by Dominic Fox
Verso 294pp ISBN 9781786635471 £16.99
James Bridle turns a nice chiasmus: ‘the systems we have built to collapse time and space are being attacked by space and time.’ It’s an apt rhetorical device for conveying a sense of tragicomic reversal, of things rotating about a pivot: the first shall be last, and the last shall be first. Much of New Dark Age is a kind of compendium of real-world moral fables having this sort of structure, in which bureaucracies designed to bring things under control generate unmanageable entropy and corruption, the pursuit of efficiency leads to impotent entanglement in myriad unintended consequences, and the quest for knowledge founders on the shores of a vast and inconceivable ignorance. What is perhaps surprising is that Bridle suggests we should welcome the awareness of cognitive limitation that these calamities portend: the ‘new dark age’ of his title is not a threat, but a promise.
The opposition of human imagination to mechanical efficiency is a mainstay of Romantic humanism, which is perennially invigorated by stories of machines gone wild, perverse automata breaking down, running out of control or unexpectedly turning on their masters. If you wish to see, as Bridle does, the ‘re-enchantment’ of a world too much given over to planning, calculation and rational control, then you too may find such reversals of fortune strangely heartening. As US Senator Eugene McCarthy once said, ‘the only thing that saves us from the bureaucracy is its inefficiency. (I discovered this while playing the world-history simulation game, Civ V, in which McCarthy’s witticism is displayed when you succeed in inventing the Civil Service.) The trouble with New Dark Age is that it tries to wrap everything up into an overarching narrative about something that is supposedly happening to us right now, a story about how we have allowed our imaginations to be captured by something Bridle calls ‘computational thinking’, and need to recover our capacity to endure and engage creatively with uncertainty and unpredictability (you know: ‘negative capability’). The ‘we’ of this story is situated at the pivot-point of the fable, the point where things come swinging around and start hitting us in the face. There are some plausible reasons for this historical positioning – we are indeed in the middle of several simultaneous crises of our own making – but by the same token there is something suspiciously ahistorical about the proposed solution.
What is ‘computational thinking’? It is something a little like Heidegger’s concept of technological ‘enframing’, stripped of its grandiose historico-philosophical backstory. Increasing computational power has led us to frame our problems increasingly as the kinds of problems which happen to be computationally tractable, even when – as it often turns out – they really aren’t. We have ceased to believe the evidence of our own eyes, or the products of our own imaginations, and are bewitched instead by a kind of fallacious objectivity, an alienated form of knowledge which is legitimated only by our mistaken belief in its inhuman puissance (Blake might have called this the ‘single vision’ of Newtonian science). Hence the stories everybody enjoys so much about users of GPS-guided navigation systems driving themselves into lakes.
There is a kind of immanent critique of such overconfidence in computational methods embedded within computer science itself, which provides the means to think precisely about computational intractability – both the ‘absolute’ kind, as in the Halting Problem, and the ‘relative’ kind, as in the infeasibility of decoding a strongly-encrypted ciphertext within a practically useful timeframe. But ‘computational thinking’ on Bridle’s account does not appear to mean the things that computer technologists themselves think, but rather the way ‘we’ think about technology – a shared ideological mirage (to which technologists are also, admittedly, often susceptible).
Bridle generalises somewhat broadly about this ‘we’, in places misreading an ephemeral bedazzlement as a deep malaise. In practice, anyone trying to navigate around a large urban centre with the help of Google Maps on a smartphone will be accustomed to the necessity of distinguishing between the map and the territory, repeatedly comparing what the app tells them about where they are and which way they are facing with what the street names and landmarks around them seem to suggest. Once you have seen the little blue marker indicating your position suddenly scoot a few streets across while you remain standing still, the news that GPS offers advice of variable accuracy and utility will no longer seem all that surprising. Over time, users of advanced technologies learn to distinguish them from magic, as their practical affordances become more clearly defined and reveal their characteristic limitations and failure modes. If, in a few years time, we are deep into a fresh ‘AI winter’ precipitated by the collapse of inflated expectations concerning the cognitive abilities of recursive neural networks, then Bridle’s claim that we place an exaggerated trust in the power of algorithms will seem simultaneously vindicated and quaint: ‘we’ will perhaps have moved on to being wowed and seduced by something else.
