Skip to Content.
Sympa Menu

pcdb - [pcdb] complexity, patterns and permahub

pcdb@lists.ibiblio.org

Subject: Permaculture Database

List archive

Chronological Thread  
  • From: christophe mckeon gonzalez de leon <chromatophore@gmail.com>
  • To: pcdb@lists.ibiblio.org
  • Subject: [pcdb] complexity, patterns and permahub
  • Date: Sun, 10 May 2009 18:57:10 -0700

hi,

i've read through some but not all of the past threads concerning the
website and database on this and the permaculture list. i'll be trying
to absorb some more soon. so if i'm covering ground that's already
been covered, apologies. one thing that i've noticed is that the
central concern seems to be the database itself. as i am thinking more
along the lines of a practical global permaculture web portal with
many users as opposed to the DB at the core of such a system, i may be
somewhat diverging from the aims of this list, but i hope you'll bare
with me. considering those aims, i think that starting the design from
the database perspective may not the right way to go.

i have a feeling that starting from the user/interaction perspective
with an eye towards mining emergent user generated structure for data
as people populate a more loosely defined system might be more
fruitful. considering the massive complexity of the interactions which
a very detailed upfront data model would be trying to capture, i'm not
sure that attempting it is worth the trouble, and doing so may also be
prone to initial design error. after all we are talking about ecology.

we also have to think about the user base and how people are going to
be comfortable both accessing and supplying data. how are permies now
communicating? mostly in the flesh, or in forums and mailing lists.
the wikis do not seem to have really taken off. so, i don't think too
many people are going to be jumping up and down for data entry jobs,
much less unpaid ones, especially when they would rather be gardening.
what that boils down to is, how do we make data entry fun and easy? a
straight non-structured wiki might be just a bit too loose for us to
be able to do the kind of data-mining i'm thinking of.

one way to go might be pattern languages. pattern languages can, i
think, capture a great deal of what permaculture talks about at the
technical level (maybe not the ethical one). from what i've read so
far in the permie literature pattern languages are mentioned but in
several different ways and not always by name. guilds are basically
patterns but are generally formally not thought of as such. principles
like 'preferring edges' are patterns. then sometimes 'the patterns of
nature' are mentioned but not in the context of pattern languages. if
we provide a generalized entity we call a pattern into which we can
pop a lot of what permaculture has to offer, we'll have a powerful
unifying principle for the site/DB design.

when inventing pattern languages christopher alexander was trying to
find a simple formal system which anybody could learn in minutes, but
which could deal with a massive number of design constraints while
remaining tractable and fun/easy to use. initially he had tried to
work with computers and more mathematical models for the design of
things like teapots with far *less* complexity than the ones we are
dealing with but could not make headway. he finally realized that the
appropriate technology was just human language with a bit of extra
structure thrown in. maybe he would not be able to capture every
little detail but so what, trial and error, careful thought,
communication and the simple use of well chosen sets of interacting
patterns was the way forward.

i think that if we make the pattern the central data entity, then a
lot of complexity vanishes for both us the developers, and the user as
well. the downside of course is that the data model would be coarser
than an attempt to, say, try to directly capture interactions between
species in the design of the DB itself. so for instance a guild
pattern might have a place for photos, drawings and text, a list of
species involved, as well as fields for soil type, shade, pH, but all
in all only a small handful of these to keep things simple. this will
keep everybody happy (including us the developers), and people will be
more likely to both contribute and consume information. note that the
photos drawings and text remain fairly central, which would be inline
with how permies are already communicating & also inline with the
spirit of patterns.

another way to frame the problem is to say that by having more people
contributing simpler information we will gain more than by having
fewer people laboring over more complex data entry, because the
inherent fuzziness of the problem domain lends itself well to
statistical analysis. or from a practical angle, it might be better to
be able to see broadly that, say, all of the west coast above portland
is failing to grow foo alongside bar, than to try to model in detail
why that is the case. details of course can be looked for in the
text/photos, again natural language trumps data modelling.

