Right, this seems to me to be a tiling window manager with a touch interface. This is kind of an odd combination too, because most tiling WMs are keyboard oriented, whereas most non-tiling WMs are pointer oriented. Maybe that's just historical and doesn't really make much of a difference though.

But with i3, I have keybindings to "open the last browser window i used" or "open the last gvim window i opened." I can tile them in different workspaces as I please extremely efficiently and quickly, and move them around as needed. I can move much faster using the keyboard than when I'm in my Windows machine. Having this tiling ability on a tablet seems odd to me.

I am kind of getting tired of the "attacks" on the traditional desktop interface. People just don't want to admit that just because tablets exist they are not a superior interface. Just a different one.

The interface they propose is beautiful but it would really have to walk a thin, thin line between usability and screen space.

I do like the idea of a tag-based filesystem though. Obviously then you get into the problem of managing your tags (a tree-based filesystem solves this problem naturally).

I would personally kill for better mouse integration in i3 (which I use on all my daily-use machines), and I would be extremely happy to augment it with touch screen support (which I would largely want to use for tile manipulation).

In fact, a main reason I use i3 is because it is literally the least ideological tiling window manager when it comes to mouse use. The productivity gains I get out of tiling don't come from using the keyboard, they come from not having to deal with the (imo, especially in the post-1024x768 era) very messy overlapping window model. Things are where I expect. I can move them from one screen to another and sane things happen. Plugging in and unplugging a display doesn't make hash of my organization. Things are never buried under incomprehensible layers. And so on.

> I would personally kill for better mouse integration in i3

I agree, it was one of the things I was missing back when I was using i3 + Ubuntu as my daily machine. Now, I'm using OS X with Amethyst[1], which while it's less powerful than i3, combined with OS X's window management features I honestly feel like I get the best of both worlds. It's quite great, to be honest.

[1] https://github.com/ianyh/Amethyst

> I would personally kill for better mouse integration in i3 (which I use on all my daily-use machines), and I would be extremely happy to augment it with touch screen support (which I would largely want to use for tile manipulation).

This sounds interesting. Would you care to elaborate?

I know at least KDE's Baloo indexer uses tags on files when available. If I search for the word "cookie" I get every filed named cookie, anything tagged cookie, contacts, bookmarks, and any parsable text document containing cookie including emails and IM logs.

I think the broader problem is that there is this awkward question of where tags belong in the application heirarchy. Right now most media files embed tags while text files do not. Binary files certainly do not, and folders are not true files and thus are not on-disk things to put tags on, they are usually filesystem constructs in a data tree somewhere.

Your file indexer could maintain the tags database, but then they are not file-portable. They could be part of the extended file traits in your filesystem, but then filesystems lacking support (including online uploads) would drop them. Thus we are at a point where tagging is done on a per-format basis, and thus some files are just not taggable but instead require indexing.

You would probably want to throw a semantic relevancy engine on top of that to associate vocabulary with similar terms, so if i search for food i can get things like cookie as a related search term.

In case, for the tag-based filesystem, there are tmsu [1].

I will try it soon with ranger [2], I think managing tags with it won't be a problem as you can call the shell where you want.

[1] http://tmsu.org/ [2] http://ranger.nongnu.org/

Looks great but what happens when you move files from one system to another? Can one keep the tags?

It looks like it's "hard" to do correctly, so TMSU doesn't do any magic but will try it's best through the "repair" command:

https://github.com/oniony/TMSU/wiki/FAQ#why-does-tmsu-not-de...

>I do like the idea of a tag-based filesystem though. //

Ditto, I use tags heavily for web content and photos and would love to do the same for local files.

Seems that it might be possible to overload soft-links to do the job - create a hierarchy of tags and then put a soft link for each file in the relevant place. That way you could use current filesystems and filemanagers whilst you develop the concept and apps.

Someone mentioned file searches in Ubuntu (using Baloo) but the default file manager needs to have an interface to apply tags as a MVP og file tagging.

I want hierarchical tags, e.g. "comp.lang.ruby.rails" instead of just "ruby" and "rails". And of course, when I search "comp.lang.ruby", I want to see results from all of its children.

The problem with traditional hierarchical organization is that objects often belong to multiple categories. With hierarchical tags, that problem disappears because you can easily assign multiple tags to the same object.

Actually, we can already do this with hierarchical filesystems and hard (not soft) links.

Do they really have to be hierarchical? Your example would work with AND.