Bridle’s argument concerning ‘computational thinking’ is intertwined with a more general consideration of complexity and alienation, notably in the chapter titled ‘Conspiracy’, which discusses conspiracy theories as failed (yet still symbolically indicative) attempts at a ‘cognitive mapping’ (the term is borrowed from Fredric Jameson) of an inexpressibly complex social and technological reality. Here, conspiracy theory is presented as simultaneously disordered, in its ‘bizarre, intricate, and violent’ attempt to cram that complexity into a narratively-tractable form, and exemplary in that it reveals ‘that our ability to describe the world is a product of the tools at our disposal.’ Bridle suggests that sufferers of Morgellons Syndrome, who imagine that their skin has been invaded by malignant fibres of unnatural origin, are in their own way perceiving and registering the threading of fibre-optic cables through the body of the earth, making symbolic sense of an ecological disquiet which they map into their own immediate frame of reference: ‘conspiracy theory is the dominant narrative and the lingua franca of the times: properly read, it really does explain everything.’
Conspiracy theories do not qualify as ‘delusional’ according to the accepted clinical definition because they are not abnormal for the emergent networked subcultures that foster and reinforce them: they are, locally to those subcultures, matters of consensus reality and thus, by social constructionist fiat, valid descriptions of the world. Bridle’s solution to the epistemological quandaries this induces is not to allow that there might be some other criterion of a theory’s validity besides its being accepted as valid by your local Morgellons sufferers’ support group, but to suggest that we mingle theories eclectically, sampling ‘from the myriad of explanations that our limited cognition stretches like a mask over the vibrating half-truths of the world’. He ends by arguing that the ‘grey zone’ of epistemic indifference is ‘a better approximation of reality than any rigid binary encoding can ever hope to be – an acknowledgement that all our apprehensions are approximate, and all the more powerful for being so.’
There is an ideological move being pulled here, which is to project onto reality itself the qualities of vagueness – ‘vibrating half-truths’, non-rigidity, approximateness – that typically arise in theories when they are confronted with phenomena of greater complexity than they are capable of expressing. Vagueness occurs within our descriptions when our powers of description are not adequate to their object. If we then describe reality itself as intrinsically vague, we can satisfy ourselves that our faltering apprehensions are nevertheless in keeping with the wibbly-wobbly, timey-wimey way of things. The important thing is not to be too ‘rigid’, because the world isn’t really like that. The underlying assertion of what the world is really like is, however, a matter of faith. (It can’t really be otherwise, given that we’ve abandoned any notion of knowing what anything is like precisely). It preserves us from the embarrassment of having to tell Morgellons sufferers that they are really wrong about the material causes of their distress, but it also saves us from having to commit to really going along with their beliefs (or doing anything about them). We can treat them as part of a rich tapestry of diverse exercises of ‘limited cognition’.
A similar move is entailed when Bridle projects onto information systems such as YouTube’s recommendation engine an intrinsic perversity and malignancy, out of which horrifying phenomena – nasty Peppa Pig videos! – arise according to the inscrutable logic of the algorithm itself, without considering the adversarial terrain into which such systems are deployed. Search engines such as Google’s, and recommendation engines such as YouTube’s, present a rich target for agents seeking to game the engine’s utility function in order to obtain a higher profile and a greater share of end-user attention. Some players in this game are just trying to get a larger slice of the advertising revenue pie; others, engaged in pranks or ideological warfare, are trying to jam or degrade rival signals as well as promoting their own. All of the perverse outcomes Bridle observes on YouTube are generated by human strategic behaviours, and essentialising these into an unfathomable darkness coagulating at the heart of the machine is really seriously missing the point: the machine is stupendously boring, a glorified filing clerk-cum-statistician. The darkness is in us.
