Hacker News Reader

I'm reminded of a tidbit from an old AskTog[1] article:

"We’ve done a cool $50 million of R & D on the Apple Human Interface. We discovered, among other things, two pertinent facts:

* Test subjects consistently report that keyboarding is faster than mousing.

* The stopwatch consistently proves mousing is faster than keyboarding."

There are probably all sorts of caveats to this (and the research may have changed since then) but I'm reminded of it whenever I see people extolling the virtues of their ascetic, keyboard-centric dev environment. I'm not saying they're wrong, just that in lieu of benchmarks and studies, I'm skeptical.

[1] http://www.asktog.com/TOI/toi06KeyboardVMouse1.html

Interestingly, I don't feel faster with the keyboard. I do, however, feel more comfortable when I stick to one mode (keyboard-only or mouse-only). It's switching that feels awkward. I also don't use Xmonad out of pure Geekiness (though I started that way). I use it because I find it more comfortable. I work at the same time with OSX, and while moving windows with the mouse is probably just as fast, it annoys me.

Now the reason why the keyboard is slower than the mouse may be sub-optimal short-cuts. For example, when editing, one very often needs to change lines, or otherwise navigate the text. A sensible short-cut to support that could be alt + jkli (assuming a Qwerty layout). Then you don't have to leave the home row at all. But if you have to use the actual arrows, it's slower, and that may make the mouse the better choice.

You seem to be describing vim's hjkl (left, down, up, right, IIRC)

I am, except modifier key, makes the trick possible on (customizable) mod-less editors. The idea is not mine, by the way: http://xahlee.org/emacs/ergonomic_emacs_keybinding.html

Also look at the Nipple-mouse. (The IBM Track Point thing on Thinkpads). It's placed so your fingers don't have to leave the home row.

I think the caveat is primarily muscle memory. If an action is in muscle memory it's almost always faster. I don't think about how to switch desktops or save a file so the keyboard is faster. However fancy cursor movement(like skip ahead one word) is pretty much useless for me since I have to think about what I'm doing.

Sure, a mouse is faster if all you're doing is pointing at stuff and pushing one of a few buttons. Studies will confirm this. That's pretty low-level, though.

A keyboard is faster if you want to, say, "jump to the head of the definition of the function body you're currently in and send it to be evaluated", because a mouse just doesn't have enough buttons. With a keyboard, you can name operations. And then name new operations, grouping those names... (Why yes, I do use Emacs.) If I was using a mouse-and-menus for that stuff, the menus would stretch to the moon.

The next time somebody argues that a mouse is faster than a keyboard for programming, I'm going suggest they make their case solely by pointing at things and clicking their tongue.

Sure, editing is one of the exceptions Tog mentions. (I'm an Emacs user, too.) My point was that just because you think or feel that something is faster doesn't mean that it is.

Also, when it comes to programming, the biggest bottleneck for most people isn't typing -- it's thinking.

P.S., there do exist mouse-heavy programming environments: http://en.wikipedia.org/wiki/Visual_programming_language

I agree on both points, actually, but when people mention that study, they tend to use it as a "See! Using the mouse is faster!" handwave-y dismissal, without mentioning the fine print.

I'm not a neurologist, but I wonder how differently the brain's spatial centers and language centers interact with the part(s) that perceive the passage of time.

Also, when it comes to programming, the biggest bottleneck for most people isn't typing -- it's thinking.

That's so true for me. I'm not pumping out code nonstop where the efficiency of keyboard vs mouse comes into play. Being more efficient with moving around a source file only matters if I actually knew what to do once I get to a particular spot of code and many times I don't. I struggle more often with the high-level design of a program than the details. And when the design is sorted out then everything else falls in place. That's when more efficient navigation may be more important to me.

Mice are better for spatial navigation, but keyboards are potentially much better for structural navigation - whether you're moving by search (regexp find, or whatever) or language elements (navigating via tag stack, etc).

Agreed. Improve your mind and not your tools. Don't spend a couple of hours learning to use a tiled WM instead spend that time reading up on documentation. I believe that will give you the biggest speed gain.

I think there's more to it than just mouse vs keyboard. The keyboard-centric power interfaces like Emacs, Vim and those tiling window managers have non-intuitive shortcuts and combining sequences that really make things go fast. Compare using a keyboard with Emacs to using a keyboard with Notepad.

