Propaganda and Decision Ecologies

How might future research into digital culture approach a “post-digital” age?

One of the many problems comes from the discourse of ‘the digital’ itself: a moniker which points towards units of Base-2 arbitrary configuration, impersonal architectures of code, massive extensions of modern communication and ruptures in post-modern identity. Yet, it would quite difficult to envisage a ‘post-computational’ break from these discourses – and with good reason: for the actual specific structures upon which computational experimentation arise, are never really discussed at length. I’d like to consider the notion that before we ever entered a digital age, we live in an ecology of decisions, or better yet, decisional ecologies.

————–

First a note on the context. My thesis attempts to address how computational aesthetics is possible when understood through the conjunction of two frameworks: the history of meta-mathematics and a realist ontology. I’ve probably lost everyone right there, but in a different way, the research only seeks to incorporate the mathematical story of function, logic and proof (and the failure thereof) back into the agency of aesthetic expression. Not just for the basis of computation, but also its aesthetic. This might be the case for how artists use computation politically or sensually, but also why aesthetics should be present at all in, what is essentially, an automated vehicle for proving theorems.

It is widely understood that the theoretical basis of computation, derived from Alan Turing, is to create a formal system of logic, which when automated, would solve particular mathematical problems put into function. What is not necessarily understood is the mathematical context to that basis: the foundations of mathematics was already precarious, way before Turing published his landmark 1936 paper, On Computable Numbers, with an Application to the Entscheidungsproblem. It is a precariousness which has been built-in to computation from its very inception.

The key word of that paper, its key focus, was on the Entscheidungsproblem, or decision problem. Originating from David Hilbert’s mathematical school of formalism, ‘decision’ means something more rigorous than the sorts of decisions in daily life. It really means a ‘proof theory’ [1]

Decision is what happens when a formal system of function is constructed in a sufficiently complex way, that an algorithm can always ‘decide’ a binary, yes or no answer to a mathematical problem, when given an arbitrary input in a sufficient amount of time. It does not require ingenuity, intuition or heuristic gambles, just a combination of simple rules and a careful avoidance of contradiction.

The two key words here are ‘always’ and ‘decide’ – the progressive end-game of 20th Century mathematicians, like Hilbert who were addicted to the buzz of demonstrating proofs, sought one simple totalising conceptual system, to decide every query, silence any dissidence, and work towards absolute knowledge. All Turing had to do was make explicit formalism’s implicit computational treatment of formal rules. Later on Turing would call this an ‘effective’ or ‘systematic procedure.’

Effective procedures decide problems, they resolve puzzles and provide winning positions in the game of functional rules and formal symbols. In Turing’s words “a systematic procedure is just a puzzle in which there is never more than one possible move in any of the positions which arise and in which some significance is attached to the final result.” [2]

Already in 1936, Turing showed how machinic decisions as mathematical ideas could model and replace human ones, and how, given a sufficient complexity certain effective procedures, (like Universal Turing Machines) could simulate the functional decisions of other effective procedures. Ten years later, Turing and John von Neumann would independently show how automated physical machines, and general purpose computers offered the same thing. From that moment on, decisions manifested themselves in materials. Programs were simply proofs. Code is function. Before there was Shannon’s information theory and the encoded logic of messages, we had Hilbert and Turing’s computational structuring of information in an underlying form of decision.

Yet, on a meta-level, Turing was fascinated by what decisions couldn’t do, just as much as their ability to automate proofs and provide flexible universal function. Unlike Hilbert, Turing was not interested in using computation to solve every problem, but as a curious endeavour for surprising intuitive behaviour. The most important of all, Turing’s halting, or printing problem was influential for him, precisely as it was undecidable, a decision problem.

We can all picture the halting problem, even obliquely. Picture the frustrated programmer or mathematician starting at her screen, waiting to know when an algorithm will either halt and spit out a result, or provide no answer. The computer itself contains all the totalising knowledge, the programmer just has to know when to give up.

But this is a myth, inherited with a bias towards human knowledge, and a demented understanding, of machines as infinite calculating engines, rather than concrete entities of decision. Turing didn’t understand the halting problem in this way, instead he understood it as a contradictory example of decisions failing to decide on each other, on the account that there can never be one totalising decision or effective procedure. There is no guaranteed effective procedure to decide on all the others, and any attempt to build one should be regarded with suspicion. Turing suggested that ‘propaganda’ was more appropriate to this meta-level than simple proof. [3]

Programs might be proof-deciders, but there is no general decision-procedure for solving all problems. Undecidability then, is what happens when formal systems and decisions (whether conceptual or physically embedded in computation) can never ultimately decide on a solution, in the absence of a general systematic procedure. Decisions, including human ones, are doomed to decide, not to know.

