Let 'er load.
Posted June 9, 2004, Michal Migurski of Stamen Design.
Updated August 19, 2004.
Get updates on this project via newsfeeds
(rss,
atom)
from gateio.
Version 1.2 actionscript and Flash source, available for free download and modification:
stamen_news-1.2.tar.gz.
This work is licensed under a Creative Commons License.
This is the newest incarnation of a project that's
been flying close to the ground here at Stamen for almost six
months now. The stripes you see at top are a near-realtime
cumulative view of Google's "In The News" sidebar (if you don't
see blue and red stripes, you need
Flash), organized by date and popularity.
This data visualisation experiment is intended to give a
high-level view of who's making news at the moment, and who made
the news at specified times in the past. While I was working on
it, I saw the excellent
Newsmap, a project that gives an instantaneous overview of
Google's headlines, rather than the time-linked information I've
been collecting. Since launching it, I have also come across
James Harry's "Week In
Review", proof-positive that no programming experience is
needed to create beautiful data visualizations.
The size of each color chip refers to the relative
amount of coverage an item has received on Google News on a given day,
based on total appearances in the "In The News" sidebar sampled
every quarter-hour:
Note that these items are representative of what Google
thinks is a proper noun, so there are definite anomalies in
the information. For example, on the week that Sun and Microsoft
announced an end to their legal feud, "Sun Microsystems" made an
appearance, "Microsoft" did not.
The color refers to relative growth or decay on
that day - an item that has gained news-share since the previous
day is red, while one that has lost is blue (note: you can
change the color scheme by using the color switcher widget at
left). Color chips are arranged left-to-right, in the
order of that news item's first appearance in Google
News: the oldest items will be at left, and newer items are at
right. I make an effort to find redundancies - e.g. "George W.
Bush" and "George Bush" are considered the same, based on a
combination of text similarity and my own judgement.
Clicking on a color chip shows a small graph of
that news item's performance over the previous month. The
"Top Names", "Gainers" and "Losers" link lists can also
be used to see this information.
The "Back" and "Foward" links move you forward and
backward in time by days or weeks, back to about
February or so, when I started counting.
Raw data is available and free for use. More info and
interface improvements (suggestions welcome) are forthcoming.
Things I'm considering:
- Finesse the transition when moving forward or backward in time
- Provide more useful information in the little sparkline graphs (were you aware they can be clicked on?)
For now, enjoy!