The same thing for those mouse-centric applications. Linux window managers have a "shortcut" of holding a modifier combination and dragging anywhere within a window to move it, or dragging the nearest corner to resize it. This makes arranging floating windows fast and fun compared to Apple's more intuitive requirement of having to drag the titlebar, or that tiny little handle at the bottom right of the window.

Some of the mouse-centric visual programming environments you mention, particularly the ones that have many years of use behind them, also contain shortcuts. Autodesk's Flame compositing environment for instance, traditionally driven via a pen and tablet, has a whole bunch of shortcuts to drive the interface without touching the keyboard. You can connect dataflow nodes by "kissing" for instance - dragging them close to each other, allowing you to connect many nodes with a single movement.

For many people, it's the mastery of these shortcuts - keyboard, mouse, pen or touch - that gives them a sense of pleasure and power when they use that piece of software.

For me, the best part of a tiling window manager is immersion. It's no-nonsense. There's no eye candy to remind me of all the brain candy out there. When your whole screen is nothing but black-and-white text and ugly emacs-colorized source code, it's easy to forget that Facebook and Hacker News are just a few clicks away.

A mouse is faster for unfamiliar tasks, but when an action is committed to muscle memory, keyboarding becomes fast.

I use both OS X and ion3 (depending on whether I'm using a Mac or a Linux box). Ion3 is a tiling WM and about as far removed from the OS X GUI as you can imagine. Nonetheless, my workflow in both is exactly the same: one large Vim window, one large terminal window. I never move the windows around once they're open, and I can switch between them using the keyboard. Once you've got an editor you're comfortable with, there doesn't seem to be much advantage to using a tiling window manager in terms of productivity.

Things like browsers are a separate issue, but if I'm switching to a browser I'm already distracted. :)

Plan9's acme editor was based on Oberon, FWIW - you can probably find more info there. I have a lot of respect for the design, but have never gotten comfortable with it. (I've used Emacs for years, vi for several years before that.) Its "every word in every buffer is also a command if you click on it the right way" seems like the best mouse/keyboard UI synthesis I've seen, though, if you're doing actual work with complex tools you can invest time in learning. A lot of GUI design seems optimized for people doing simple stuff and hoping the UI will be self-evident / "intuitive". I wouldn't expect AutoCAD to be "intuitive", for example - the problem domain itself is far too complex.

Also, I use dwm, and got a netbook recently (dual-booting Windows 7 and Debian). On a screen that smal, tiled wms really shine, and Windows 7's interface feels like a joke in comparison. The low-hanging fruit on a tiling window manager makes the biggest difference, though - having N different workspaces, automatically arranging windows, and being able to hit alt-3 to switch to the third workspace / alt-enter to open a shell is most of what you need to know.

Another thing in favor of a tiling WM (and a keyboard centric setup in general) on a netbook is that it is often more convenient to use a netbook without a mouse. I am happy to reduce my track pad use in general, especially hold-button-and-drag operations.

No kidding! The trackpad on mine is too close to the spacebar, anyway. I just have it turned off (Fn-F7) most of the time. dwm + dmenu + emacs has me pretty much covered for mouseless usage.

I was skeptical about netbooks, but saw a great deal on craigslist and decided to try one. With dwm, Emacs, and terminus, I have two 80x45 buffers, which is quite adequate. (I also bike nearly everywhere, and it weighs so little that I don't hesitate to carry it with me all the time.)

Like the author, I use the Awesome window manager. I think you have a perfectly valid perspective, but just want to speak for the other side.

The main benefit I get from Awesome is having 9 workspaces ("desktops"). I use my mod key (in my case CAPS but it can be Alt or anything) plus # to get to any workspace. I have web browser on one desktop, email on another, IM on another, terminals on another, programming on another, and it's really, really easy to switch around. This was a HUGE improvement for me over Windows (and even Gnome), just in terms of managing multiple windows/applications running at once.

Also nice to be able to get two windows to share the screen with a keystroke, if I need to look at two things at once. And in the same vein, tiling is really nice for IMs, because on the workspace for IMs, I can see all my chats at once. Normally for other tasks, like web browsing, I just switch into the tiling "mode" that doesn't tile (i.e. application is "maximixed").

