I realize that before reading this post (made entirely out of Kittler's "There is No Software"'s extracts), it would be good to put it in context with current debates in software criticism. As Troels Degn Johansson recalls in "Mise en Abyme in Software Art: A Comment to Florian Cramer" (Read_me 2004), Kittler's claim that software has no existence without its hardware support has been countered by Florian Cramer, who understands the art of software in a conceptual way (thus related to Conceptual Art, but not only, as hte history that Cramer is retracing goes way back in time to a tradition of thinkers who used computational methods to extend their knowledge).

Just a note: this debate is a little weird since when you consider software as a "mental act", you are really talking about the program, not the software, and Florian Cramer, even if sometimes he doesn't make it clear, sure makes a difference between the 2. The problem, not discussed by Johansson, is that reducing software to a set of instructions (thus conceptual), all language-based, is forgetting the fact that software, implemented in the machine, has a socio-technical reality: it is an object and does have to be studied in the realm of the relfexion on technique. This does not mean that this study should be deterministic and not language-oriented. But there should be a reflexion on the differences between the symbolic machine that is language and the physical machine that is the computer (that may not be reducible to the symbolic). There was a little discussion on this blog about the "symbolic computer". Are the pragmatic achievements of language and the machine to be considered on the same level? Can we study programming language codes the same way that we analyze codes and conventions in languiage (mostly via semiology, these days)? 

So I respect and enjoy Cramer's contribution to software studies, because his works brings an exciting historical background to software studies through bringing on the forefront a cultural heritage of thinking in terms of programs and computing. But I don't think he provides direct methodological tools for software criticism per se; what is brings out is new cultural material via opening new fields in historical and aesthetics studies. His reading of computer codes as cultural conventions seem, if not wrong, a little fast. I want to say: "sure! but what else?". So I don't really see how Cramer's discourse could be opposed to Kittler's; they seem more complementary than opposed. Or maybe I am missing the point of the debate Johansson is mentioning?

Now back to the original post.

=+=+=+=+=+ 

Friedrich Kittler's rant in "There is No Software" is pretty exciting. Extracts:

"When meanings come down to sentences, sentences to words, and words to letters, there is no software at all. Rather, there would be no software if computer systems were not surrounded any longer by an environment of everyday languages. This environment, however, since a famous and twofold Greek invention, consists of letters and coins, of books and bucks.⁴ For these good economical reasons, nobody seems to have inherited the humility of Alan Turing, who, in the stone age of computing, preferred to read his machine's output in hexadecimal numbers rather than in decimal ones.⁵ On the contrary, the so-called philosophy of the computer community tends to systematically obscure hardware by software, electronic signifiers by interfaces between formal and everyday languages. In all philanthropic sincerity, high-level programming manuals caution against the psychopathological risks of writing assembler code.⁶ In all friendliness, "BIOS services" are currently defined as "hid[ing] the details of controlling the underlying hardware from your program."⁷ Consequently, in a perfect gradualism, DOS services would hide the BIOS, WordPerfect the operating system, and so on and so on until, in the very last years, two fundamental changes in computer design (or DoD politics) have brought this secrecy system to its closure."

"Electronics as literature; the linguistic agent ruling with near omnipotence over the computer system's resources, address spaces, and other hardware parameter."

"Whereas in the good old days of Shannon's mathematical theory of information, the maximum of information coincided strangely with maximal unpredictability or noise,⁹ the new IBM measure, called logical depth, has been defined as follows:

The value of a message [...] appears to reside not in its information (its absolutely unpredictable parts), nor in its obvious redundancy (verbatim repetitions, unequal digit frequencies), but rather in what may be called its buried redundancy - parts predictable only with difficulty, things the receiver could in principle have figured out without being told, by only as considerable cost in money, time or computation. In other words, the value of a message is the amount of mathematical or other work plausibly done by its originator, which the receiver is saved from having to repeat.¹⁰

Thus logical depth, in its mathematical rigor, could advantageously replace all the old everyday language definitions of originality, authorship and copyright in their necessary inexactness, were it not for the fact that precisely this algorithm intended to compute the cost of algorithms in general is Turing-uncomputable itself.¹¹"

"According to Brosl Hasslacher of Los Alamos National Laboratory,

...this means [that] we use digital computers whose architecture is given to us in the form of a physical piece of machinery, with all its artificial constraints. We must reduce a continuous algorithmic description to one codable on a device whose fundamental operations are countable, and we do this by various forms of chopping into pieces, usually called discretization. [...] The compiler then further reduces this model to a binary form determined largely by machine constraints.

The outcome is a discrete and synthetic microworld image of the original problem, whose structure is arbitrarily fixed by a differencing scheme and computational architecture chosen at random. the only remnant of the continuum is the use of radix arithmetic, which has the property of weighing bits unequally, and for nonlinear systems is the source of spurious singularities. This is what we actually do when we compute up a model of the physical world with physical devices. this is not the idealized and sere process that we imagine when usually arguing about the fundamental structures of computation, and very far from Turing machines.¹⁴

Thus, instead of pursuing the physical Church-Turing-hypothesis, that is of "injecting an algorithmic behavior into the behavior of the physical world for which there is no evidence,"¹⁵ one has rather to compute what has been called "the prize of programmability" itself. This all-important property of being programmable has, in all evidence, nothing to do with software; it is an exclusive feature of hardware, more or less suited as it is to house some notation system. When Claude Shannon, in 1937, proved in what is probably the most consequential MA thesis ever written¹⁶ that simple telegraph switching relays can implement by means of their different interconnections the whole of Boolean algebra, such a physical notation system was established. And when the integrated circuit, developed in the 1970s out of Shockley's transistor, combined on one chip silicon as a controllable resistor with its own oxide as an almost perfect isolator, the programmability of matter could finally "take control" just as Turing had predicted.¹⁷

Software, if it existed, would just be a billion dollar deal based on the cheapest elements on earth. For, in their combination on chip, silicon and its oxide provide for perfect hardware architectures. That is to say that the millions of basic elements work under almost the same physical conditions, especially as regards the most critical, namely temperature dependent degradations, and yet, electrically, all of them are highly isolated from each other. Only this paradoxical relation between two physical parameters, thermal continuity and electrical discretization on chip, allows integrated circuits to be not only finite state machines like so many other devices on earth, but to approximate that Universal Discrete Machine into which its inventor's name has long disappeared."

More texts by Kittler (German and English).