-feature that only distracts you. It's total computing bliss, everything just works and does what I want.
That had given me really unique solution ideas. Thanks a lot. I also have some desktop environtment ideas by my own and you made me discover more problems. I hope i'll publish my own DE in one day
I am happy you kicked this hornets' nest, Nick. There must be ways of working with our computers that let us do what we want, securely and privately, while reducing the overhead. We may be blind to this overhead but you're showing us how the current motif does impede or frustrate our work. Thank you.
I was with you until you showed your idea for "files" and saying everything should be OS-level instead of having apps, it doesn't make any sense in my opinion.
The whole paradigm of 'Hardware > Kernel > OS > User space apps' isn't going to change, we live in a capitalist world and people won't give up on making third-party apps as a business or otherwise. Users have so many niche needs that you can't possibly implement and maintain enough apps at the OS level for every single one of them. And why would you? It's a very restrictive way of doing things and locks you in to ONE choice which might not fit your needs or preferences, instead of many.
Now I agree that for a same file-type such as video for example, you shouldn't need to have one app for cutting/editing, one for special effects, one for subtitling, etc.. and in that sense I think Davinci Resolve has a great philosophy with its workflow and workspace tabs, and I wish more apps worked that way.
Same with images, I think viewing and at least basic editing, sorting, tagging, should be available within the same app and built-in to the OS, and to be fair we've seen efforts being done towards that on basically every OS out there.
As for the system tray, that's completely subjective, and you've mentioned your dislike for them in quite a few videos at this point without providing a poignant argument. I personally enjoy the idea of a system tray, and so do many people out there.
I'll say that your idea of having a widget of some sort on a desktop would allow that space to be better used, and it would make a lot of sense for say a media player, or a file manager, but this would not work well for every app out there, and you'd be essentially asking every app in existence to comply with this. It would also mean background apps would use more resources since they need to display information on said widgets.AND you would have to go back to your desktop to access background apps, instead of having them accessible and visible at all times in the corner of your screen.
Either way, I think UI/UX is going to change drastically once A.I improves further and becomes more integrated into our systems. A lot of our current troubles will vanish.
Both because the A.I will be able to do a lot for us and more efficiently than us, but also because said A.I systems could be used to find new creative ways of improving the human interfaces.
the issue with the sys tray is inconsistency. For it to be well implemented, there should be some standards.
Like every user facing app that runs in the bg should show up there and have a close button, they should probably also have to use monochrome icons to fit in with the rest of them.
Not every single app needs a widget (like a cloud sync thing probably doesn't), but it would be nice if there was a standard for having widgets in a control center kinda way.
@@raidev_I agree, better standards would be nice.
@@raidev_ Why in god's name do they need to be consistent?
@@Christobanistan because they look ugly, cluttered, and often confusing in terms of behavior
This reminds me of Ross Scott's video on the GUI, and i do really like the approach of reinventing the wheel and trying to make it better
Ideas are good, but what i would like is one of these times a programmer gets their hands on some of these concepts and tries to actually implement them, seeing how good these workflows are in practice
I love UX discussions, we don’t have nearly enough of them
Just because you don't like them doesn't mean they suck. All of them are there because they're the best options.
No, they’re here because of inertia. No one will ever convince me that the system tray is a good solution to handle background apps
@@TheLinuxEXPDone correctly they can convey a lot of information with a smaller visual footprint, i.e. different colors or icons for different service states. Your proposal makes them less space-efficient and downgrades their visual priority. As far as info the development of a dedicated notification sidebar/popup system seems to bridge the gap for providing more detailed information about updates.
@@TheLinuxEXP are you a gnome dev?
@@TheLinuxEXP Well, widgets on the desktop truly suck in comparison. They're hidden most of the time! In the tray they can convey just the most useful info, then you can click to see the full widget. A lot of other background information is conveyed by notification popups, which aren't so bad if you configure them to go away quickly.
Anything is better than a widget you can't see _anything_ of till you hide literally all your programs.