Do you use a terminal multiplexer or something similar?

I only ask because my set up was exactly like what you just described. Once I buckled down and started using one, it negated the need for a lot of the tags I had with dwm.

I've never needed one, probably since a tiling WM allows you to accomplish something similar, as you point out.

I know screen has some nifty features, but for the things I do, a tiling WM is a superset of the functionality screen would provide.

Screen's primary payoff is when your connection to another machine goes down a lot for whatever reason. A good wm can handle the multiplexing better but they don't restore context like screen.

I love this! I use dtach (with dvtm) instead of screen, but it's serves the same purpose. I'll be working at my desk and then have to head to a meeting (with my laptop) to another floor. My ssh connection will drop but once I log back in I just need to reattach and I'm right where I left off.

I know, and I usually work locally.

I use dwm and tmux (rather than screen). My muscle memory knows that I keep a tmux session local to the computer I'm physically located at on #1, remote (if any) on #2, side tasks on #3-7, Emacs on #8, and a web browser on #9. (I use a slightly different setup on a dual-head station, but usually w/ two Emacs windows, etc.)

I recently started using Bluetile (http://www.bluetile.org/). It has all of the keyboard bindings, but there is also a dock that allows mouse control of major features. It's been a pretty easy transition for me because of this.

Something such as a unified desktop (OS X, Haiku, Gnome, KDE) will make you more productive rather than a window manager that optimizes for pixel real estate.

UNIX is my unified desktop. I only use the window manager to manage windows.

> Something such as a unified desktop (OS X, Haiku, Gnome, KDE) will make you more productive ...

Its not an either-or thing. XMonad works perfectly well Gnome, panels and all. It only replaces metacity/compiz, not the whole environment.

The same can be done with Openbox.

I am not so sure that makes it a good idea though. Why undermine something so deliberate?

I use awesome (a tiling wm) in a very mouse heavy way. Tiling does not mean you're stuck with using the keyboard. It does mean that you don't have to worry about managing screen real estate nearly as much.

Amen.

When are people going to get it through their heads that using the mouse is faster than the keyboard? This has been determined by Apple's UI research and known for some decades now.

Source: http://www.asktog.com/TOI/toi06KeyboardVMouse1.html

"We’ve done a cool $50 million of R & D on the Apple Human Interface. We discovered, among other things, two pertinent facts:

    * Test subjects consistently report that keyboarding is faster than mousing.
    * The stopwatch consistently proves mousing is faster than keyboarding."

This is the reason that the Plan 9 folks gave for making their UI mouse rather than keyboard based.

It takes two seconds to decide upon which special-function key to press.

Why would you link to something so foolish?

How can this be? I'm just not convinced.

The main thing I use my WM for is switching applications. Any application I want to switch to is modkey+number, and I know the numbers.

That's why you do research, because "It feels faster for me" usually doesn't lead to the best decision making.

The problem is research costs money, and people with money are more likely to be interested in average users. Nobody's going to do research on obsessive HN-reader-type users who are willing to go out of their way to memorize lots of Emacs commands in order to ensure their exocortex is fully optimized.

That's not an answer to my question.

Let me state my point more bluntly. I've tried switching applications with mouse and with keyboard, and it's faster with the keyboard.

"It feels faster for me" actually might be a decent form of decision-making. The bottleneck in software isn't user-input speed, so it's kind of moot. If people feel better using the keyboard for input, why not let them feel better?

Faster for what, though? If you're doing something best served by pointing at things, sure. Web browsing, sure. Programming? I don't buy it.

It's the mode switch that's frustrating. It interrupts my concentration when I discover I need the mouse, when all I'm trying to do is code.

I use a mouse when I have to, but I'd probably use it more if I could get one of those nubbin-mice in my desktop keyboard. 99% of my job is typing. Switching between mouse and keyboard is annoying.

At home, I tend to do the same kinds of things I do at work. However, I picked up a Logitech bt media keyboard with a touchpad off to the side--and that works fine. I wish I could find a wired version. Cherry builds keyboards that look worthwhile, but they don't sell them piecemeal to little guys like me.