———–

Now there’s a lot in that, but what does any of these mean for aesthetics? One could point towards a couple of moves. One of the potential insights gained from the conflicting consequences of decision, is it re-ignites something about the philosophical remit of aesthetics, descended from Immanuel Kant: aesthetics is not in the business of solving problems. Moreover, an aesthetic effect is rule-based, yet paradoxically its power, politically and sensually, lies in having no explicit rules to follow. Aesthetics might be in the remit of applying and endorsing decision problems.

Yet, we should also be wary of sticking with Kant’s functional basis for grounding aesthetics in the basis of transcendental thought alone. This is where the realist ontological aspect of the thesis kicks in, because whilst the meta-mathematical basis for computation is important for understanding the origin of computation as decision, it doesn’t grasp what sort of critical effects these decisions have within culture, contemporary arts practice and big business. It doesn’t grasp the level of adaptability required for negotiating different types of decisions in computational culture, and how the binary logic of ‘provable’ and ‘non-provable’ simultaneously offers the means of subversive expression whilst condemning us with hegemonic means of control. It does not address the ecological reality of decision that takes place.

What is clear is that our world is no longer simply accountable to human decision alone. ‘Culture’, is no longer simply guided by a collective whole of human decisions, nor is it reducible to one ‘natural’ collective decision. Rather the collective world is comprised and composed of an ecology of decisions: a collection of decision-centric autonomies, which surround and often presuppose any individual or collective decisions that get underway, both social and technological. Before there was ever the networked protocol, there was the computational decision, an effective procedure to decide in the first place. Decision ecologies already takes place before we enter the world, implicitly coterminous with our lives: explicitly determining a landscape upon which an individual has limited manoeuvrability.

Decisions are everywhere and in everything. Look around. We are constantly told by governments and states that are they making tough decisions. CEOs and Directors make tough decisions for the future of their companies. ‘Great’ leaders are revered for being ‘great decisive leaders’, not just making decisions quickly and effectively, but also settling issues and producing definite results. Even the word ‘decide’, comes from its Latin origins of ‘decidere’, which means to determine something and ‘to cut off.’ Algorithms in financial trading know not of value, but of decision: whether something is profit or loss. Drones know not of human ambiguity, but can only decide between kill and ignore. Making a system which decides, between two or more routes, means cutting off and excluding all other options leaving a final result at the end of the procedure. Making a decision, or building a system to decide a particular ideal or judgement, must force other alternatives outside of it. Decisions are always-already embedded into the framework of action, always already deciding what is to be done, how it can be done or what is threatening to be done. It would make little to sense to suggest these entities ‘make decisions’ or ‘have decisions’, it would be better to say that they are decisions: they decide what choices can be made within them.

The question is no longer simply ‘who decides’, but now, ‘what decides?’ Is it the cafe menu board, the dinner party etiquette, the NASDAQ share price, the new Google Search algorithm, railway network delays, unmanned combat drones [4], the newspaper crossword, the javascript regular expression or the differential calculus? It’s not quite right to say that algorithms rule the world, whether in algo-trading or in data capture, but the uncomfortable realisation that real entities are built to determine provable outcomes time and time again, in the midst of automating questionable procedures: most notably ones of profit, revenue and ideology. One of the more stronger examples is George Dantzig’s Simplex Algorithm, (whose origins began in multidimensional geometry), which always decides solutions for large scale optimisation problems. It’s proliferation and effectiveness has been critical for 30 – 40 years of successful capitalist production, deciding almost everything from bus timetables, work shift patterns and trade shares. According to optimisation specialist Jacek Gondzio, the simplex algorithm runs at “tens, probably hundreds of thousands of calls every minute.”

Turing was almost prophetic in calling it propaganda, despite him writing about mathematics and not aesthetics or politics. For what is propaganda if not an effective method of disseminating and structuring information through mass ideological control, decision and persuasion? Political Propaganda, as Frederic Charles Bartlett knew [5], was primarily a decisive method of suggestion, not simply designed to control psychological behaviour, but to acquire specific, effective results through purposeful action.