What if I have two dozen ruby-related tags, not just rails? In order to search all of them without hierarchical tags, I would have to either add the "ruby" tag to all of them, or run a search with a lot of ANDs.

I already do this to some extent with my 11,000+ bookmarks on Pinboard. Entering the Usenet-style tags is really easy thanks to autocomplete, which also helps me avoid ending up with misspelled tags. Pinboard currently lacks searching by tag prefix, but it's trivial to script it locally using the API.

Talk of tag/DB based file systems always reminds me of WinFS[1]. Microsoft were trying to build a tag (well relational database) backed file system since 1990... eventually abandoned - presumably due to back-compat issues.

1. https://en.wikipedia.org/wiki/WinFS

Lots of reading material about the conception and death of WinFS: http://www.metafilter.com/134799/WinFS-what-it-could-have-be...

This is exactly the idea of Desktop Neo. If tablets and phones are used for consumption, the desktop can be focused on productivity.

I don't want to replace the complexity of the desktop with interfaces that are easier. I want them to be better, to be more efficient for professionals.

Concepts like panels, tags and touch input are probably way harder to understand than windows, folders and mouse input. But it doesn’t matter, because professionals spend so much time on desktop computers that it's worth the effort to master the interface.

> I want them to be better, to be more efficient for professionals.

This is a good goal. But gestures and multitouch will not lead you on the path to that goal. Consider tiling window managers like awesome or i3, that already today help programmers (myself included) be more efficient. They achieve this by reducing time-consuming dragging, dropping, resizing and clicking actions either by replacing them with keyboard shortcuts or simply by automating them away. A UI where you have to move your hands even further (distance keyboard-screen > distance keyboard-mouse) will actually decrease productivity (and likely cause RSI problems as well).

I realise this is a mockup/concept design. But if you want to pursue the idea further, I very much recommend using a tiling window manager for an extended period of time, to learn from that paradigm.

The mockup interface of i3 + graphical rofi[1] + touch/mouse integration.

This is a very powerful paradigm to pursue: An intuitive version of tiling WMs that appeals to everyone. Go, ziburski, go!

[1]: https://davedavenport.github.io/rofi/#window-switcher

I do think you are on a good path with tags.

Back when I went to Gmail from a folder type email representation, it was hard at first. But search proved superior. Basically, a lot of organizational time investment had no return. Getting better at search always has a return, and you save all the laborious folder management.

I've taken to one or two big directories filled with reference material, docs, pictures, etc... search generally pays off.

Smarter tagging is likely to work in similar ways.

My beef with search-based, tag-based UIs are they are not great at browsing content. Sometimes I want to just review and look at older content, not search for one particular thing.

I will plug in a date, but yes, agreed.

As a developer, I unfortunately have to live in a fundamentally different paradigm where file systems are inescapable (unless OSes change dramatically under the hood), keyboards are essential, etc. etc.

But when I'm not trying to write code I can see this working.

This is a long response, but I think it might be useful to read.

There are a lot of attempts to redesign the desktop or do some clever thing to increase productivity but all of them tend to forget the fundamentals of interacting with the computer. The forget why what we have is successful and seem unclear as to what they are trying to move towards as an improvement.

That probably seems like a bold statement so lets look at a few things.

Lets look at why something like paper is still around. Do we care that one piece of paper might hide behind another in a stack of papers on a desk, is that a terminal failure of the medium? I'd say no. Just like windows hiding others, it's not a fundamental issue that needs solving.

Why else might paper be successful? I'd suggest it's not the paper itself, it the way we interact with paper. We don't use our hands to mark the paper, we use tools, we use pens, brushes, pencils, chalks and all manor of things. Paper is a tool that allows us to get the best from a range of our 'marking' tools.

Like wise, if we are to interact via touch with the desktop, it stands to reason that to get the best from the experience we need to have a range of tactile tools to do so. We need our keyboard, our mice, our touch pads, touch screens and stylus'. Concomitantly, we need our desktops to allow us to get the most out of these interaction tools.

To my mind, the problem with the desktop isn't that it needs a fundamentally more restrictive use pattern, it needs a better use of the available control mechanisms.

Let's think about why Vim is so enduring and successful. Vim only does one thing, manipulates text and it's got a mode of operation totally alien to many who've used a text editor. Yet for decades now it's what many gravitate towards as the pinnacle of interacting with text on a computer (shout out to Emacs folks, this is just an example!).

