Notes for Friedrich Kittler “Protected Mode”
Key concepts: multiprocessing, protected mode, Von Neumann architecture.
Related theorists: Foucault, Tanaka-Ishii, Klaus-Dieter Thies, Turing, Von Neumann.
(156) The air-to-surface battle of 1991 demonstrated once again that among postmodern strategies of appearance, none is as effective as the simulation that there really is software.
Writing under as subject to Microsoft; worm, snake view now that command and data indifference of VNM split by protected mode.
(156) For one writes—the 'under' says it already—as a subject or underling of the Microsoft Corporation.
Alludes to early days of microprocessors when literary theorists and hobbyists could hack hardware.
(156-157) This worm's-eye view did not always prevail. In the good old days when microprocessor pins were still big enough for simple soldering irons, even literary critics could do whatever they wished with Intel's 8086 Processor. . . . The silicon chip, which was as stupid as the hobbyist and user, could accommodate all of this because the Von Neumann architecture recognizes no difference between commands and data. . . . This [binary coding] is an activity that only Alan Mathison Turing—when he finally had his universal discrete machine of 1936 at his technical disposition one World War later—is said to have preferred over all mnemonic aids and higher programming languages. But one this expulsion of spirit and language was completed, the machine's stupidity equaled that of its user.
Prejudice against thinking like machines, no matter how pleasurable to Turing and others (Simonyi in Lammers), built into design and other knowledge-power actions defining technology; pusler project and pmrek invitations to reconnect with machine operations.
(157) These good old times are gone forever. In the meantime, through the user of keywords like user-interface, user-friendliness or even data-protection, the industry has damned humanity to remain human. Possible mutations of this humanity into paper machines are obstructed by multiple malicious tricks.
One-way function of programming languages; one-way subjectivity of consumers: why inline assembler example is significant, as well as paying attention to initial one instead of zero required for correct hardware operation initial power on state of pmrek.
(158) At the risk of having already
done crazy long ago, the only thing one can deduce from all of this
is that software has obviously gained in user-friendliness as it more
closely approximates the cryptological ideal of the one-way function.
The higher and more effortless the programming languages, the more
insurmountable the gap between those languages and a hardware that
still continues to do all of the work. . . . The sum hides the
addends, the product the factors, and so forth.
(158) In any case, the subjects of the Microsoft Corporation did not simply fall from the sky, but first had to be produced like all of their media-historical predecessors—the readers of books, film audiences and TV viewers. The only problem now is how their subjugation can be hidden from the subjects in order that they fall in step with the global triumph march.
Politics of knowledge discerned from technical handbook written by Siemens engineer Thies.
(158-159) As concerns the politics of knowledge, perhaps only Siemens' engineers can tell it like it is, as Klaus-Dieter Thies did in the 80186-Handbuch: . . . Thus in multi-user systems, it is necessary that the programs and data of individual users are isolated, just as the operating system must be protected against user software. . . . in order to entangle civilian users in an opaque simulation.
Hiding other processing like reading silently relates computing to individualist democracy; multitasking to fool user usefully different from warfare logic, deception afforded by protected mode afforded by higher clock speeds.
(159-160) From the 80286 on, Intel's
processors are equipped with a Protected Mode that (in the
words of the Siemens engineer) protects the operating system from the
users, and through this protection, first allows the users to be
deceived. What began as the simple capability of switching between
the supervisor and the user stack in Motorola's 68000—naturally, a
secret rival system—is extended to system-wide procedure in the
separation of Real Mode and Protected Mode. . . . Thus it is
precisely in the silicon on which the prophets based all their hopes
for a microprocessed democracy of the future that the elementary
dichotomy of modern media technologies again returns.
(160-161) The innovation of Intel's Protected Mode consists only in having transferred this logic from the military industrial realm into that of information itself. . . . Although there may no longer be any written prohibitory signs that guarantee a power gap, the binary system itself encodes the distinction between commands and data, what the system permits and what, conversely, is prohibited to user programs. John von Neumann's classic architecture, which made absolutely no distinction between commands and data—unnecessary in an era when all existing computers were still state-secrets—disappears under four consecutively numbered privilege levels.
Foucault power argument from mute efficacy of technical implementation: look at ICs to understand society as form of technology studies.
(161) With this authority and through
this authority, conversations could in fact still be conducted;
however, technically implemented privilege levels draw their power
precisely from mute efficacy.
(162) As a result, a double shadows the analysis of power systems, that immense assignment that was Foucault's legacy. To begin with, one should attempt to abandon the usual practice of conceiving of power as a function of so-called society, and, conversely, attempt to construct sociology from the chip's architectures. For the present at least, it is a reasonable assumption to analyze the privilege levels of a microprocessor as the reality of precisely that bureaucracy that ordered its design and called for its mass application. . . . It is no coincidence that in the 80386, it is precisely the input and output commands that are protected by the highest privilege level—in an empire in which the public views the rest of the world only through the haze of television news, even the thought of foreign policy is a privilege of the government. . . . Whether there are better ones is beside the point because they would in any case also have to be bureaucracies; but a competition between different systems and different bureaucracies would as such already allow the subjects of MS-DOS to breath a little easier.