Propaganda operates as if it can produce idealised solutions to problems, but in its operation, must hide uncomfortable paradoxes which allow its communication to occur in the first place. Perhaps a post-digital realisation, of culture might address newer forms of propaganda emergent in computational culture: not posters, pamphlets, zines and broadcasts, but videogames, gamification, devices, spy-ware, apparatuses, services and subscriptions: each one only allowing certain outcomes to be realised, each one already deciding (or propagating), a limited number of routes, which users mistake for their own freedom. If there is one thing Silicon Valley would love to solve it’s how can we tell if a problem always has a solution: and whenever they come up with one, it usually has a market to satisfy.

But I’ve stopped writing about aesthetics. Or have I? Remember when the far Left used to be good at propaganda? If we were never post-ideological, then we have never been post-propagative either. Decision ecologies present numerous problems, primarily as they decide what can be done and what can be negotiated in crafting anything whatsoever. What is required of aesthetics nowadays, is not an explicit rejection of decision, but an implicit affirmation of a decision problem. Whenever artistic practice seeks to undermine decision and open up a space for discussion, it must suddenly realise that purposeful action is required: which is to say that aesthetic practice might present better propaganda. Even open source software is propaganda of sorts: a principled ideal embedded in computation, where free use as a precondition of its modification, is already decided on account of its dissemination.

This sounds absurd, but why not? Decisional ecologies do not have a teleology, yet they are all we have, and aesthetics has a historical prevalence, not from burrowing towards the formal truth of rules, but addressing why rule based judgements and decisions have no rules to follow. Exploitative hacking for example is the very sort of purposeful action which undermines decision in a given systems, whilst simultaneously demonstrating new automated proofs which were previous undecided. [6]

Decisional ecologies are grounded in social and technical entities, which makes certain analogous practices stretch over disparate mediums. The problem with computation is that the skill set required to make regulated systems undecidable is set quite high. But one can learn from other areas.

For example, one can cite Julius von Bismarck’s famous Image Fulgurator (2007-present) series of interventions. von Bismarck’s invention projects an optically triggered hidden image (usually at press events), which is only detectable after the event, after it is triggered by flash snapshot. The Fulgurator itself is a modified SLR camera which has a reverse trigger flash sensor, only secondarily triggered by external camera flashes. This device stealthily projects an image, the moment a flash is taken, effortlessly infecting every possible snapshot, without anyone knowing.

Crucially, von Bismarck’s interventions actually take place in the events themselves: a cross on Obama’s podium in Berlin, or a giant ‘NO’ beamed above the Pope. It gives such technical structures something that they cannot decide on (to prevent the technology falling into commercial hands, von Bismarch even took the decision to patent it).  In such events where decisions are tightly controlled, and contingencies are calculated, von Bismarck demonstrates an exploitative method which undermines this particular concrete environment of decisions, and exposes a certain undecidability: a hole or gap within the procedure previously unnoticed.

Because decisional ecologies are never total, it is always possible to undermine totalising systems of propaganda, no matter how extensive. But doing so is not pre-given, and requires not totalising knowledge, but a negotiation of what present decisions can currently afford. Such is the focus of a decisional ecology, it can only spit out more decisions: what matters are the ideals and the future conflicts that are embedded in its place within this ecology.

—————–

1 David Hilbert, ‘Probleme der Grundlegung der Mathematik’ [Problems Concerning the Foundation of Mathematics], Mathematische Annalen, trans. Elisabeth Norcliffe 102 (1930), p. 3, cf. 1-9.

2 Alan Turing, “Solvable and Unsolvable Problems,” in Science News, #31 (1954) p. 18, cf. 7 – 23. citations hereby taken from Jack Copeland, The Essential Turing: Seminal Writings in Computing, Logic, Philosophy, Artificial Intelligence, and Artificial Life: Plus Secrets of Enigma, edited by Jack Copeland, (Oxford: Oxford University Press, 2004), p 590. cf. 582-595.

3 Quoted from  Turing, “Solvable and Unsolvable Problems,” p. 588 (emphasis added)

“This statement is still somewhat lacking in definiteness, and will remain so [...] The statement is moreover one which one does not attempt to prove. Propaganda is more appropriate to it than proof, for its status is something between a theorem and a definition. In so far as we know a priori what is a puzzle and what is not, the statement is a theorem. In so far as we do not know what puzzles are, the statement is a definition which tells us something about what they are.”

4 See Daniel Suarez’s science fictional (but self researched) account of robots and drones automating military killing decisions in Daniel Suarez, Kill Decision, New York, Penguin Books, 2012.

