All our technology on Earth has been built from scratch essentially using human power, but surprisingly, majority us (individuals) do not know how to make many useful pieces of technology from scratch. This knowledge gap is what is a problem for individuals who are "obsessed about doing what they want" rather than doing what others want them to do.
The idea is to apply the principles of generating driving directions from a road map to automate generation of technological procedures from a map-like knowledge base.
The knowledge-base would ba a directed graph, the nodes (or vertices) of which would be a set of procedures and definitions (one set made up of a union of two sets - set of procedures and set of definitions). Definitions would represent the materials or products. The edges (or vectors) of the directed graph would represent inputs or outcomes (depending on the direction of vector) of each procedure.
Each procedure would be a list of actions necessary to produce a definition.
Each action would be described by one sentence.
I am talking about a Google Maps as an analogy. Assuming that Google Maps Driving Directions provide enough information for a driver to get from point A to B, one may think that the instructions generated by Google Maps are written in language that a driver will understand. The driver needs certain experience (such as experience of turning right or left) to successfully drive car according to the instructions. I am aware that by analogy, the technological instructions will be impossible to generate without abstracting the actions, and using the words that only the people with experience in actually doing these actions will be familiar with, and assuming that they will.
// There are a great number of various commonly known technological procedures that take something in, and give something out. Why not to connect them all into a map? //-- Inyuki, Jan 18 2010 An earlier version http://quod.lib.umich.edu/d/did/Not the same thing, but a similar impulse. [mouseposture, Jan 18 2010] Evolutionary Education Evolutionary_20Education [phoenix, Jan 19 2010] Protocol Online http://www.protocol-online.org/"Your lab's reference book" [Inyuki, Jan 20 2010] The Toaster Project http://www.thetoasterproject.org/Making a toaster from scratch. [tatterdemalion, Jan 20 2010] why ™, why not a kicky ® or even a ©?-- WcW, Jan 18 2010 ISO9000 ? TQM ?
This will not, however produce that most precious of commodities, innovation - merely repetition.-- 8th of 7, Jan 18 2010 1) As an academic project, this seems quite interesting. Applying graph theory to real-world networks is all the rage, I believe. I don't buy it as a practical reference work, though. 2) A lot of the sort of information you're proposing to collect is owned by people and organizations who attach a monetary value to it.-- mouseposture, Jan 18 2010 [+] interesting-- FlyingToaster, Jan 18 2010 [WcW], it acronymizes "Technology Maps" well.-- Inyuki, Jan 18 2010 And all I was noticing was the tired cliche, daft on me.-- WcW, Jan 19 2010 What? I'm missing something, could you give a simple example? Is this like James Burke's "Connections"?
This reminds me of an early (silly) attempt of mine to write down all possible chess moves. I was 8 or something and soon realized my hand would fall off before I finished. Where are your starting points? Caveman? Modern man? Man with a machine shop? Man with an internet connection and a credit card? Google can give driving directions because it assumes you know how to move in a basic sense. With technology the "movement" is the interesting part and there are several ways to move from A to B, so which do you pick?
For example, I would like a red shirt, how do I make one? Grow cotton, pick cotton, make spinning wheel, spin thread, create waver, weave thread to fabric, create sewing machine, sew fabric into a shirt, find mineral for dye....etc.-- MisterQED, Jan 20 2010 Yes, you are right, [MisterQED]. Generally there are too many possible ways how you could get a red shirt, that it is impossible to list them all.
However, under narrower circumstances, such as a molecular biology laboratory for instance, there are a wealth of pretty common procedures to accomplish common tasks, and these procedures are written into protocols (see link "protocol online"), and these protocols are used to get something from something.
Certainly, there is a combination of chemical/physical reactions applied to certain materials to produce a red dye, and to dye a thread for manufacturing a red shirt, but usually they are not connected into a one procedure, but exist as separate techniques that take something in, and give something out: take one materials produce others.
People who plan to manufacture a product take these procedures and connect them into a sequence that produces a desirable result. There are a great number of various commonly known procedures that take something in, and give something out. Why not to connect them all into a map?-- Inyuki, Jan 20 2010 “If you want to make an apple pie from scratch, you must first create the universe.” Dr. Carl Sagan-- tatterdemalion, Jan 20 2010 "first, catch your hare" Isabella Beeton-- pocmloc, Jan 20 2010 One tricky piece would be trying to represent a multi-dimensional cluster of near and far relationships on a 2d surface without it ending up looking like a mess. With so many interconnecting concepts, connecting at different levels, you're going to have to filter out all the noise, based on some kind of relationship strength metric, or only show relationships of certain types - like showing roads, or rail, or underground connections.-- zen_tom, Jan 22 2010 So really this is more suited, not the the red shirt, but to just the red dye. You need a chemical, to create it you need another set of chemicals. You don't have one of them either, so to make that you need a different set of chemicals, all in the hope at some point you will have all the chemicals you need to produce the dye. OK, that is interesting (+).-- MisterQED, Jan 22 2010 The problem is that a car can go everywhere on the road, making connection algorithms easy. I don't have any idea what that means, but it seems relevant.
Basically, your vectors will be in hyperdimensional space and cars are not, making mapquest possible but technology maps vastly more difficult. Anyways, I think I read about a computer that does it's own genetics research using an idea like yours and it even found some new stuff. Soo.. it's undefined but realistic. Magical and kinda boring. +-- daseva, Jan 22 2010 [MisterQED], yes, the map would be made out of the basic well-tested procedures that take something in, give something out, and thus can be connected into a directed graph.
Making a red shirt isn't a basic procedure, but is a sequence of basic procedures: taking chemicals -> getting chemicals -> ... getting dye that can be an input argument for the sufficiently general procedure of dying a cotton thread. Then hopefully there is a procedure that takes this type ofcotton thread as an argument and outputs a cloth of type that is suitable for making shirts... If there aren't procedures whose types match, then there is no connection between them in the graph, and the procedure of making a red shirt can't be generated from the map.
Actually, robust multivalued functions in a procedural programming language that have ever been written naturally define a directed graph where inputs and outputs can be matched up if their types match. This would imply that precisely defining the types for real world variables such as materials, would be a first step in creating such a map.-- Inyuki, Jan 22 2010 random, halfbakery