CHI-WEB Archives

ACM SIGCHI WWW Human Factors (Open Discussion)


Options: Use Forum View

Use Monospaced Font
Show Text Part by Default
Show All Mail Headers

Message: [<< First] [< Prev] [Next >] [Last >>]
Topic: [<< First] [< Prev] [Next >] [Last >>]
Author: [<< First] [< Prev] [Next >] [Last >>]

Print Reply
Jeff Brandenburg <[log in to unmask]>
Reply To:
Jeff Brandenburg <[log in to unmask]>
Wed, 3 Mar 1999 11:26:47 -0500
text/plain (98 lines)
Steve Bang wrote:
> I've followed hypermedia since the mid-80s and some early research
> considered underline issues.  One of the strongest arguments made back then
> against underlining is that the underlines available on most systems cut
> through the descenders (in letters like g, j, and y) and affected the
> readability of the text.  One alternative proposed back then was to use a
> gray bar below the hyperlinks.

This argument was founded on technological limitations that no longer apply,
I think.  It's easy to draw a continuous underline; you can even do it in
video (character-generation) hardware.  Drawing an underline that breaks
around descenders pretty much requires bit-level manipulation, which is
quite expensive by early-1980s standards.  Now, of course, everything
Navigator or IE displays is bitmapped, and doing the right thing with
underlines is almost automatic.  The one remaining problem is underlined
words (like variable names) that include underscores.

> Professional desktop publishers typically avoid the use of underlines
> because there are better (i.e., more aesthetically pleasing) alternatives
> available (bold, italic, and colored text) as well as the above mentioned
> issue that the underline cuts through descenders.  If an underline is needed
> (not very often), it should be created using graphical lines inserted below
> the descenders.

Professional desktop publishing is massively different from Web design.
Leave aside for the moment arguments about presentation independence and
user-centered appearance control.  Web pages are almost always displayed
on a device whose resolution is in the range of 70 to 150 dpi; professional
publications *start* around 1200 dpi.  Techniques that are aesthetically
effective in print are not necessarily useful in screen displays -- for
example, it's nearly impossible to display legible italics with a 12 pt
or 10 pt font.

In other words, the alternatives that desktop publishers (properly) prefer
to underlining are not often practical for screen displays.

> Another argument in favor of not using underlines for hyperlinks is that it
> helps resolve some of the abiguity between the use of graphics for text or
> picture links (which don't use underlines to indicate hyperlinks) and the
> hyperlinks created in HTML text.  This is one reason I prefer to turn off
> underlining.  [Note: MSIE users CAN explicitly force links to ALWAYS use
> either the "hover" style or underlines to indicate links (this feature can
> be found under View|Internet Options...|Advanced).  The argument can be made
> that web designers are more concerned about novices who may never venture
> into advanced options.]

The argument can also be made that this ambiguity reflects a problem with
picture links, not underlines. :-)  In The Olden Days, image links were
surrounded by a border, blue (for unvisited links) or purple (for visited
links).  Such cues were deemed offensive to design, and they're becoming
rarer now.  I miss them.

> I do not feel obligated to use underlines just because that was the only way
> available to indicate hyperlinks from the beginning.  Would I as a designer
> have chosen to move from colors (or hover) to underlining, if underlining
> had been introduced later other alternatives?  If the earliest use of
> hyperlinks in graphial Web browsers had used only colored text to indicate
> links, I don't think we'd be having this debate today.  If anything we'd be
> arguing over the use of colors other than a default link color.
> Keep in mind that we are in a transition stage still.  Hover will become a
> standard CSS style on both browsers soon.  I think there is sufficient
> visual clues available to design links the way that you want as long as you
> are consistent.  The use of the hover style, colored text, and the pointer
> hand (cursor) should be sufficient to indicate the existence of hyperlinks.

I have to disagree strongly.  Text colors alone might be effective if
links were the *only* colored text in otherwise monochromatic designs;
they never are.  If every page indicates links via text color, and every
page picks its own colors based on its own design scheme, it's almost
as bad as having no cues at all, IMHO.  I struggle enough with sites
that pick their own visited/unvisited link colors, even when their links
are still underlined.  When I don't have the underlining cue, or when
lots of non-link text is underlined, I'm likely to leave in disgust.

As for hover:  when I was in school, students were *discouraged* from
using their fingers while they read! :-)  Surely a design requiring
users to scan over the screen with a cursor can't be as usable as a
design that lets users discern links visually at a glance.

I do have to use some sites that are designed with this perspective.
I still occasionally find links that I never suspected; there may well
be others that I still haven't found.  This process of surprise and
discovery has its place in art and entertainment, but it's not what
I'm looking for in a business or reference site.  If you're trying
to sell something, or trying to provide information, it seems wise
to mark the user's paths as unambiguously as you can, even if the
result isn't precisely as beautiful as you'd like.

> Just my thoughts,

...and mine, of course. :-)

> Steve
        -jeffB (Jeff Brandenburg, Persimmon IT)