12 December 2005

The craft of view-scaling

my recent neologisms
like gWarming
and vshort
and aCock (for Alexander Cockburn)

are admittedly-borderline experiments

in a broader campaign
that might be called 'view scaling'




an illustration:

an online teevee schedule
is a 'view' of a database of shows
with fields for plot-summary
actors, year-made, etc

and in the 'now showing' over-view
you get very-abbreviated entries
for each of the many channels

but you can click on any individual show
to see the full database entry




so an abbreviated movie blurb that appears baffling
becomes clear when you see
in the expanded view
that the first-named actor is jackie chan

that it's a jackie chan movie
is the critical bit of info

but the database drudges
had to put that critical bit in the
normally-secondary
'actors' field

so the abbreviated listing
which omits the actors
favoring instead the plot-blurbs
left out the critical bit




sometimes it's the director that's critical
a bergman film
a woody allen film

or the school
a dogma95 film
or french new wave




and this stubbornly human
database-intractability

is universal

we can never say in advance
that one database field will always be
the most important to feature
when only one field can be displayed




and i'll call this the 'view scaling' problem

when you need to cram
five pounds of data-sugar
into a four-pound screen-canvas sack

something's gotta give
and you want the least-important pound
to give first




but often you want this trimming
to happen under robot-control

so databases need a custom-set importance-flag
on each field
to optimize the trimming




and further
as views dwindle down to a precious few characters

you need ways to trim the phat
from individual words

so 'very xxx' becomes vxxx
and jennifer lopez becomes jLo