[TriLUG] Who runs Red Hat and KDE

Ilan Volow ilan at clarux.com
Mon May 26 07:00:21 EDT 2003


On Mon, 26 May 2003 01:07:11 -0400
"Stephen P. Schaefer" <sschaefer at acm.org> wrote:

> 
> Brent Fox wrote:
> [elided]
> > If a majority of our users were asking for us to release new desktop
> > packages every week, I feel certain that we would do it.  The reality is
> > that most users are not asking for this.  And yes, we do feel the GPL
> > makes us different from other software companies.  I think our
> > commitment to the GPL is well documented.
> [elided]
> 
> I'll toss in my perspective as a happy Red Hat user.  I've been dealing 
> with Unix for more than two decades, and my attitude to graphics is 
> this: equanimity and detachment.  Yes, there can be some occasional 
> nutrition (kstars!) in the eye candy, but the effort is usually 
> extravagant for the obtained result.

The problem is the folks who currently program linux UI's don't understand how to use graphics, and often misuse them to "make things purty". The whole prettiness == usability fallacy runs rampant in the linux community. Confusing dialogs and inefficient layouts are merely given anti-aliased text and are then proclaimed "perfectly ready for the desktop". I think at least part of the problem has to do with the fact that most of the people who design linux UI's have their brains still stuck in runlevel 3 mode, and allow 30 years of command-line unix baggage to leak into the graphic interaction. That's why I think that the future of desktop linux needs to be based on the classic MacOS UI--for a long time it didn't have a command line, so you don't really have those harmful command-line influences. The developers of the mac had to think graphically, and it produced a good interface as a result.  

  The number one benefit for me of a 
> monitor over a character cell screen is that I can get 160x50 characters 
> instead of 80x24.  Emacs "windows" (i.e., boxes of text) get me 80% of 
> what I want from a user interface.  I'm reading/composing e-mail in 
> Mozilla right now, but my screen consists almost entirely of boxes of 
> text with borders.  But the fonts? you say.  Regardless of how superbly 
> the fonts are rendered, the whole monitor/keyboard business is tiring in 
> comparison to a well printed book.  Only the interactivity and immediate 
> access justifies the computer.

Paper is entirely non-modal, it doesn't have a bootup time, anything is undo-able (if you are using pencil), it doesn't require batteries, has instantaneous response time, and it has a resolution that puts the highest pixel resolution monitor to shame. It has an excess of the qualities needed to make a good user interface. I'm not surprised you find it better. 

> I'm not anti-picture: pictures obviously have their place.  But what 
> conceivable deep tweaking of the window manager could recoup the time 
> spent by an individual?  (Which hasn't kept me from trying.)  That time 
> *is* justified if a marginal improvement is enjoyed by a wide subscriber 
> base, but, of course, all the subscribers have to spend the time to even 
> learn about the change.  Further, I don't see much work justified by 
> fundamental human factors measurement, as described by Jeff Raskin in 
> "The Humane Interface" - or by any serious alternative methodology.  As 
> for allowing the programmers to be productive, Red Hat's theming does 
> not in my experience break the functionality of individual applications. 


As a student of human factors, and someone who's been ear to some of the HCI rumblings about BlueCurve, the problem with the whole "unifying the desktop thing" is that when you take a look at their modifications, they are not uniting anything, they are only painting over the differences with pretty themes. GNOME and KDE have different behaviors in different situations, and if you make it look like GNOME and KDE apps are the same exact one environment, you fool the end user into thinking that all the applications running under this "unified environment" will act the same given the same situation, when they really won't. If, for example, an action in one environment saves something, and the same action in another program automatically closes the user's window and deletes everything they're been working on, it's a recipe for disaster if you make them look the same. To quote Matthew Thomas (mpt), at least when things look different, there is the hint that they will act different. 


In the context of the Humane Interface, Red Hat did something really stupid and made things worse with BlueCurve. 

A major point of the Humane Interface is the habituation a user naturally develops when they an interface (and taking advantage of those habits to make things more usable). With two different desktop environments like GNOME and KDE with two different behaviors, a user can't develop advantageous habituation to the UI because both environments act different. The lack of advantageous habitutation is a Bad Thing, and most linux distros with the two toolkits have had this problem for years. They should have picked just one environment and stuck with it. 

But what's even worse than not taking advantage of habituation is making habituation work against a user. If a user, let's say, uses GNOME app for 7 hours a day, they will naturally develop habits of doing things that they will try to carry over into other applications. What happens if after a year of using that application they try to apply the habits they have formed with the GNOME application to a KDE application thats look exactly the same? They will overzealously try to apply the habits they acquired in the first app to the second app because, after all, it looks so similar. BlueCurve fooled them into thinking both these apps were the same environment where the laws of physics were there same. But the habits they developed in application #2 really don't work the same as in application #1. In fact, those habits tend to destroy things on occasion. We now have bad habitutation that works against the user, whereas before we just had lack of good habituation that works for the user. And if the user doesn't know about the difference between KDE and GNOME (because they were so well hidden by Red Hat), they just assume all the apps in this "unified environment" are inconsistent with one another and will do weird or horrible things at any moment. 

In the end, I'm really past the point of flaming in regards to distributions like Red Hat making unusable software. They will keep doing the same stupid things again, and again, and again. I can't stop them from doing that crap in their own distribution, but if they think for one second that I am going to let them target the desktop and terrorize Aunt Tillie without a fight, they are very much mistaken. Flaming is redundant if you're close to all-out war. 

I'm writing a public license that enforces usability and, among other things, bars distributions such as Red Hat who refuse to make usable software but who target the desktop anyways from distributing or modifying the code, or copying any successful or unique UI designs. 

I probably sound crazy, but for me, the freedom for end users to get stuff done with a minimum of fuss is the most valuable and sacred freedom that exists when it comes to software. I will do what I have to to protect this freedom. 

--Ilan

-- 
My choice after I quit film school was either to be a script writer for porno flicks or a linux UI designer. And to tell you the truth, there's hardly any difference. 



More information about the TriLUG mailing list