5 F. C. Bartlett, Political Propaganda, (Cambridge, Cambridge University Press, 1940).

6 See Sergey Bratus, Michael E. Locasto, Meredith L. Patterson, Len Sassaman, and Anna Shubina, (2011) “Formal Applications of Formal Language Theory – Technical report TR2011-709”, Dartmouth Computer Science. See also; Sergey Bratus, Michael E. Locasto, Meredith L. Patterson, Len Sassaman, and Anna Shubina, “Exploit Programming: from Buffer Overflows to “Weird Machines” and Theory of Computation”, ;login:, December 2011 and for a more detailed analysis on fundamental x.509 security protocols, see Dan Kaminsky, Len Sassaman, and Meredith Patterson, “PKI Layer Cake: New Collision Attacks Against The Global X.509 CA Infrastructure”, Black Hat USA, August 2009, <http://www.cosic.esat.kuleuven.be/publications/article-1432.pdf.> last accessed September 1st, 2013.

8 thoughts on “Propaganda and Decision Ecologies

  1. That was a good read – thanks! Enjoy both the content and manner of analysis although there are few things I have a problem with that perhaps you can clarify.

    Around aesthetics – not that I am a mathematician, but am curious why you position logic and proof – computation, outside of aesthetics? I don’t see on what grounds (clearly not aesthetic) you could eliminate aesthetics from it? Perhaps in the same distinction between an aesthetic and Aesthetics that you use later?

    When you are discussing Kant and say aesthetics is not concerned with problem solving it also has to be accepted that this does not mean problems don’t have aesthetics. I think this is the point you are making when you distinguish between the ability of aesthetics to solve and to apply logic?

    But what I am unclear about is that if logic as you position it is capable of residing in materials only post-Turning then what was the state of aesthetics prior to Turing. Was it incapable of applying logic? I would think not as aesthetics was a political method prior to Turing surely? Really interested to see how you understand that so looking forward to your reply.

    1. Thanks for the comment James. Sorry for lack of response to all the comments on here – had an unexpected couple of days away, and no signal (which was quite good in fact).

      Around aesthetics – not that I am a mathematician, but am curious why you position logic and proof – computation, outside of aesthetics? I don’t see on what grounds (clearly not aesthetic) you could eliminate aesthetics from it? Perhaps in the same distinction between an aesthetic and Aesthetics that you use later?

      This is my fault for not making more of the background stuff explicit (which is difficult in a 2200 word draft). It’s not so much placing computation outside of aesthetics, but understanding how the undecidable processes within logic and proof may account for a computational aesthetics – and in a more concrete form – computational art. It’s certainly not an elimination, but of a affirmation of repeated non-failure.

      When you are discussing Kant and say aesthetics is not concerned with problem solving it also has to be accepted that this does not mean problems don’t have aesthetics. I think this is the point you are making when you distinguish between the ability of aesthetics to solve and to apply logic?

      No, no – that’s the opposite of my point. Aesthetics is, in this argument, a decision problem. Aesthetics does not have any rules to follow, to complete a function, (which is another way of saying that aesthetics isn’t functional), but does originate from normative rules, and is a rupture of those rules. It is the rupture of unexpected computation, between the moments when computation is expected to compute something, or decide a result. I’m sure many mathematicians think effective solutions are the real aesthetics, but in my eyes, that is more condemned design, rather than sensual rupture.

      But what I am unclear about is that if logic as you position it is capable of residing in materials only post-Turning then what was the state of aesthetics prior to Turing. Was it incapable of applying logic? I would think not as aesthetics was a political method prior to Turing surely? Really interested to see how you understand that so looking forward to your reply.

      Very good point. And to be honest I don’t have an answer – as this is very particular to computational art. However, from my analogous example of the image fulgurator, I’m trying to point towards certain situations whereupon something material is exploited, and something unexpected arises. Or how certain artistic practices of unexpected behaviour are deliberately foregrounded for aesthetic purposes and aesthetic judgements (hence distinguishing unforced errors from arts practice). Certainly aesthetics did reside in materials before Turing, but post-Turing, decision had a potent grip on infrastructure. Now certainly the image fulgurator is not computation, it is not beholden to proof and logic – it does not warrant the level complex computability that is required. However, the same can be said that computation is not a closed off realm of separate complexity, which is not singularly contained in the world. Thus, whilst I don’t have an answer, I guess I should point towards a certain broad application of decision of which computation has provided a salient hand.

  2. I also enjoyed the text – especially where it begins to open up broader political questions around totalizations, and freedom of choice? I’d be interested now the politics of this opens up to other mathematical problems (set theory, for instance). I am interested in how the “problem” of undecidability allows us to break out of formalisms and ideological totalizations – are you simply saying that decidability is ideological (and complexities and paradoxes are hard-wired)?