Gnome and Apple system tray are useless, KDE is the only acceptable one because you can actually configure it. I like your idea though.
cinnamon, mate, xfce, windows nt 4.0 and newer system trays?
I never really thought about the inconsistency of system trays, but you're right. I almost never know whether I'm supposed to left- or right-click to do my desired action.
6:45 You reinvented Microsoft Office Word with OLE technology
I'm from the days of the command line so I'm always converting applications, files, and the "desktop" to where they actually are in the directories (not "folders") on the hard drive. The desktop metaphor always felt at best patronizing or at worst condescending to using a computer. I don't think I can change. Sure, a GUI is pretty, but I still want to know where everything is, especially for organization and space conservation.
Microsoft has been pushing people off the desktop metaphor for 20 years at least, and onto sticking stuff in Documents, Pictures, etc. I never put anything there because it becomes hard to find and cumbersome to access.
I too, was a "directory" snob for a long time. 😂But eventually you gotta give in and accept "folder." 💔
"Paper workflow is almost done" is one of the most laughable statements ever uttered
Do you use paper and pen in your daily life when working at your desk, or a computer? There.
@@TheLinuxEXP Primarily pen and paper for journaling the work done but that's a blue collar job. We do use a computer though when working with stuff from like 2 customers because they have rfid tags so it essentially is there to run one single app for half an hour and be a music player for the rest of the day. And for the majority of the times I have to interact with desk workers they tend to use both paper and computer
@@TheLinuxEXPso many people do! I literally work with them. there's a lot of paper still moving around and I highly doubt that the bureaucracy in your country has managed to move behind papers
@@TheLinuxEXPI am a game developer and I as well as the rest of the programmers in my office always keep a pen and a notepad around for writing down tasks, drawing stuff, etc. It's just incredibly intuitive and fast to use. Not sure how all of this applies to an argument as a whole, but I don't see pen and paper going away anytime soon
@@TheLinuxEXP Yes. I use a Kobo e-ink reader with a stylus as a notebook. Previously used a Rocketbook with erasable ink.
I get your point on the task oriented approach and it make sense, but I think is simplistic approach, in the sense that while those tasks struggle to accommodate a complex workflow as document edition it will quickly turn into an app as we know it. Optimizing the task experience is what we call now an app. Gimp moved from the floating window/widgets into the window, we might explore there on the reasons, maybe just users wanting something they know but to me was a mess to find options in the old Gimp.
I actually love this idea of rethinking how we use our apps, and you make some great points here.
Personally I'd like to be able to work in a more task/project-oriented manner, so for example fire up all the necessary apps/windows to work on a given thing and then close them all down in a way that can be restored again later for future work on that task/project.
I can approximate this using e.g., "workspaces" in VS Code, multiple profiles in Chrome, virtual desktops, etc, but this is still pretty clunky. I'd love to have something that works for all apps globally, streamlines context switching and keeps resource usage to a minimum.
If it ain't broke, don't fix it.
Couldn't disagree more with your statements. You basically say let's throw 40+ years of UX out the window in favor of an experimental idea that it's in your head, which you don't know 100% how it would be, and the only way for you to convey it was through a quick, painful to look at draft.
I like your idea of the workflow-based environment. I'd love to try something like this
Your opinions are as insightful as your ideas are creative. I'd be very curious to try out even those suggestions I strongly disagreed with.
Great video overall. I'd love to watch a deep dive into any of these particular topics (say, how launchers have evolved and how they're done by different desktop environments), as well as how you'd go about redesigning them in more detail.
The real world is the most natural UI and has the best UX. The future is offline. I love how you’re thinking! Let’s collaborate and make a new OS that outclasses everyone else.
Not really.
Unfortunately, I can't make nature check my work to see if it functions properly, not without _magic_ being real.
wow. the desktop metaphor is to me like the paper clip: it's done. unless the hardware changes (we stop interacting with computers using 2D screens, keyboards and mice), the desktop metaphor, with its buttons, menus etc., is as good as it gets. not the windows/KDE style, though, the GNOME style.
