Components. Two rationalities
(this text was probably written in 1974, in the French version, translated circa 2006, then it deals with electronic components, not sofware)
The component, encounter of two rationalities
The component is the meeting point of two families of corporations and individual workers, numerous enough to justify standards creation, up to markings and representations, with metric frange, features sets, physical dimensions, voltage and current levels (typical, maximal, minimal).
This double origin, is expressed through two description types, functeional and organic, which never replace one another.
From the the user standpoint, a purely formal description is enough, in principle, and often, in practice : this is a 10 microfarad condenser, bearing 50 . that is a double NAND gate. But often, some organic precision will be more effective than a long functional description, if enven it is not a must. For instance, the mounting plan will specifu "cesium condenser" or "bobinated (...) condenser". Sometimes, at least, the nominal reference will be the most useful if not the one only possible : use the MM537 circuit of National. Conversely, the maker could, in its production department, be happy with purely organic descriptions, i.e. a description giving the basic elements and materials to be used and the operations to apply them.
But, here also, some functional features will be more efficient. And finally, even it that may look shocking, some features are neither organic nor functional. This is the case, in particular, of nouns and dimensions. This last category of descriptors is more and more present, in order to warrant and enhance the liberty of both partners. As such, it shows the growing autonomy of mediation or, if we want, the thickening of the structure between its two versants, organic and functional.
Indeed, users will tend to use the available components of unforeseen functions. As well as he can also use (and so give a functional meaning) to products heretofore useless for him and affording only a purely organic description. Bird defecations on Pacific Ocean islands (organic description) may become a very efficient nitride fertilizer, and be called guano (nominal description). Some components elaborated for Space conquest will be use by Industru. A Parisian clothes shop has, for long years, found its success out of a war surplus, the "toile d'avion). A purely functional description would prevent such new developments.
On the other side, manufacturers have similar interests. They, also, try to widen the market for their products. And, for instance, a software developped to answer the needs of a given corporation (functional description) will be "generalized" and be edited by a software vendor to all the corporations of a particular market, or to all corporations globally, bur for a specific function (e.g. production control). In this case, the functional description will go to more and more abstraction and generic functions.
In parallel, users as well as vendors, starting from a component defined at start organically (with precisions) to fulfill a given function, will try to fulfill this function using cheaper ways. The vendor will try to keep his price level, while strenghtening his brss margin, or to lower the prices to be more competitive. As for the users, they will try to satisfy their nees at lesser costs, or to do better with the same budget. And the organic description also will become more abstract.
In a sophisticated economic system, where markets connect large sets of vendors as well as customers, abstraction impacts deeply on production as well a use. The user knows that he can find sources for a given product, while the vendor way rely on a market to enhance his production facilitgies (let alone his R&D). Then, everything reduces into a formalizable description of products, totally expressed by performance and selling price. But, in practice, this objectivation and abstration is limited for technical and economico-political reasons.
Standards reinforce this trend. Its abstact mood is sometimes hated by men of practice. Nevertheless, it is a must. As well as descriptions, standardization has an organic side, a functional side and, in between, an structural and nominal edge.
Organic : an "appellation contrôlée" of Bordeaux wine, for instance, does not imply by itself that this wine has this peculiar taste, but that it comes from a specific vine, and has been prepared according to a specific method.
Functional : a device may be called by some names only after having passed tests proving that it corresponds to specific use levels.
Nominal and structural : ranges, scales, unt systems. From these descriptorfs standpoints, a component or product is seen in its neutrality relatively to its origin as well as its destination.
Here comes into play Esthetism and its perils. This temptation finds little ways of expression. Neither vendors nor users are the vast and elastic spaces customary in the liberal economic therory in its elementary presentations. We observe dominant positions, major and minor actors. They all new, more or less clearly, that standards options may or may not boost their firm. War is more present than artistic creativity.
Emerge also some false standards, the "de facdo standards" : the powerful ones impose their own law, the weak ones accept it, more or less complacently. Aesthetically unpleasant, this parasit/host situation is not necessarily a curse, and may be profit to everybody. Mais true standards, democratically designed and chosen are in the line of progress and coherence.
This objectivation of components concurs with more supple markets, simpler analyses of value and language, reduced stocks. Dangerous indeed, the closure of components upon themselves and the set of standards is good for all.
Actually, if we can say, at first glance, that markets generate standards, as we say that "functions generate organs", we must conversely admit that components generate markets. The two populations, makers and users, are not as "given" from stard as we could think. This mere distinction between vendor and customer is a consequence of markets, which imply a minimum of standards to allow any competition at all. At stard, one could say, corporations (or more exactly communities) are indifferenciated, autarcical and generating for themselves objects and organs to solve the problems at hand.