so for instance, the questions that i see have been of central
importance in discussions on this list, like how species or even
minerals interact & how that should be modeled? these could be handled
albeit more loosely & statistically, by guilds (just another pattern),
just let user comments/geocoding/ratings/tags along with
photos/diagrams/etc be the guiding hand in adoption/search/analysis
rather than a more fine tuned but potentially more brittle data model.

because the whole site would i hope be restful (see below), developers
who for example want to look at species data correlated against
success rate, soil type, & location, for use in some other more
complicated DB or app, will be able to do so easily. there should
also be some mechanism for allowing variations within a pattern of
materials/species/etc so that there is not a proliferation of similar
patterns.

as far as user interaction goes. it should be easy to upload media,
maybe even easy to sketch directly in the browser using flash or the
canvas tag or something to illustrate or comment on a pattern. it
should be easy to see the relationship between groups of related
patterns. a large shady tree might show up in a building pattern and
also in a forest garden pattern, if the user can see that they are
related they are more likely to include them both in a design in a
meaningful holistic way.

i think that central to the design of the website should be the notion
that it is the interaction of patterns and the relationships between
them which give them their generative power. relationships between
patterns could be user specified, for example using alexander's idea
of sub-patterns and super-patterns (maybe also sibling patterns), but
also relationships could be pointed out to the user by the system, for
instance when two guilds share species.

some other useful data entities/webpage types that i think should be
included are:

people, locations (geocoded & with google map integration),
institutions, courses, physical materials, species, physical
permaculture sites, other websites (basically links but with some
extra data about the sites), physical resources like piles of wood
somebody threw out (w/ reference to a location), articles (wiki
entries) i have started to model these relationships on paper, i can
share some of that if you guys are interested in what i have been
saying.

the species entries would just contain permaculturally pertinent
information, links to (or maybe even content pulled from) wikipedia
and other relevant sites/databases would be automatically generated so
as to keep redundancy to a minimum.

another idea is to have guidebooks people could collaboratively write
about specific subjects, like "an intro to permaculture". these books
could easily inline material from other articles or patterns, or link
to them, and could be printed as free pdf's as per everything else
(see below).

along similar kind of meta-page lines, another entity might be a
'design', comprising written word, media, as well as collections of
patterns and notes on how they are working together. permaculture site
entities could have an associated design page, people could comment,
link to their own designs etc. something to keep in mind here might be
a way for users to keep a diary (blog entries of some kind?) of how
the design is actually coming along. maybe that would be one of the
functions of the permaculture site pages.

i'm thinking the whole thing would be a cms type deal with some wiki
aspects, and editing constrained to registered users (who can *easily*
sign up). as well as a blog which registered users can post to and
which aggregates other permie blogs. i think all submitted material
should be implicitly under a permissive creative commons license which
users agree to when they sign up, and that there should be pdf
generation of pretty much the whole site or subsections thereof, for
use in places where computer access is limited for one reason or
another.

i'd like it if the entire DB would be accessible via a rest interface
(http://en.wikipedia.org/wiki/REST) to make it easy for other services
to use the information. i'd personally be interested in working on all
of this in ruby on rails, unless there is some pre-existing technology
which can handle it, but i am a bit suspicious of canned solutions.
the skin i'd like to have fully in css, the graphic designer should
not have to touch the markup.

i'd really like to hear all of your thoughts and i hope you are
interested in tightening up some of these still very sketchy ideas
with some further exchanges. there's still a lot to work out if we are
to proceed along these lines, especially in terms of user interaction.

best,
_c

p.s. it looks like i have a bit more reading to do:
http://www.livingneighborhoods.org/ht-0/generative.htm
time is something patterns do not capture well. i've been
reading holistic management, and time is important there
too, maybe some ideas to integrate?




Archive powered by MHonArc 2.6.24.

Top of Page