Yeah, you’re probably right
Yes the Windows/KDE style. Stop shoving global menus and large buttons down my throat. Ever stop to consider I want things small instead of modern and diverse? Vote yes on California prop 8.
@@supercellex4D sure, if you prefer to live like it's 1998, you're free to! ;)
@@danielalvesldiniz 1998 was a better time. There was no COVID, no spyware in the OS, no IME or UEFI, the music was better, CRTs had better response times than LCD, programmers coded with optimizations in mind, and there was no shadow-censorship like what RUclips does. I’ll happily live in ‘98, and even suggest you come with me!
A few thought:
1) In terms of your "taskflow paradigm", your "AI" button where you type commands into suffers from the same issue of start menus. What happens if I want to add a drop shadow but I don't know how to express that. What happens if I need to find that one printer setting to print in CMYK but all I know is RBG. Menus are bloated, poorly optimised and organised but lets face it, a great many programs are complex. And you're really never going to be able to get away from that. Nor should you.
2) The "taskflow" paradigm also suggests greater interoperability between apps to the point that you don't have apps, and that is just never going to happen. Which means, in a way we already have something similiar to this in Tiling Window Managers which can be configured to always display certain apps on certain virtual desktops but then enable Window Tiling for the things that pop up out of the blue.
3) We do have examples of trying to get away from the desktop paradigm. Gnome and Android phones/Chromebooks. Thing is, Android Phones and Chromebooks don't multitask well, due to screen space so they are kinda designed not to be able to have multiple windows open at any one time. Where as on a PC it's required. And Gnome, well you either love it, or you hate it. Oh and remember Windows 8?
4) Tiling Window Managers show that you can work in a completely different fashion being more keyboard driven, heavier emphasis on the terminal and terminal tools but even still you have windows, menus, launchers and even system trays. So I posit that these elements aren't bad, they just need refinement.
One of the hurdles of changing the desktop is familiarity. I would say most people know how to use the current desktop because that's the way it has always been and trying to change that would be very difficult. If you make too big of a change then it'll be too confusing and disliked but when you take a more subtle approach then you just end up with an incohesive mess as it slowly transitions into the new method.
I know there's lots of disagreement about some of the ideas presented but I do think it's really healthy to take a step back and to analyze everyday processes. It's good to question the way it is when being told simply "because that's the way it is".
I hope you don't get discouraged from questioning the status quo. It's something really important to do and I think everyone should ask themselves if this really is the best way of doing it.
I'm sure if the computer interface was created today then it would be different, smartphones may be the best example of that, but trying to create an interface that is tailored to every single person's workflow is just not feasible. There will always have to be compromises and sacrifices.
The single biggest thing is that there are too many different ways that people will use their computer, too many different tools, too many different processes that people use to accomplish what they're trying to do. And even then, the same person probably won't be using their computer the same way every single time they use it.
That's probably why the desktop starts blank, to give the freedom to create it the way it needs to be setup for each users workflow.
Honestly, I think the answer is probably spatial AR interfaces.
The App metaphor still makes sense, because we interact with the world using objects. Picking up a book, controller, basketball, getting into a car, etc. Most things are objects for specific purposes.
The limiting factor here is the "desktop" itself. Having a cost and energy effective AR solution would eliminate that restriction (or at least improve it to the point that it's restrictions are similar to all the other objects we use).
All of the current elements do seem unconnected and take up lots of screen space. There does seem to be a good reason to make things more efficient.
The existence of program like krunner and powertoys runner fixes most of my problem. I just need runner and system tray on my desktop, i don't need launcher or taskbar. It can also find files in indexed folder. What's nice with krunner is if i want to launch a music player i can just type "music" and find the app because it not only search for the app name but also app description.
I feel like the future interface is runner + AI suggestion. The runner will show a few item list of things you might want to do.
edit: sry i commented before i watch the entire video lmao. 11:32 yes, this is what i have in mind about the future interface. System-wide AI with just button, search bar, and voice command to interact. Simplest interface.