I suggested earlier that Bridle’s Big Answer to the myriad problems his book addresses, from the lightning crashes of glitching high-frequency trading algorithms to the dark satanic mills of Amazon warehouses, was ‘suspiciously ahistorical’. I think that this is so because the book is fundamentally a defence of an aesthetic and ideological position – Romantic humanism – which constructs its antagonist opportunistically from the materials contemporaneously at hand. Accordingly,‘computational thinking’ has very little to do with computers, finally; it’s more a convenient proxy for alienated reason, to which Romantic humanism has always opposed the spontaneity and authenticity of direct perception and imaginative creativity. In discussing ‘big data’, for example, Bridle makes the peculiar claim that ‘the big data fallacy is the logical outcome of scientific reductionism: the belief that complex systems can be understood by dismantling them into their constituent pieces and studying each in isolation’. It is wholly conventional for a Romantic humanist to oppose ‘scientific reductionism’, insisting on the faculty of human imagination as providing an essential unifying context for understanding, but ‘big data’ is simply not reductionist in the required sense: it studies the constituent pieces of its data set not in isolation, but aggregated to the point of indistinction. Its mode of understanding is not combinatorial – we understand A, and we understand B, and therefore we understand the system made up of A and B – but actuarial: we observe that As and Bs tend to come along together, for reasons we probably should be more curious about.
The essential promise of ‘big data’ is that it may sometimes be possible to surface characteristics of the domain under analysis that we don’t initially know how to specify in detail: we can find things we we didn’t start out knowing how to look for. Bridle later quotes the researchers behind Tri Alpha, an approach to machine-enhanced investigation of the space of possible fusion reactor designs, who describe their system as ‘attempting to optimise a hidden utility model that human experts may not be able to express explicitly.’ There is something of the sublime in this: that which is ‘inexpressible’, for which we have no concept, can take form as a ‘model’ which then becomes a tool for further investigation and reflection. Bridle seems to see this for a moment, then want to move on: on the one hand, ‘these technologies open up the possibility of working effectively with such indescribable systems’, forcing us up against the limits of our existing powers of description and definition; on the other, ‘admitting to the indescribable is one facet of a new dark age: an admission that the human mind has limits to what it can conceptualise’; and, of course, ‘not all problems in the sciences can be overcome by the application of computation, however sympathetic’. Rather than being driven by the sublime to conceptualise new things, developing new formal and aesthetic capabilities in the light of novel experience, Bridle wants to revert immediately to a posture of immobilised epistemic humility: we should acknowledge that we just can’t understand what lies beyond the limits of our language, and seek contentment and enchantment in the fact of being encircled by impenetrable gloom.
From a modernist perspective, this is a reactionary argument: it closes down lines of advance, forbidding us to make use of the experience of the sublime to retool and extend our conceptual apparatus. From a Romantic humanist perspective, it is one which conserves the value of certain distinctively human strategies for living with uncertainty and ambiguity: spirituality, imagination, the consoling and illuminating power of fable. At the conclusion of Lars von Trier’s film Melancholia, a slingshot effect propels the rogue planet Melancholia, which had swung by Earth in a near-fatal near-miss, back into a collision course with our planet. Recognising that they are finally, inescapably doomed, the surviving characters (minus the cowardly scientific rationalist, who immediately commits suicide once he realises his calculations concerning the trajectory of the planet were mistaken) construct a ‘magic cave’ – a makeshift teepee – in the garden and take shelter within it, assuring their child that this structure will protect them. This act of make-believe, in the face of certain annihilation, is represented as a warmly human demonstration of care and kindness, in explicit contrast to the scientist’s cold emotional desertion. (Of course they all die anyway). Bridle, like von Trier, would rather we were story-making fabulists than truth-seekers engaged in ‘the endlessly destructive pursuit of ever more finely grained knowledge, at the expense of the acknowledgement of unknowing.’ But what is the value of this ‘acknowledgement’, to which we are constantly exhorted? What consequences are we supposed to draw from it?
The conclusion is ultimately, and I think not coincidentally, religious: consider the lilies of the field. Give up the ‘illusory promises of computational prediction, surveillance, ideology and representation’ and focus on the ‘here and now’ in which we can recover an untrammelled moral agency. It is perhaps easier to live with an unknowable future than one in which we know, with as much certainty as it is possible to know anything, that the climate of our planet is changing in ways that are going to threaten the continued existence of human life. But if we do know this – and we really do – it is through precisely the kind of immense, detailed, data-driven modelisation of reality that Bridle spends his entire book trying to persuade us to regard as pernicious and deceptive, and any exercise of our agency in the here and now that is not informed by this knowledge is both culpably self-deceptive and self-defeatingly futile. Magic teepee thinking is always an option, of course, and perhaps if we’re simply doomed anyway it doesn’t matter. But this seems like a weak premise for what is meant to be a compelling and inspiring moral argument.