    1. Hi Geoff.

      Glad you enjoyed it. Yes, I hope I am opening up those questions. Turing’s negative solution to the decision problems (that is remains a problem) has a historical equivalence to Cantor’s unsolvable Continuum Hypothesis (proved to be undecidable by Godel and Cohen – later on)

      There are two connected issues here: one is the connection between decision and solutionism (as Morozov terms it) which is very similar to what I’m talking about. That for every natural, political, human, cultural and systemic problem, silicon valley and other start-up cohorts think that computation can solve it, with coercion, social ranking, points or digital nudges. Decidability is ideological, if, in Marx’s sense, that users, and digital natives alike, adopt and adhere to decisional principles that are not their own. All the adapting is on the side of the consumer.

      The second is the difference between choice and decision. Sometimes they are treated differently: that we make choices and then decisions. But hopefully, what a decisional ecology should pinpoint is that decisions often decide, and thus structure, choices before they are made (another ideological point): moreover, this isn’t about a lack of knowledge – real technical things decide effective outcomes before we even choose whatsoever. And yes, complexities and paradoxes are hardwired into computational decision – one cannot ‘undo’ the general purpose of general purpose computation.

  3. I’ll join the choir: interesting read. And, also how our comments reflect different aspects of your text. Personally, I was carried away by your presentation of Turing, first of all – hinting at the problem of decidability as an ecology (episteme?), and not least his insistence on what decisions couldn’t do/decision problems – e.g., failing in deciding on each other. As you present it, this is something that materializes at a later time in history, but as such, the problem itself is more profound and exists “before we enter the world.” This widens the scope of Turing (and mathematics) considerably :-)

    As you suggest, this allows us to critique video games, services and other decision systems, and not least the application of certain decision systems in a really wide set of systems (as you describe).

    But my next (incomplete) speculation is also: How would you relate this to theories of language? I cannot help but notice the ironic (language) play in the decision problem that may be related to not only ‘”who [speaks] decides”, but now, “what decides” [speaks]“‘ (a différance?). And, of course, language is often referred to as having a profound relation to our being in the world – as you describe decision problems. This relation between the production of meaning, and the production of decision is intriguing (perhaps what produces our reality today?)

    In relation to aesthetics, is the “affirmation of a decision problem” a mimesis of the production of reality? (i.e., a poiesis)

    1. Thanks Christian,

      Glad you also enjoyed it. The use of ecology is two-fold: to understand decision in a realist sense – that problem solving automations to decide things outside of human access (and thus also support it) – and secondly to enforce the interconnected nature of decisions – that decisions often decide (and fail to decide) on each other. As you say, it materializes at a later time in history – we only seem to retroactively acknowledge that something could have been otherwise, but only after it has been decided. The question is how the artist can create a situation which causes a decision problem?

      As far as language is concerned – I’m wary of getting bogged down into the philosophical straightjacket of language (as Badiou would put it, but then he would as a mathematician). Certainly undecidable problems affect formal language theory in computation: where executable code has within in it, an implicit exploit which is always-already capable of being ruptured (made explicit). There is definitely a critical question of who/what decides, and who/what decides in the interest of the commons – but I’m interested in the social and technical of ‘Being in the world’, as well as the very decisional things that “do” the deciding in the social. It’s almost not enough that decisions produce reality, decisions decide reality, containing meaning, production and the agency of the artist.

      In relation to aesthetics, is the “affirmation of a decision problem” a mimesis of the production of reality? (i.e., a poiesis)

      Yes, sort of. The affirmation of a decision problem is a very particular type of affirmation. Unlike Deleuzians (a.k.a. Massumi) I’m not one for affirming literally everything as novelty, because I have to be careful about distinguishing the artistic practice of decision problems, from unintentional decision problems in general. Thats not to say they aren’t linked – both occur as real events, and both are capable of being transformed into each other. Rather, the affirmation of the decision problem is about a judgement (also undecidable) from the artist and beholder about the contingency of a particular decision problem. It’s a late night – I might explain this better on Monday. :-)

      1. I like your phrase “It’s almost not enough that decisions produce reality, decisions decide reality, containing meaning, production and the agency of the artist.” I think I agree, but I’d like to know more about this from your point of view/theoretical ground. Yes, lets talk some more… it is getting late on a Saturday night for such speculations :-)

Comments are closed.