Overview

In this essay Kittler frames the argument that software ultimately serves to conceal what is important in a computer system. The most clear way of using this seems to be in the manner in which software restrains and restricts the capacities of the user. Programmability is seen as a force that enables this concealment, and is, instead of an advantage, considered an indictment. While software is limiting and restrictive, so too is hardware, so it is difficult to tell exactly where and what the problem is.

I would say that ultimately the restrictions of software are the the same as the restrictions of using any system of abstracting models.

Prevalent through this essay is a theme of disgust over the notion of design, and a desire to appeal to bare mathematical concepts. Unfortunately, Kittler wildly misuses his reference to fractal theory, and his treatment of Turing’s computability too seems dubious.

A useful summary is located on mediamatic.net.

Notes

Kittler opens by noting the pervasiveness of computers, and that writing is more frequently (he says that the bulk of writing) is stored in computer memory, where it is no longer perceivable by humans. In fact, we do not even write anymore, but we use tools that are able to write by themselves.

Kittler is concerned with this technology driven evolution, and the technology that has enabled it. He specifically looks at the relations between Turing machines and microprocessors. The Turing machine can imitate any other Turing machine and compute any computable function. This fact means that computation is independent of hardware and that nature itself may be considered a Turing machine.

This is relevant from a perspective of languages, and Kittler suggests that programming languages have transcended ordinary language and have formed something of a tower of Babel founded on computational equivalence. He uses an analogy referencing fractals and the idea of self-similarity, but that does not seem to bear much resemblance to the idea in a real mathematical sense. What he is describing is a similarity among models and languages. Beyond this, the question arises of what language *does* when it has reached this universal state.

What follows is a close look at the process of executing WordPerfect on DOS. The language issues that seem to be problematic are the syntactic qualities stemming from DOS as a platform: the “exe” and “com” extensions, as well as the pervasiveness of acronyms. One of the problematic points is where the OS ends and the program begins. This relates again to the BIOS of the computer, which is another layer operating underneath even the operating system. In turn, underneath these too is additional hardware, in which information is only represented as differences in voltage. In turn, Kittler argues, the formalization of these is mathematical theory, which is composed of sentences with words and letters.

From this sequence of observations, we are to find that, really, there is no software. It is difficult to tell how this argument is supposed to coalesce. If we are to apply this sort of reduction to everything, (our bodies are cells, governed by biology, then by chemistry, then by physics), then we wind up with reductions that are of very limited use at all.

The current trend in design is that of concealment of the technicality of the underlying machine elements: BIOS conceals the hardware, the OS conceals the BIOS, and software conceals the OS. This concealment is “completed” by GUI design, and hardware level security. HCI has taught us that GUI design certainly does not always conceal the machine nature of the computer, it in fact reveals it in different ways. The fallacy of this is best exposed when programs catastrophically fail and reveal their inner workings in a most jarring manner (the Windows blue screens are most notorious in this regard).

Kittler sees the effect of the prevalence of software as pushing this trend of concealment. He invokes the idea of software as similar to one-way cryptographic functions, and as a result, cannot be easily reproduced or computed. Again, history has shown that modern programming languages, instead of being progressively harder to decompile, are much easier than in the past (especially languages such as Java or C#), similarly, all software is still universally accessible on one level or another, the communities that crack software or hack electronic devices such as game systems, portable music players, etcetera are tantamount evidence to this.

Software is seen as being able to conceal itself on every front, and have that concealment further enforced by patents and copyrights. The quality of undecidability and complexity has led too to the legal status of software as material, despite its immateriality. At this point in time, due to the internet and rapid ability for software to be copied, this nature has become much more complex, with the material elements being held to very tightly by publishers.

The capacity for software to function is limited and dependent on the power of the hardware, which ultimately limits the capacity for software to emulate other systems through its available memory.

Ultimately, Kittler concludes that software is ill fitted to take on the increasing complexity of problem solving that will be demanded in the world. Programmability limits the effective potential of any platform due to its opening and lack of focus. Hardware has the strength and power to simulate much more effectively real world systems.