THEORY: Where hypertext design should go

[Jorn Barger]

Summary: long rich timelines with links and quotes, that reward
many repeat visits, compiled as you research the topic

I want to draw attention to some non-obvious features of my
latest mega-timeline (more than 200Kb, 1000 links):
http://www.robotwisdom.com/ai/timeline.html (intro)
http://www.robotwisdom.com/ai/timeline1.html (to 1937)
http://www.robotwisdom.com/ai/timeline2.html (1938-1969)
http://www.robotwisdom.com/ai/timeline3.html (1970 to present)

First off, the unifying theme of the timeline isn't even that
important-- if it looks to you like a grabbbag of unrelated
items, that's fine, because most of the items were chosen for
their intrinsic interest, so everyone should be able to find
something rewarding.

Back in 1997 I compared hypertext links to the colored shards in
a stained glass window:
http://groups.google.com/groups?selm=5umnil%248en%241%40Jupiter.Mcs.Net
...arguing that the best design would use link-shards that clearly
transmit/summarise the pages they link to, and arrange them in an
attractive unified pattern.

Keeping a weblog teaches one to choose 'bright shards' (ie, to
write vivid link-text, often including pullquotes), but webloggers
are constrained to a most-recent-first arrangement of those shards.
An unhappy consequence of this is that weblog archives grow dusty,
unread.

A simple alternative with better staying power is the timeline.
Almost any informational webpage can be associated with a date on
verious topical timelines. Filling out the timeline with related
sites in chronological order can often _add value_ by revealing
the longterm evolution of the topic.

(Any newspaper with online archives should offer timeline-based
article indexes. Especially for science-related stories about
archeology or evolution, etc.)

The Web already has zillions of timelines, but most of them are
unreadably boring, mechanically repeating the conventional view
of history... and almost none of them leverage the format by
adding offsite links for each timeline-entry.

So I'm suggesting:

1) timelines with at least one offsite link per entry, on average

2) emphasize inherently interesting items rather than rote boredom

3) use pullquotes to draw people in

4) don't be afraid of long pages

Finally, this doesn't need to be a lot of work if you get in the
habit of compiling the timeline _while_ you research a given topic
online. It provides a handy way to keep track of useful bookmarks,
and there's a synergy that sets in because so many unexpected
connections will turn up in the process of your research.