Vims's method of operation seeks to ruthlessly release the power of the keyboard to offer huge benefits in terms of productivity and manipulation.

What is your design seeking to ruthlessly release from the power of our interactive devices? Why are our existing tools better for interacting with your design than a traditional desktop? what power are you releasing from them?

All of this is to say that whilst you may redesign the desktop and have clever (and useful) schemes to increase productivity, these designs don't pay enough attention to the things we use to interact with the computer, and so don't offer true lasting benefit and therefore don't catch on or endure.

If you redesign the way the desktop works, you need to either re-imagine the tools we use to interact with it or take measures, without pity or care for convention, to release the power of the existing ones.

Apple's done an interesting job in this area in it's move to bring touch gestures into the control of the desktop, but I'd don't think it's hardly been a fully realised opportunity.

polite spelling correction: it's "all manner of things", not "all manor of things".

Ack.

I can see the reason from the power user's point of view. I've noticed that quite a lot of power users don't use desktops (or window affordances) at all! Their only use is to provide a place for a cool background picture. I see a couple of reasons:

- having to use a mouse (or, God help us, a touchpad) is annoying when you have to switch between it and the keyboard

- using the mouse is slow

- windows on traditional desktops take way too much space (compare with how much better they look on tiling WMs)

- ALT+TAB (and equivalents) suck - you can only switch back and forth between two applications; anything else requires paying attention

- window locations and sizes are ephemeral, which again is annoying when you want to set up a good working environment

- wheel menus are awesome and the fact that pretty much nobody is doing them baffles me

Tablets suck for creative work, I agree, but the way I see this project is not as a nod towards touch interfaces. On the contrary, it's about using sane(r) window management, combined with tags, eye-tracking, voice control and wheel menus to minimize the amount of effort you need to make to issue commands or search for things. It sounds like a perfect addition to keyboard-driven work style.

In my world (music, concert, theater production) we depend on a lot of equipment which is basically a computer and a custom physical control surface, retailing for 10x the price of the computer. The keyboard and mouse interface is simply not good enough to replicate faders and knobs. The tablet interface, on the other hand, is. This has important implications for light boards, sound boards, and DAW controllers. For many of them the domain-specific IO by itself isn't actually that expensive and the software can run on commodity computers. We were only locked into custom hardware because we needed its control surfaces to work efficiently.

Existing desktop workflows may not be improved, but existing custom-control-surface workflows can be made vastly cheaper and more flexible, consolidated onto relatively commodotized hardware, and in some niches this is happening.

There's a lot of this sort of thing showing up in the DJ/VJ space and with visual and audio performance in general.

I think there are definitely strengths and weaknesses to the use of touch-oriented control surfaces. On the negative side, they require you to look at the control surface a lot more often than a setup featuring jog wheels and sliders. But on the positive side, being able to customize the control surface and allow a level of more "analog" control compared to pointing and clicking or typing.

There are a lot of less expensive options that use more common USB interfaces to map hardware controls to software functions which can really help if your only other option is that pro (read: expensive and proprietary) gear you mentioned. Still, you're always limited by the number and layout of pads, buttons, sliders, and dials that can be mapped to software functions.

I've played around a bit with software like TouchDesigner which is meant for creating custom touch interfaces and programming the ways that each control affects parameters in software. It's like a more open and flexible alternative to the old Crestron and similar systems where you could define the control surface in software but were limited to their processors and peripherals.

It's something I don't really do much in my work (since we use those aforementioned Crestron-type systems) but in my spare time, I like playing around with using software-defined interfaces to make interactive audiovisual projects.

I'm surprised that no one's made a small add-on control with faders/knobs to work with Photoshop for font-sizing, opacity, and all the other things usually done with the tiny palette interface.

Someone pointed me towards this: http://palettegear.com/how_it_works

I think some of that results from the limitations of mobile OSes: everything is tied to an app store, jailed, etc. A mobile device is not a "real computer" not because of any hardware limitation but because the OS is designed for it to be a dumb terminal to access the cloud and consume content and rapid-interaction services.

It maps well to the sorts of things you want to do with a tiny device in your pocket, but it's not suitable for producing anything other than selfies.

This is the hole in the "mobile is the future" argument. Mobile (as it exists today) is only the future if nobody has anything substantial to say or create and we are all just passive consumers.

This is really great desktop UI work and I do hope someone takes a look and gets inspired. It would really be nice if Linux desktop efforts stopped trying to deliver Windows 95 or early versions of MacOSX in 2016 and actually innovated.

