[prev in list] [next in list] [prev in thread] [next in thread] 

List:       kde-usability
Subject:    Re: Few miscellaneous suggestions
From:       James Richard Tyrer <tyrerj () acm ! org>
Date:       2004-07-24 19:53:48
Message-ID: 4102BE4C.9010904 () acm ! org
[Download RAW message or body]

Brad Hards wrote:
> On Sat, 24 Jul 2004 16:36 pm, James Richard Tyrer wrote:
> 
>>Brad Hards wrote:
>>
>>>On Fri, 23 Jul 2004 07:51 am, James Richard Tyrer wrote:
>>>
>>>>In this case, although the reasons have validity, the action is wrong.
>>>>Keramik is 'really cool' but it is not very usable.  This is a problem
>>>>that needs to be fixed.  Actually, this issue should have been considered
>>>>more carefully before it was made the default.  Now, it needs to be fixed
>>>>ASAP.
>>>
>>>Do you have any objective evidence that one window style is more or less
>>>useable that any other window style (or theme, whatever)?
>>
>>Yes.
>>
>>My assertions are based on on the viability of icons, widgets, and window
>>decorations.  And, I base my conclusions on classic information theory. 
>>Stuff with less information, or with more noise, isn't as easy to see. 
>>When you compare (the window decorations and to a lesser extent the widget
>>style) Plastik to Keramik, Plastik not only have more information, but
>>Keramik has more noise. It is simple: S/N ratio.
> 
> I disagree. Your use of information theory argument is both misleading 
> (because a UI isn't a single channel, so it doesn't hold) and subjective 
> (because your references to "signal" and "noise" are unsubstantiated by 
> objective test results).
> 
> If you don't have controlled test results, you are just expressing an opinion. 

Actually, the S/N comment only applies to the Keramik window decoration.  It is 
just another more technical way of saying that the "busy" window decorations are 
not a good thing when usability is concerned.  It should be obvious that the 
icons on the buttons and the text is the signal and everything else is noise. 
Also, note that it is actually LOG(S)/LOG(N).  And, the FFT method below would 
also be applicable.

However, information theory does not just apply to only one dimensional signals. 
  Multi-dimensional signals can also be analyzed using standard methods.  In 
this case, for an example, an icon can be considered to be a two dimensional 
digital signal with each pixel having three orthogonal qualities.  Standard two 
dimensional DSP (Digital Signal Processing) methods can be used to analyze an 
icon for information content.

A simple metric would be to decompose it into three matrices one for each color, 
take the two dimensional FFT of each one.  You can obtain the digital equivalent 
to a volume under each FFT by a scalar addition of all of the entries in each 
FFT result matrix.  So that icons of different sizes can be compared, you 
actually want the average so you should normalize the sum by dividing it by the 
number of scalers summed (the area of the icon in square pixels).  Then take the 
square root of the sum of the squares of these three (normalized) scalar sums 
and you will have a very rough metric of the information contained in an icon. 
Note that for a first example, it should be obvious that a gray square would 
have a value of 0 indicating that it had no information.

--
JRT
_______________________________________________
kde-usability mailing list
kde-usability@kde.org
https://mail.kde.org/mailman/listinfo/kde-usability
[prev in list] [next in list] [prev in thread] [next in thread]