The Blacksmith and the Hacker

yarn-weaving-sweater (by dmuren)

What is the difference between a sweater and a pile of yarn? Nothing is added to the yarn to make a sweater; the yarn weighs the same before knitting as after. Yet the sweater is obviously different than the pile. The energy of knitting might seem to be the difference, but we could make make just as many random knits in the yarn -- expending the same energy -- and end up with a knotty pile of yarn. A knotty pile is definitely different than a regular pile, but it's also significantly different than a sweater. So, if it's not added matter, and it's not added energy, what's left? Answer: Information.

The difference between a sweater and a knotty pile is that in one case the knits are applied in an orderly pattern. A pattern is a repeatable method for applying information to matter. Using another sewing example, think about the pattern used to make a dress. It is information, printed on paper, but it can be applied to cloths of different colors or patterns to produce multiple finished dresses. These dresses are functionally identical, even though their colors or textile patterns may be different. The reason they function the same is because their patterns -- the information that organizes their matter -- are identical.

So, in garments at least, the matter and energy that go into an object are not so important as the information applied along with them. This should not come as too big a surprise: throwing eggs, flour, sugar, and water in a bowl to make a cake will never match the results of following a recipe. Methods, techniques, tricks, hacks, recipes, and other storage media for the information of making have been a part of being human since the first tool was made for the second time.

What should come as a surprise is that these three components are not given equal consideration by the manufacturing industry.

Energy is easy to quantify. It is the heat used to melt, treat, dry, cure, cut, burn, or bake objects. It can also be the mechanical energy that runs the machines that knit a sweater, cut a wafer, or stamp a fender. In many processes, adding more energy makes them more precise and accurate (cutting metal at higher speeds, for example) or results in a higher performance product (in making ceramics, or metal alloys, for example).

Matter also easily understood by industry. Matter is the raw iron and plastic, or the processed wire and sheet that feed a manufacturing process. Both the form and composition of the matter influence the final properties of the made object. Purity (in metals), impurities (in semiconductors), crystal structure (in everything), and nanostructure (in semiconductors) and metastructures (in fabrics) all affect the final product.

Information is a third component of every object, but it is introduced during the making, and it is not as easy to measure; Information does not weigh anything, or pass trough an electrical meter on the way into the product. Often, information is introduced in multiple ways, at multiple stages during production, and is not easily extractable in a final product. A drop forged wrench has at least four different informational domains encoded during making, and bearing on its ultimate utility.

forge_steel_grain (by dmuren)
The formulation of the steel used is the first domain. Metal crystal grain size, orientation, and deformation from forging are the second domain.

bulk_shape_surface_hardening (by dmuren)
Final machined shape and surface finish are the third domain. Finally surface composition or differential metallic composition -- as from case hardening, or plating -- is the final domain.

All of these domains are created by controlling the addition of matter or energy, and the removal of any one of them would seriously (and adversely) affect the performance of the final object. Additionally, the order in which these domains are encoded is important -- a metal cannot be alloyed after it is case, and a hardened workpiece is much more difficult to cut to a reliable shape.

So, in this example, the function of the tool depends, not so much on the matter itself, or the quantity of energy used in making it, but on how usefully and effectively that energy is used to encode information into the matter. And yet, we still think of this process as one primarily of gross processes (melting and forging) and energy (heat to melt metal, force to drive hammers). We think that a blacksmith is different than a hacker -- why is this the case?

Let us consider two blacksmiths to illustrate the point: One, a modern swordsmith,working with modern tools and technologies and the other, a swordsmith working in Hyderabad, India more than 2000 years ago. Both of them are making the same material -- what has become known as Damascus steel (more specifically Wootz steel). This metal has a highly ordered crystalline structure of alternating bands of very hard carbide (a ceramic) particles within a relatively softer (and more flexible) tempered martensite (an alloy of steel) matrix. While the final product is the same, the methods by which the two arrive at that product are very different.

The modern blacksmith subscribes to a method outlined in this patent by Daniel Miller. Basically, a metal bar with the appropriate blend of alloying elements present in it is alternately heated by burning natural gas and cooled by a cryogenic liquid in an inert atmosphere in order to grow grains of carbide within the bar of steel. After the carbide is grown, the steel is worked with hammers and grinding to form the final shape.

The blacksmith of antiquity also hammers and grinds his blade, but the method by which he obtains the carbide-enhanced steel is radically different. John Verhoeven and Alfred Pendray only recently reconstructed this method in the late 1990s. Because precise control of temperature, furnace atmosphere, or purchased steel composition was not available in pre-common-era India, the information in the final Wootz ingot was encoded through the selection of already information-laden admixtures. For example, in addition to iron and charcoal (for carbon), the alloying crucible was filed with leaves from specific trees, wood chips, and residues from river beds (salts). Each of these elements contributed some of their information to the mixture within the crucible as it was fired over wood for hours. When cooled, the mixture was now separated into one ingot of Wootz, and one thick layer of slag -- the matter of the additions, less its information.

Both smiths created comparable artifacts. From the viewpoint of industry, however, the modern method is superior for a number of reasons: It scales up easily, it requires only simple inputs, and it is highly predictable. The older method does not scale to mass production (crucibles alloying is a non-continuous, expensive process), requires complex ingredients (specific leaves, salts, etc), and is less obviously predictable (what scientific model is there for the interaction of molten metal and leaves).

However, looked at from the point of view of a programmer, these options appear in a different light. While the modern method is predictable, it is a brute force effort; Getting high quality natural gas, liquid nitrogen or helium (the cryo-liquid), and high purity initial alloy is a high energy, high precision effort, much like trying to design a better piece of software simply by adding more features, or more workers coding on it. In contrast, the ancient method is what a programmer might call an elegant hack; A highly ordered final product created using clever choices of raw components and a minimum quantity and quality of energy (a wood furnace for crying out loud).

Why should these viewpoints be so different? Because industry persists in thinking that matter and energy are the primary building blocks of things. Programmers have long realized the benefits that come from working in an information-heavy medium. The recognition of elegance as a desirable quality fundamentally changes the way that things are designed, and what things are possible. Humblefacture must embrace a three component view of the construction of things. Without the idea of information as a component, there can be no elegance in a design. Without the idea of elegance, there is little reason to work toward things like energy efficient manufacture, culturally appropriate manufacturing methods, or cradle-to-cradle design. Development of information-centered manufacturing processes is central to all the aims of these, and the best chance for maintaining our current level of technological output while transitioning to a humbler mode of making.

1 comment:

  1. Nice post but there is so many advance lab and medical equipments are developed and in trend like blood bag tube sealer, lab centrifuge, bod incubator muffle furnace by the best muffle furnace suppliers for the research and hospitals.