Also, the screen is smaller.

A lot of tiling problems on the desktop would disappear if monitors were two or three times (or more...) as large on each side and wall-mounted.

Being able to see most of your windows at the same time and work in any one without moving or switching would be revolutionary.

I'd never heard of GUI/10. I don't particularly like the windowing scheme, but I think a ten-way touch interface which flipped between being a multi-pointer and a text keyboard - possibly with a touch point in one corner to switch modes, and not that show/hide thing tablets do now - would be a very interesting thing.

It wasn't practical when 10/GUI was first discussed. It's getting more and more practical now.

>It would really be nice if Linux desktop efforts stopped trying to deliver Windows 95 or early versions of MacOSX in 2016 and actually innovated.

Any innovation has to be much better than current practice. If it's "interesting, but..." it's not enough to get people to switch.

Talking about Blender: I think that the way windows work in Blender could make a great OS window manager. Devide the panels the way you like and select a program or part of a program you like to show inside the panel.

There are window manager who can do this but imho not as simple as Blender.

It would definitely take some changes to make it work at an OS level, but Blender's modular window organization system would be awesome for more uses! It definitely is power user only, such a UI style would overwhelm most users with the raw number of choices.

What do you think generates the most money for Apple and Google... You being reduced to a consumer that buys movies, music and apps in their stores — or you fiddling around in Logic or Photoshop?

This is a shameless plug, but I've been developing something involving 'scrolling with a resting thumb' to bring up a quazimodal menu, intended for applications where one is creating content in a spacial context… (i.e. here's a gif demo with Photoshop: https://twitter.com/vivekgani/status/659437589896630272 ) The intent though is for folks that want some middle ground between having to remember a bunch of keyboard shortcuts and mousing out to click something. This is an actual working app that works with Sketch & PS and more apps soon - if you'd like to try it email [email protected] and I'd be happy to share it.

The OP had a similar thing with the radial marking menu, though I've shied away from that approach since it doesn't scale with lots of icons/options unless you do hierarchies - also I simply just wanted to try something different. One random thing I stumbled upon yesterday is an old Autodesk Research video looking testing these versus other methods: https://www.youtube.com/watch?v=YHZB0d20640

I think this is the best objection I've seen to what could otherwise be a pretty cool idea to try and refine.

The answer to designing complex, professional-level applications in that paradigm is "make it fullscreen, and use a keyboard and mouse". Tablets just don't give you the precision and screen space to work with extremely complex tools.

I think Windows 10 actually manages this well. I'm using a Surface Book, and despite its various issues, this tablet/notebook hybrid actually works quite well. Plug the screen into the base and I have a full blown, high res screen with a keyboard and dedicated GPU for my coding/Photoshop/Blender. But if I just want to consume, I can pull the screen off and sit on the couch.

> Why is everyone out for the blood of the desktop/windows paradigm?

Honestly I think a large part is because microsoft dominations the desktop market and is failing in the tablet market.

Suggest looking into the work that the PennyArcade guys have done with Microsoft to make the Surface Pro good for illustrators.

Agreed, the folder metaphor works as its a direct visualization of the hierarchy of the computer.

I remember what Yahoo! was originally like: a nice, hierarchical categorization of websites. That was great when there were 1,000 websites. Maybe even okay for 10,000.

Google quickly became an obviously superior approach.

Back when Yahoo! was a categorized directory of sites, my computer had probably 1GB of storage. I didn't have a digital camera or many thousands of email messages. I didn't have as many documents that I either created myself or was given by someone else. I didn't have ebooks. Or mp3 files.

The hierarchy that we have is not necessarily a given and even if it exists on disk, the UI metaphor does not need to be tied to that paradigm. Perhaps there is a better way for the things we do today?

I think the central problem is that few userspace programs get as much engineering and tooling thrown at them as the filesystems. Itunes and similar apps which implement their own abstractions for organizing content have always trailed the filesystem for me in the following aspects

a. Hard to access the data outside the single ui

b. Performance of data import / export trails raw folders / file systems

c. Hard to figure out how to atomically backup the content in an incremental way.

d. No programmatic access.

Perhaps, but I think the answer lies more in backend tweaks than a UI/UX overhaul. File/folder tagging, for example, could integrate easily into existing UI paradigms without having to overhaul (or remove) the files/folders analogy. I agree that dealing with folders gets too complicated above a certain threshold, but to me that signals a need for additional (thoughtful) tooling.