The difference arises when are recognized the repetitive features of some objects, the common needs of diverse individual or collecdtive users. Definition and identification is part of this differenciation process. And then the possibility of efficient manufacdturing by specialized corporations or group of firms.
Components, and still more standard components, reach a rather hig differenciation level. But differend differenciation types may coexist. Here, me must always respect a ryhtm on its way up. Early standardizing is not costly, as it operates on an still weakly defined border, pre-competive, a no man's land where straight lines may easily be drawn. Give a look to a political map of Africa. But this standardizing runs the risk of simplism and arbitrary choicesn, of ignoring the local terrain dynamics, of subsequent evolutions of technologies and men. Conversely, a late standard questions inured habits, solidified positions, let alone the heavy investments in mass production of standard products, as well as thought habits and behaviours.
And blood flows to draw borders throughout tilled fields. We have described this emergence of components under the industrial light, wih the production/use dialectics. Similar upgoings could be trace with organs, in a biological view. We shall not elaborate on it here.
Let us conclude with some notes on the growth of organs, taken in its generality.
At the low limit of simplicity, parts of a thing are not differenciated, be it on the organic or functional grounds. In other words, the thing can and may be cut following any plan without changing in nature (no organic differenciation) and all the parts of the thing fulfill the same functions, based, by globalisation, on the function of the thing itself.
When some differenciation emerges (or is perceived), we always try to justify
it, to explain it, i.e. to show
-either that this break in continuity is demanded by some finality (trivial case : line drawn on a paper sheet to represent à line)
- or that the different parts resulting of the differenciation relate to definite functions, and as such becope organs properly.
This recognition of correspondences, or this development in diversification is finally a fundamental progress in the knowedge of the object (or, in other cases, of the objecdt itself). Neither partition in organs, nor correspondence of physical parts to specific functions are ever complete or perfect.
The two moves are parallel. To fulfill better and better its function, organs are driven to inslate themselves, very neatly sometimes : he surrounds himself with a membrane (biologicial membrane, skin, car body, electro-magnetic protection). It develops its own homeostases, own rytms, own electrical feeding, own security redundances. It rounds up its shape in order to optimize its internal yield. It even specializes its parts in relation to the rest of the system : tendons (XXXXXXXXXX) concentrating muscular forces, blood vessels, electronic busses and interfaces. This autonomy may become so strond that, but for some environment considerations, the organ can be transplanted from one organisme to one another, or kept operating in isolation (Carrel chicken heart, spin-off corporation).
An so, being a generalization of one thing dominated by another one, of a component included in a system, or an organ in an organisme, we are allowed to go further and to find, upwards (as demandes Teilahrd), a general model of this up-marching : the fathership/filiation. A father dominates his son. A master his disciple. But the father becomes fully himself only in driving his son to adulthood, and soon after a father in his turn. A master is great only if he wants his disciple to become a thinker strong enouth to dispense himself of the masters authority, to become in his turn a master for other disciples.An so, I make growing all the mediations that I create. From mud on earth, creator image of my Creator, I make emerge a ball, a statue, a tool, a machine, an automaton, an intelligent automaton, and more and more so (at least by analogy). For the time being, the gaps is still enormous between the most advanced machine and the child I generate with my mate. But the convergnce, in spite of lapses and refusals, shows itself strong enough to foster my faith.
Sub-system warranting specifically adaptations of the system to a given type of aim differenciation (Le Moigne 1/33).
- The smaller a module, the more probably it will resemble other modules operating
for similar functions.
- A module must combine a number of operations big enough to be an autonomous center, economicalle viable (Blumenthal 1969).
The technology concept in HD1
A"technology" is a rather genral mode of implementing a family of
functions with a set of organs. Examples en HP : TTL, ECL, Mos for logical functions.
Mechanic gears Vs. electonic components for decimal computing.
There is a direct link between the concepts of technology and components : technology is a set of technical choices apt to be assembled in systems, using coherent parts, more or less standardized.
But, in HD :
- algorithm types, for sorting, pi computing, theorem proving, automatic system to build a domain
- a method is a tehnology, first to solve a problem through human minds, then by machines
- task division is a technology.
Darwin axiom : a good technology eliminates the less good ones. Indeed, it masters the fields more rapidly. If resources lack, are rare, a good technology knows better how to put them in use.
Never forget that the development of a technology has its investment costs.
Representation technologies :
- for instance different representation modes. One can find
. better algorithms (in abstracto)
- new regularities and the way to make use of them.
Let be a M object.
At first, an elementary generator (copy). Total length : M length (see Delahaye)
Then we have a non trivial generator. Total length : M (reduced) + generator length.
Then a set of generatorsand the way to combine them.
(See these gain aspects in HP)
Any knowledge acquisition is a gain. It means that we have reduced. To "master" (control) is to reduce the (functional) complexity of the things we control. Idem for the building of new classes, of better multipliers.