Re: I think we should know each other.
From: Eric Hunting
To: Bryan Bishop
Date: 2008-04-29 10:30:24 am
If we are to realize an integration of society to informatics and then
to production -to create what writer Bruce Sterling has recently
dubbed 'an Internet of things'- then the key rests in how we digitize
the spectrum of modes/methods of process representation for our
collective knowledge of fabrication in some standardized and modular
way. This also relates to the subject of courseware development since
that field faces a similar issue of how to deal with a very great
diversity of knowledge with a uniform structure for its presentation,
communication, and the evaluation of knowledge retention. You could
say courseware represents this same problem going in the opposite
direction. (this is why, in TMP2, I feature an Open Courseware project
in the same section as the Open Source Everything project. They are
both important as means to cultivating a Post-Industrial culture and
their development will likely feature much cross-over between the two
pursuits) As you noted, defining a specific semantic structure for
this is a vast and challenging task. It's no wonder you assumed at
first it would be a task demanding AI to accomplish. It looks
virtually impossible on the face of it. I think, though, that in a
study of how we really do this representation today and have done it
in the past -which really is something few people have ever seriously
and systematically examined- rest the keys to this. I don't think we
can realize any ideal system. What's probably more important is that
any system devised allow for free evolvability because, just in the
act of pursuing this, we're going to learn some new things about how
our own civilization and culture work that we didn't realize before.
The forest we never noticed for the trees, so to speak. And that,
along with the advancing technology of our tools, is going to feedback
on the later design of such systems.
The graphic representation of fabrication -usually craft fabrication-
is something that has always interested me. It has an ontology that
relates to some very unexpected things. For instance, to understand
the history of DIY literature today you also have to know something
about the history and culture of comic books because they both evolved
in parallel with the same systems of graphic representation -
particularly temporal representation- largely because there was often
a cross-over in the artistic talent pools they relied on for
illustration. One of my all-time favorite books is a fine example of
this; Peter Auschwenden and John Muir's How To Keep Your Volkswagen
Alive. It's like an underground comic book of the 1970s crossed with
an engineering textbook and this style became a hallmark of 'softech'
and sustainable building literature throughout the 70s and 80s. (I've
often said I which I could get an artist like Auscwenden to illustrate
for TMP) We also see the influence of -or perhaps the origins of-
Taylorism in this through the similar modes of compartmentalization of
tasks over time. Taylorism has often been accused of being a
collection of psuedo-scientific baloney feeding-into the sociopathic
tendencies of the executive class but it has its basis in the idea of
the quantification of the physical parameters of tasks and the
modularization of the processes of production, thus being the starting
point for the modern science of ergonomics and the basis of much
automation research.
Another unexpected side to this is the evolution of toys, model kits,
and product instruction manuals. In fact, in these this context comes
full-circle with the design style of Lego and Japanese model kits now
representing the state-of-the-art in the graphic representation of
fabrication and now being mimicked in contemporary DiY literature,
engineering and scientific visualization, and industrial and
architectural design presentation. Indeed, the efficiency of these
forms of graphic representation now allow for very different
strategies of production and distribution. For instance, the growth in
'flat pak' furniture design epitomized by Ikea is based on the desire
to off-load portions of product end-production to the consumer in
order to reduce production costs and allow for more energy and
materials efficient transportation of goods in flatter more compact
forms. This is only possible where the visual representation of this
final assembly is made comprehensible to the completely non-skilled
consumer and often irrespective of their native language. For
companies like Ikea, billions of dollars are now riding on this point.
That this actually works at all is pretty amazing, when you think
about it -even if a certain number of people often complain they never
can figure out these kinds of assembly instructions.
Curiously, the formal world of engineering and industry does not seem
have ever pursued as sophisticated a scheme for communicating its
knowledge as the craft world nor ever sought a standardization in
semantics as in the world of science. There are standards in drafting
and patent documentation and, quite recently, standards in
documentation for engineering committees but in general this culture
has been more concerned with the restriction of the flow of technology
information as a means to maintain market share (and for engineers,
job security...) or cultivate monopolies and so the implementation of
technology has tended to depend heavily on the engineer as interpreter
of very disorganized information and the general spread of
technological knowledge has tended to be illicit, based largely on
deliberate information theft and the reverse-engineering of
competitors products. So we have a kind of technological Tower of
Babel today. This is probably why we ended up with the situation NASA
and the new space entrepreneurs now find themselves in with having to
reverse-engineer museum and junkyard hardware from the Apollo era
because, after 40 years of neglect, no adequate documentation exists
for much of the technology that once put men on the Moon. One would
think this impossible for technology representing one of the historic
peaks of human accomplishment, but there it is.