While the "all-encompassing program" paradigm might be heavy and unfeasible for most applications, I could see it being used for some simplified general-purpose office suit. In this scenario, you'd open, say, LibreOffice and be greeted with a blank canvas in which you could add a "spreadsheet element", a "text element", a "picture element" or what have you, and be able to configure and edit it according to your needs. In a way, LibreOffice Writer and other editors sort of already do that with OLE/Formula as objects _inside_ the text document, so the jump to _juxtaposed_ elements shouldn't be too big.
Hi Nick, I don't think the desktop metaphor is that bad. If you sit at a clear desk, you would place objects on the desk along with the tools you are going to use, then return these thing back once the task is done. The problem seems to be the getting and returning these things back to their proper place once you have completed the task. My workflow is to only put shortcuts to files and tools on the desktop. A better shortcut would be useful. A shortcut that not only points to the file your working on, put includes other information such as tools, tasks, temporary meta info etc. Once the task is finished you can delete the shortcut and the files (data) remains in the correct place in the file system. You could actually call this advanced shortcut a "Task".
I have started making notes of all my bugbears and hope to start writing a specification of how I would like things to work. If it reaches a certain level of clarity, I might kick off a project to make it real.
The “Simple file with add-in file types “ paradigm reminds me a lot of piping, which in many ways is a command line paradigm that died with the creation of the desktop methaphor
.... And maybe we could select our UI Blocks from a list, or if that's too wordy, perhaps a menu of UI Blocks icons, with a handy way of keeping them organized by categories (Multimedia, Office, Productivity, Graphics...) or have a little text box that you can just type in the name of the UI Block (which would require you to know it, of course.) 😅 And we are right back to where we started!?
I like the idea. But I don't think that we need to do away with everything. What we need as you pointed out is to better use what we have. And in my case do it better with multiple monitors.
For example, I have a main panel with launcher and activity pager on the left, time and day in the middle, and the tray on the right only on the main screen.
I have a secondary panel on the bottom with a twist. I use activities to group home and work tasks from research to fun. Each activity has a unique screen on each monitor. And the bottom panel displays the specific apps for that activity.
The next level has been problematic, for instance, I am using Dolphin Bookmarks to group my work files for easy access. It fails when I change the folder or move a folder to a more optimal place. I have yet to figure out Virtual Desktops.
When I back to one monitor while traveling, I use tile manager.
As for AI, I am still exploring how it might serve. Perhaps GPT4ALL might serve.
Really miss the Windows XP and previous start menu. Nothing stuck in a tiny box to endlessly scroll. If I have hundreds of entries, it'll just open a new page next to the old one, both spanning the full vertical screen space.
I'm happy you mentioned the menu as the "command palette" is something apps have been using for many years now so it's disappointing no one has really put any effort into implementing this in a distro/desktop, unless I've missed an applet somewhere then someone please let me know because I'm desperate for this although I'll have to give unity another look now.
IMO the current standard file menu is one of the last remaining really old fashioned paradigms from a time when you used to slowly hit "Start" -> "Programs" -> "Application" etc to get where you want vs an app launcher like krunner/ulauncher/spotlight.
For system tray, every app should have a colored and a black and white version (both dark and light color for it to be adapted to the curent theme), so that they all use the same style (all flat black and white or all colored depending on user choice), except if we add an exception for one (the OS or the user do it, not the app itself). One useful exception of putting color is the one of screen recording, but it's a display option we can change, not the only available icon for the feature.
It's an interesting take. As a programmer I'm going to review them based on feasibility.
Global AI integrated text-based menu search. Very good, reasonably easy to implement. I use text-based search for IDE actions and I wish it was OS wide.
Task based workflow. That's nice, I have a post on the KDE discussion for the future of Activities where they have an integrated lifecycle. A set of 3 apps can be started, put to sleep and woken up for example.
Changing filetype as you go looks like an incredibly complex integration between standard apps. Either you lock people into using certain tools like apple or need the FOSS community to agree on how to handle files. Looks unlikely but I'd be up to being proven wrong.
System tray is not the best, gnome approach is in the right direction and this should get standardised in KDE at some point.
