Looking at the different Maker blogs, I could see that there was an emergent set of information standards forming ad hoc by a popularity-driven process of selection as well as limitations -chiefly in graphics as people generally can’t produce their own drawn illustrations so resort to photos and video. The people who first started publishing recipes for making things on sites like the Make blog and Instructibles had no preconceived notion of the format of presentation for those recipes. So what they have been creating to date is a mashup of metaphors and examples they remember from DIY/hobby books, TV shows like This Old House, cooking shows, and so on. And yet there’s an emergent consistency in these as people react to each presentation in terms of open popularity voting and critique and adapt their future presentations to that. But this process has no goals and people aren’t really put as much attention into the critical task of showcasing technique -disseminating knowledge- as they are in demoing their own designs. So I envisioned this community that made producing this kind of Maker knowledge and collecting it in a database -while making a living on its publishing in print and in the creation of kits- their vocation. The knowledge would all be open source, but publishing and kit-bashing would be for-profit, letting people make a living and providing a means to subsidize the free dissemination of the knowledge on-line and through outreach activity.
Eric Hunting meditates and strategizes on the way forward from here to there.
“I’ve been thinking more on this lately -particularly as I’ve recently been working on an article on characterizing/defining Post-Industrial design. This is a topic that seems very prone to spiral into abstraction -especially when the goal is as grandiose as gathering all industrial knowledge into a central repository of some sort. Perhaps we can ground the discussion a little by defining how we might represent any artifact in terms of a body of information. For the moment we’ll resign ourselves to a characterization for human use and leave the machine encoding to another level as we need to grasp this in human terms before we can seriously figure out how to work our machines into the knowledge loop.
Any manufactured artifact is -more or less- fully characterized by the following collection of information;
-annotated design images, digital models, physical models
-samples -in the form of stored examples, photographs
-cultural/historical cross-reference (see science historian James Burke)
-standardization parameters (technical specifications that are standardized across a collection of artifacts)
-design discussion record (for designs that are group or community projects) -designer/design community (contacts, forums, references, etc.)
-drawn plans -CAD, drafted, etc.
-materials and components catalog
-materials source guide
-cross-reference to materials production
-tool catalog (including software tools)
-cross-reference to tool designs
-step-by-step production process instructions
-firmware (here defined as software embedded in or integral to the function of the artifact)
-optional kit assembly instructions
-user feedback record
-materials and components catalog
-technique cross-reference for artifacts that are tools
-user group community
-Software Family (cross-reference index to software artifacts made to compliment the artifact)
-recycling technique cross-reference
-upcycling technique cross-reference
-impact analysis record
-Spime (borrowing the term from Bruce Sterling, here defined as the collective media and information network associated with any product and supporting its feedback channels, knowledge cross-references, interdependencies and ‘prosumer’ communities)
For each set of the above information you have variations and iterations. A variation of a design is an adaptation or modification of a design to suit customization/personalization or some variation in application. An iteration of a design is an improvement of the design. So we might represent the ‘lineage’ of an artifact as a tree where each trunk is a line of iteration and each branch a variation branching off into a line of iteration from a point in the parent line of iteration. In some cases lines of iteration split into competitive parallel branches. At other times divergent variant branches can converge to become future iterations of other variant branches or even the main design branch.
Not also that some of this information flows in time independent of the iteration. For instance, the user feedback records of an iteration grow as long as that iteration remains current, and form part of the reference information used to make subsequent iterations and form part of its ‘spime’.
All subcomponents and all processed materials can be characterized as artifacts, interfacing into the hierarchy of other artifacts through the components/materials catalogs. Even things we think of as ‘materials’ like lumber, plywood sheeting, steel sheet, plastic stock, etc. all have ‘designs’ that determine their form factor and have often been standardized. The A4 sheet of paper is a specific design of sheet paper. In some cases components can be specified to be cannibalized to serve as a material. (this is upcycling) In some cases iterative evolution of processed materials and components can force an iterative evolution in the artifacts they are used in.
Raw materials can also be characterized ‘like’ other artifacts with the exception that they have no design, just samples/examples, a range of specifications defining quality, and a range of natural sources. Their ‘recipes’ are about extraction, not fabrication. This is not exclusive to just stuff in the ground. Some raw materials -like lumber- are ‘grown’ and so employ technique in their extraction and management of quality. Basically, you can say that lumber production is a ‘mining’ technique that extracts carbon from the ground and air and turns it into wood. In the old fashioned manner its like strip mining. Fell timber harvesting and tree farming are the more renewable approaches that have different technique.