Another area of this that has long been a focus for me is the
evolution of modular building systems and architecture, because of my
personal need for and interest in alternative and non-toxic housing.
Modular component systems are vexing because they've always been a
more efficient strategy to make just about anything and yet, in the
construction trades in particular, most of those proposed or developed
have been dismal failures while in areas like electronics they've been
astoundingly successful. In fact, in the context of the computer this
was so successful that it resulted in an entirely new industrial
paradigm that had gone largely unrecognized today; the industrial
ecology.
Most modular hardware systems have often been developed by designers
and inventors with a poor grasp of the logistics of industrialization
and so frequently fail miserably when their developers give
insufficient focus on market cultivation through application
diversity, never bridging the gap in unit production volumes necessary
to justify the tooling for their production. There's a presumption
that simply because something seems more rational in design it should
automatically be embraced by the market. Reality doesn't work that
way. So these top-down modular systems have a hard time getting off
the ground. Modular component systems that evolve from the 'bottom
up', so to speak, by being driven by market forces -as was the case in
electronics- have tended to be much more successful even if less
efficient. How these emerge is a bit of a puzzle. In the field of
electronics it was associated with the development of an abstract
symbolic language employed as a very high level design tool. In other
words, the modularity of electronic components has its origins in the
symbolic engineering language of schematic diagrams with components
being developed to directly translate that language into hardware.
Electronics engineers originally had to make the components to match
their schematic designs from scratch and on-demand. By designing
circuits for schematic modularity they inadvertently created a certain
standardization and reusability of components parameters that afforded
manufacturers a potential market volume that would justify mass
component production. These parts were 'functionally modular' in terms
of their ability to equate to uniform performance parameters.
Standardization in physical form-factor followed suit as compelled by
the need for ready physical component integration, though has never
become quite absolute. In other words, every electronic component
represents an independent domain of possible form factors bounded by
the topological and electrical constraints of its mandatory component
interface standards. An IC may have an infinite number of internal
variations in structure but to be useful must have standardized
external sizes, shapes, and pin arrangements. There's a programming
analogy here -one that's becoming more literal as we move toward the
use of generic data processing hardware like field programmable gate
arrays running virtual circuits as the basis of full computers. (with
FPGA programming is the same as circuit design and code optimization
has a topological component to it) Circuit design is done on an
abstract level independent of specific physical hardware form even if
those symbolic elements equate to specific pieces of hardware. This is
like high level programming. After this the standard form factors of
the components come into play through circuit board design as modular
graphic elements employed in layout and simulations to model the
thermodynamics and EMF field behavior of specific hardware
configurations. This is very much like the interpreted or 'bytecode
based' programming language in that one is still dealing in a layer of
abstraction based on simulation software. (the architectures of
'emulator' systems and circuit design and simulation platforms like
VHDL did, after all, emerge from the architectures of interpretive
programming systems) And finally we get to the physical hardware
implementation which may be performed with automated manufacturing
systems and so we're down to the 'machine code' level of a programming
analogy. It seems that electronics may be the only technology at
present where we have this complete a spectrum of standardized
semantics at each of these levels of development, and this, again,
seems related to the fact that electronic is the only industry so far
to have actually implemented an industrial ecology where this sort of
information must be pretty freely and efficiently communicated. The
only industry that does work, more or less, in non-linear systems of
development and production. This probably also related to the
astoundingly rapid pace of evolution in this technology compared to
any other.
With the Open Source Everything project idea I was thinking not just
about cultivating and disseminating Open Source knowledge to make end
products but also about the evolution of standardized semantic and
visual systems for representing their 'recipes' as I tend to call them
because you need more than just 'plans' to explain how to make things.