This definitely got my brain going and I'd love to see some of these features come to my OS.
I really like the idea with a task or maybe file based approach where you create a file and then basically can edit a file in one program and then seamless transition it to another program, where the file position itself doesn't change, but only the tools you have access to.
Kind of like piping data between different commands on the command line.
Though I have the feeling that interoperability and standard formats would be really hard problems to solve.
My guess the best way to implement this would be to have one center app that handles the file management and then you can integrate different apps like plugins.
Also there should be an option to group the different files with each other and maybe with tasks.
I imagine for video editing you have one main file with the actual video being edited and you can then attach different assets like clips or images or audio etc to that task.
I know that already kind of exists, but the special thing here would be that there is a unifying and standardized interface that is common for different programs instead of every program implementing it themselves.
That sounds like a great idea. I'll have a few questions, though. How does one share the data with someone on a traditional paradigm?
It would also mean your developers would have to implement the favorite features that come out within your widget app. It will have to be done periodically.
You would have to create your own data file structure and conventions to store your data and interoperate with the world.
Those are the biggest areas of contention that I see of the top.
Still, it's elegant.
The desktop paradigm has been around so long because, it just works. The DE designs have tweeked over time but it still just works. It allows the user to keep things grou[ped in ways they can find things rather than having to hunt and poke about and trying to guess what tools are there by trying to type descriptive phrases which is barely more useful then using a terminal. The Gnome paradigm just throws bigger icons all together in a mess you have to go hunting for.
A lot of these issues can be solved by a tiling window manager and a launcher like dmenu/rofi/wofi imo. You can even have the sort of layouts predefined and set up using workflows. I feel like the centering the OS based on workflows and not based on the applications is really interesting and very nice but there is a practical issue that I foresee: we are pretty much flipping the workflow cycle on its head where we think of workflows and design applications to fit that workflow rather that the older way where we have applications which expose features and we fit our workflow around it. The new way would make development of features and workflows extremely difficult as the individual components would have to be general enough to be reused and there would also have to be people with "grand visions" to set the workflow going.
However, I wouldn't complain at all if someone figures out it all and if we have this new paradigm to use. It will be amazing. (We who live in the tiling work already enjoy a lot of these similar usecases. Especially the people who customize their Emacs/Neovim installations so much that they possess bits of this workflow already :) )
For someone who is never satsfied with stuff im quiet pleased when stopped distro hopping on Pop os. It just has all i need as a desktop.
Tiling with gaps, a minimalistic top panel and the workflow. I feel like the design is extremly pleasing for me. Hope the new cosmic desktop will be just as great or even better.
The great thing about the HUD is it could integrate with LLM assistance, but also integrates nicely with global menus and app menus. I really, really like the traditional menu metaphor for the most part, but certainly appreciated being able to hit one (consistent) hotkey and search for a menu as well. If you don't want to break people's workflows, having something that can incrementally improve upon existing paradigms *and* hook into cutting edge innovations seems key.
The problem of course is that even in the Compiz days not all apps played nice with the global menu, a situation that is far worse in 2024 as the Age of Electron continues. Actually, a good way to get me to *want* an chatbot integrated into my desktop would be as a HUD/Global Menu support for apps that don't have one.
I really like how you start with a file-based UI, and then switch to a task-based UI. As others said, your approach with files is nice, but it's just very limited. A lot of activities don't need files.
So I think the task-based UI or activity-based UI is better. It resembles how you'd act in the real world, where you want to draw (activity), not to "edit" or "create" a drawing (file), nor to use a pencil (app) (which is the current paradigm, btw)
The current paradigm focuses on using tools, and your paradigm focuses on editing files, while the "right" paradigm would be focused on doing things, no matter if that involves editing files or using tools, or both.
I think your task-oriented computer interface is really interesting!
I might have to apply my visual design skills and make a flashy mock-up of what that would look like and how it would work! Please let me know where I could share that, so you and others could see it!
It actually reminds me a lot of the ways old GUI computers used to work like the Xerox Star and Apple Lisa.