Now, let’s consider the characterization of a technique. A technique represents a body of knowledge for how to use a spectrum of tools to produce a spectrum of products. An individual technique belongs to a family of technique embodying a roughly common spectrum of fabrication or performance. In some cases there is no product, but rather performance. For instance, instructions for how to dance produce a performance but not a product -unless you’re an anthropologist and consider dancing primarily a courting technique. The problem we have here is generalization. There are large families of techniques using large families of tools and a large spectrum of possible products. And so to illustrate a technique we must use simple artifacts as models and reference other artifacts as examples. The technique is adapted according to the specifics of recipes for different artifacts. Ultimately, all artifacts that use a technique in their production back-reference to the description of the technique as a potential example. But not all examples of a technique represent good models of a technique to learn it from -which brings us to the other fly in our ointment; skill. A skill is a learned technique arrived at through practice/performance. Skill is the ultimate goal in the communication of a technique -at least from the human perspective. We have only a rudimentary notion today of how to characterize a skill in the machine sense other than to say that it represents a mechanism for the translation/interpretation of a generalized fabrication process to the design specifics of a particular artifact. I’ve dubbed this ‘Taylorization’. (after Frederick Winslow Taylor) and characterize it as the intermediate representation of a production process relative to the topology of a production system prior to ‘compilation’ into the control language of individual machines. Thus I characterize ‘machine skill’ as a system/network of ‘Taylor programs’. But let’s not get too deep into that dark jungle right now…
For now, let’s speculate on how a human characterization of technique might break down.
-back-reference to family of technique
-description –cross-reference to science and engineering principles –cross-reference to integral technique –cultural cross-reference -tool catalog –cross-reference to tool artifacts -materials/components catalog –cross-reference materials characterizations
-illustration (images, models, performance/demonstration) –cross-reference to artifact examples -developer community –developer discourse record
-Model Process Recipes (many iterations)
-step by step demonstration –illustration/demonstration/media –cross-reference to artifact example recipes –cross-reference to integral techniques
-waste product characterization
-waste handling instructions
-cross-reference to recycle techniques and upcycle artifacts
Now, there’s an open question here as to whether one characterizes fabrication families of technique by their predominate tools or the predominate materials. Neither choice is particularly good because there’s so much overlap. The best approach might be to build upon that categorization described by Neil Gershenfeld in characterizing the tools of the fab lab; additive, subtractive, molding/forming, synthesis/stochastic (as in chemistry), reprographic (including photographic and lithographic), and algorithmic. (ie. software) This is still not ideal -there is still much cross-over between these families- but this may be a good starting point.
Though they do evolve with their predominate tools and materials spectrums, techniques generally do not iterate, though whatever knowledgebase we use to manage their information will, constantly, as both the technique and our ways of illustrating/demonstrating/performing it evolve. Techniques do, however, generate variations in extremely great numbers as tools evolve, new tools spawning new technique with them. Techniques can also be nested, a given technique a possible composite of multiple other techniques. Easy to see how characterizing all this for autonomous machines is rather daunting.
So here we have an admittedly rough but relatively complete break-down of the types of information that make up the definitions of artifacts and techniques. So what do we do with it? On the face of it, things don’t look too complicated but, the way things work today, at each line-item the specific information blooms into clouds of diverse data, representation metaphors, and media forms. In order to realize the goal of collectivizing our civilization’s industrial knowledge we need to standardize on these things in some way to produce a common knowledgebase format that is at once easy to put in a digital form that’s easy to disseminate and easy for people to use, adapt, and translate to different languages. Once we have some workable and nominally restrictive standards, we can start figuring out how this can be converted into a machine form -adding a machine-knowledge level to this database.
This is what I intended to be the primary task of the ToolBook cooperative. Looking at the different Maker blogs, I could see that there was an emergent set of information standards forming ad hoc by a popularity-driven process of selection as well as limitations -chiefly in graphics as people generally can’t produce their own drawn illustrations so resort to photos and video. The people who first started publishing recipes for making things on sites like the Make blog and Instructibles had no preconceived notion of the format of presentation for those recipes. So what they have been creating to date is a mashup of metaphors and examples they remember from DIY/hobby books, TV shows like This Old House, cooking shows, and so on. And yet there’s an emergent consistency in these as people react to each presentation in terms of open popularity voting and critique and adapt their future presentations to that. But this process has no goals and people aren’t really put as much attention into the critical task of showcasing technique -disseminating knowledge- as they are in demoing their own designs. So I envisioned this community that made producing this kind of Maker knowledge and collecting it in a database -while making a living on its publishing in print and in the creation of kits- their vocation. The knowledge would all be open source, but publishing and kit-bashing would be for-profit, letting people make a living and providing a means to subsidize the free dissemination of the knowledge on-line and through outreach activity. Using the sort of breakdown I’ve described here and guided by the objective of a ‘brand identity’ for their published media and its management and distribution on-line, this community would deliberately and actively cultivate evolving standards of knowledge representation and organization as they cultivate this knowledgebase. Which, of course, brings us to the Vajra Maker incubator community concept -a definitive place with shared resources and a shared culture focused on this task like they were building a new open Alexandrian library of technology on the Internet -where the barbarians can’t burn it down this time around. This is a really big task and I think this might be what it takes to actually pull it off. Maybe it takes a kind of casual monastery.”
Eric Hunting, via [email protected]