You have to represent a fabrication process, not just an end-form. And
you need to standardize to make the information portable, modular, and
reconfigurable across networks, across languages, etc. Making that
work would take experimentation and evolution, hence another reason
for employing a group project. I've often been critical of the sorts
of silly things the Make and Instructables blog participants seem to
focus on. I keep thinking; "What's with these people and their toys?
Where's the open source car? The open source hand-tractor? The
refrigerator? Wind turbine? Compact gas turbogenerator? Water
purifier? Things that would really shake-up this world if they were
open source." But I've also come to realize there's more going on with
these blogs than just people sharing instructions for making toys. The
real 'craft' of these blogs isn't in the stuff people are making. It's
in the language they are cultivating to communicate the instructions
for making things. You could be able to make Faberge eggs but your
social status on these blogs is based not on how impressive your
personal end-results but on your ability to communicate techniques in
a way that is effective and entertaining. So it's really the evolution
of another kind of literature going on here -a sophisticated mixed-
media literature for fabrication knowledge. Right now it's a pretty
messy, fast, and loose system but I'm noticing some consistent
structure emerging as people interact with each other's recipes and
refine them. This is the latest branch of evolution in DIY literature
and it is adapting old schemes from that to new media with some
interesting results.
Concerning T-slot; T-slot framing is a system of modular building that
is common in current industrial automation and laboratories. it's the
current 'state of the art' (more or less) in the evolution of so-
called universal building systems. It's had a very powerful impact on
the evolution of automation because it is doing -albeit slowly- for
machine tool systems what the modularization of electronics did for
the computer. T-slot is based on a system of aluminum profiles in
standardized dimensions (standardized in cross-section profile
dimensions and fittings but not in unit frame piece length) based on a
square section with a T shaped slot in each face. Simple fittings with
recessed or hidden bolts mounting in these slots allow the profiles to
be connected together in many way, usually forming simple box frame
structures but sometimes in trusses and space frames. Tubular channels
in the profiles also function as a means of distributing pneumatic or
hydraulic power or as conduits for wiring, allowing the framing to
function as a system bus with the appropriate fittings. T-slot is made
by many companies to common standards and a vast assortment of passive
and active components have been developed to integrate with it,
allowing for the increasingly easy on-demand construction of complex
automated systems and lab equipment. It's very commonly used for robot
prototypes, particularly when they have to be fairly tough, large, and
real-world functional. The three biggest manufacturers of it in the
world are MK and Bosch HQed in Germany and the 80:20 company in the
US. (which, curiously, took its name from the Pareto Principle) T-slot
appeared on the market sometime in the late 1970s or early 80s as a
building system for custom machines and industrial structures and
became a replacement for commonly used framing systems based on pre-
drilled or spot-welded angle-iron and flats, extruded or roll-formed
alloy channel, and 'trilap' joined wood in the Box Beam or Matrix
systems. It's sometimes called the engineer's Tinker Toy or Lego. NASA
was one of its early adopters, which helped give it prestige among
engineers, aiding its rapid proliferation. Intended originally for
industrial automation, today it's ubiquitous in that market and most
every machine-manufactured product in existence today has probably
passed through some piece of machinery with a T-slot structure. It has
greatly reduced the cost of developing automated manufacturing systems
and adapting them to rapidly changing product designs, making
automation more economical than it has ever been. It's produced an
automation integration and consulting industry very much the the
computer consulting and Value Added Reseller industry. However, it's
applications have steadily been spinning off in all directions and you
can now find it used for everything from reconfigurable power tools
for model-makers to office furniture and most recently as the basis of
a plug-in architecture system for prefab housing -which is a use I'm
personally very interested in.
Currently, I have been slowly researching material for a T-slot Source
Book akin to the legendary Suntools Box Beam Source Book of the 70s or
Ken Isaac's How To Make Living Structures of the 60s and have been
looking for novel 'recipes' for T-slot based structures as well as
other similar contemporary building systems, like the rod and socket
framing I mentioned and others like the N55 space frame developed by
the N55 design group in the Netherlands. (http://www.n55.dk/) I'm
currently hoping to build a recumbent bike, Dobsonian telescope, and
fabber from T-slot and hope for my own future non-toxic home to be T-
slot based. I've been hoping my relationship with the Jeriko House
company -one of the few current developers of T-slot based housing-
may lead to obtaining a workshop for this research along with that home.
Eric Hunting
erichunting@gmail.com