Re: Another Perspective on the Cause of the Keyhole Problem [LARGE]

At 11:28 AM 10/20/2003, Dave Thomas wrote:
>this requires a degree of arrogance on the designer's part. Choosing 
>certain column widths may remove some flexibility from the end-user, 
>but may also end up increasing the overall effectiveness of the user's 
>experience.

Well, I suppose that it *may*, but then again, I also suppose that it may
not.  I continue to believe that the page viewer is in a far better
position to determine the details of page layout such that his/her overall
experience is optimized.

>On this site, the designer clearly felt that the headline was 
>important, and used it to set a minimum usable width for the main text: 
>resize the window below this width, and parts of the text get lost. The 
>designer also clearly wanted to establish separate widths for the 
>sidebar and for the main text: the sidebar has a fixed size, which make 
>it look good when viewed normally, but which causes the page to break 
>for those who use large fonts to correct for poor vision. (If this 
>doesn't make it through the list, the following is an image of the same 
>page with the fonts scaled up 4 or 5 times. The sidebar is messed up, 
>which the picture and fonts bleeding into the main body {at least on my 
>browser}).

FWIW, it looks fine with my browser, but the fact that it does not look
fine with yours is a legitimate criticism of the site.  It has nothing to
do with keyholes, but well designed web sites aimed at the general public
(as opposed to e.g., corporate web sites meant to be viewed only by
corporate-approved browsers) should be designed to look good on as many
different browsers as practical.  

>So, is any of this important, or a criticism of Scott's web site?

Yes and yes, IMO.  But it doesn't have anything to do with keyholes.

>No. His site is absolutely fine.

Scott disagrees.  The site could be improved in many ways, including
addressing the problems you've identified.  However, I think it does a
pretty good job of avoiding keyholes.  It does a less good job of avoiding
reverse keyholes (i.e., adapting well to very narrow or very short
windows), and its cross-platform behavior falls short on several counts.
There are other problems, too, but fortunately the topic here is 
keyholes :-)

>But in all this talk of absolutes, and of keyholes, it is easy to forget
>the underlying truth that communication is a holistic activity, and that
>effective communications must maximize many dimensions at once.

This "many dimensions" issue is very important.  In putting together any
system, there are many things to be considered (e.g., cost, portability,
performance, scalability, correctness, maintainability, usability, etc.).
It is typically not possible to optimize for everything at once.
Fundamentally, my goal with TKP is to get that issue a place at the table,
so it's considered along with all the other issues that have to be traded
off against one another.  Furthermore, my goal with *gratuitious* keyholes
is to convince people that, by default, they should be avoided or
eliminated unless there is a compelling reason to the contrary.  By
definition of being "gratuitous," one can never argue that getting rid of
them is too much work.  

Scott