Classic power dilemma because highest protection allotted to input and output, yet this is how the user uses the machine.
(162-163) With its move out of front
offices and everyday language into the micrometer realm, power has
also changed the processes and the working surfaces. . . . In silicon
itself there can be, to borrow from Lacan, no other of the other,
which is also to say, no protection from the protection. . . . At the
level of the machine, then, protection mechanisms have no absolutely
protected hiding-place. Because microprocessors must despite
everything remain usable to users, that is, communicate with them,
Intel's Protected Mode describes a classic power dilemma.
(163) What such prohibitions conclusively demonstrate, however, is only the impossibility of perfect access control. . . . Once the difference has been rendered programmable, however, it is already vulnerable to all sorts of circumventions.
Vulnerable to circumventions, such as changing address boundaries of Real Mode (higher addresses trigger Interrupt 13) with an Assembler routine.
(164) A single subordinate clause in
the manual discloses that any address boundaries in Real Mode are no
more and no less than presuppositions programmed into the system
start-up. . . . Instead of the deliberately low default value that
the CPU automatically loads into the hidden sections of its segment
register at every reversion to Real Mode, programs could also set
completely different values.
(164) One hundred lines of Assembler, but only of Assembler, solve the problem of a postmodern metaphysics. At the risk of going crazy, they lead through MS-DOS beyond MS-DOS. Along with the infamous sound barrier at which the operating memory in DOS remains limited to a ridiculous mega-byte, all of the advantages for which Windows is praised dwindle to nothing. In a drastic paradox, it is precisely the most antiquated of all operating systems that provides the trap door out of the operating system. Intel's built-in blockages—which engage immediately in more complex operating systems such as UNIX, and subsequently even pick out those hundred program lines as illegal commands and refuse them—are powerless against stupidity.
Chaos in engineers empiricism against computability of theory, but saving power in danger of protected model also for machines.
(165) Such chaos does not reign at the
elevated level of information science, where the computability of
Finite State Machines and their ability to predict is argued over in
general, but rather at the modest level of the engineer's
(165) In other words, information science appears to be confronted with internal information obstacles. Information science must refer to the actual domain of code, even if the theory could generate completely different models (and should). And despite the will and belief of the code's developers, decodings are just as possible as they are rare. Long after the end of the print monopoly and authorship, the phantom of humanity apparently makes sure that mere opinions or even assertions of protection will continue to be recorded, as opposed to actually cracking the codes. A systems program must be created precisely to this end—to be used by programmers, to begin with, but in principle for machines as well.
Codes subject to same opacity and opacity as everyday languages.
(166) A discourse analysis whose elements are obviously not only words but also codes, would, of course, level the sacred distinction between everyday languages and formal languages. In light of the wonderful “orthogonality” that, for example, Motorola's processor series flaunts since the 68000, that would be heresy. The history of Protected Mode as a half-compatible, half-incompatible extrapolation of good old standards could, however, teach us that codes are subject to the same opacity as everyday languages. . . . However, Intel's new generation de-optimized precisely this speed advantage, while still permitting the synonymous commands to survive for compatibility reasons. Thus the code has achieved a redundancy that everyday language already boasted in Frege's wonderful example of “evening star” and “morning star.”
Opacity of machine languages as in the examples given fit with sourcery and vicissitudes of execution arguments, and go beyond rejection by Ong as artificial because they have a social history that exhibits parallels to characteristic of natural languages like redundancy; confounds desire of Hilbert to formalize everyday language.
(167) it becomes a Babylonian tower in
which the ruins of towers that have already been demolished remain
built-in. Protected Mode as both the enemy and co-existent partner of
a Real Mode that has already been superceded technically for some
time is computer history on chip.
And David Hilbert's
dreamlike program to clear out the opacity of everyday language once
and for all through formalization is undone not only at the clear,
axiomatic level of Godel or Turing, but already by the empiricism of
the engineers. Codes with compatibility problems begin to grow wild
and to adopt the same opacity of everyday languages that have made
people their subjects for thousands of years. The wonderful term
source code becomes literal truth.
(167) Another tower of Babel, on the chip.
Machines may have already taken command; impossible to test ICs independently of the producer.
(167) Turing's old idea of allowing the machines themselves to roll out their code may well have already secretly come true. Precisely because “the complex function of highly integrated circuits (aside from memory-ICs) can no longer, as in the case of a simple, logical connection, be checked by testing all of the possible input signal combinations,” tests that are independent of the producer are in order.
Where does free software fit for Kittler after quoting von Hofmannstahl: compare to Tanaka-Ishii or Chun on vicissitudes of execution.
Need to study his more recent texts, which may not be translated.
(168) Hugo von Hofmannsthal once ascribed the ability to read “what has never been written” to the “wonderful being” called Man. Similar crypto-analyses must become universal and mechanical in the chaos of codes that begins with the world-historical dismissal of everyday language in favor of a universal discrete machine.
Kittler, Friedrich A. “Protected Mode.” Literature, Media, Information Systems: Friedrich A. Kittler Essays. Ed. John Johnston. Amsterdam: Overseas Publishers Assocation, 1997. 156-168. Print.