The Blind Eye and the Vision Machine


The Stanley R. Mickelsen Safeguard Complex, North Dakota, USA.
Became operational September 1975, deactivated eight months later.

This text is based on a talk given at a symposium the London College of Communication on January 14th 2016 to mark the opening of The Forest of Things. This talk and draws together a few different ideas I’ve been thinking about over the last year around the status and place of the photograph today, and expands on some of the darker implications of algorithms and photography which I first speculated about in An All Seeing Eye. In The Forest of Things, the graduating show of the the 2015 masters degree in Photojournalism and documentary photography is at London College of Communication until January 22nd.

Empiricism, the belief that knowledge comes from direct experience, has been at the heart of western understanding for several centuries, and in turn the human eye has been at the heart of empiricism, sight valued above all other senses. The camera was conceived of as a sort of mechanical extension of that sight, which replaced some of the demands on the living eye to be physically present at an event, and which opened up knowledge which was beyond what the human eye could perceive unaided. But the camera still ultimately depended on the living eye to interpret and understand the images it produced. What I would like to somewhat provocatively suggest is not only is this is now changing towards an ever greater emphasis on the computational analysis of imagery, but that we are perhaps unwittingly also preparing the groundwork for us to be permanently locked out of the role of seeing and interpreting, whether we want this future or not.

Soon after its invention photography was readily integrated into a range of authoritarian structures. The camera satisfied the expansionist desire to know all and control all, by apparently offering us the possibility of unlimited seeing all through it’s photographs. With more time the camera of course also became part of a broader, more democratic culture, as a tool of reflection and expression. It seems a very contemporary angst that this democracy of the camera has given rise to a world where there are too many images, but it is not a new one. The Weimar cultural critic Siegfried Kracauer decried what he called the blizzard of images, and the way that this storm challenged photography’s ability to bestow meaning. Kracauer’s world was very different one from our own, a world where I would say technology and technological progress still seemed to offer the possibility of an almost omnipotent vision, an all seeing eye encapsulated in the blind stare of high technologies like the Cold War radar system featured above. Today we live in quite a different world, one which is so inundated with imagery as to make Kracauer’s blizzard of images seem like a light frost. But the fact of photography’s abundance, so often quoted, so often fretted over, is I think not nearly as interesting as the form those images take.

In Kracauer’s time the image was a material object, significant in that the image could truly be said to be a trace of the thing it recorded. Because of this materiality it was also something which was inherently visual, the photograph could be held and viewed in the hand. Today neither of these things can be said to be the case, digital technology has democratised photography and made possible an explosion in quantity, but it has also led to a more profound change in that the massive bulk of images are no longer really physical nor visual, they are alphanumeric data, pure information, inherently not visual things. I think this fact is significant insofar as it is increasingly causing photography to intersect with one of the other technologies which is defining our age, the algorithm. Photography had tended to be difficult fare for algorithms, partly because of it’s material form, no longer an issue in the digital age, but perhaps also because of it’s complexity and subjectivity. In the 1980’s, thinkers like the French philosopher Paul Virilio began to anticipate the coming of vision machines, essentially algorithms which could not only see but also understand images. Facial recognition was an early example of this, but the efforts involved were often huge and the results often crude. Today we are seeing algorithms that are ever more capable when it comes to sorting, sifting and understanding visual material, and have that material readily available in massive quantities to practice on. Up until now many of these algorithms have only been demonstrators, and appear rather like parlour tricks, more often amusingly inept than threatening. As they reach the real world their roles are for now are mostly supplementary, so far supporting people rather than supplanting them.

But even in this role these technologies pose interesting questions about the extent to which they are not only guiding and advising us, but also shaping us. As much as we feedback into and refine the algorithms we create, perhaps they are starting to do some similar to us. When man gazes into the machine, the machine gazes into man. It has been suggested that the result of algorithms playing such a prominent role in social platforms is that they are increasingly serving to shape interpersonal interactions which they have no business being involved with. In the case of dating websites it has been said that computers are now breeding people. These ideas have particularly strong implications in photography’s old stomping ground, the repressive realms of policing, intelligence gathering and warfare, major growth areas for automated technologies that reduce the intensive manpower needs of these fields, and offer to remove the personnel of the security services from harms way. Given the huge advances and investment in these areas it seems to be only a matter of time before technologies which are able to search, fix and kill without human intervention features of a battlefield somewhere. Indeed I suspect that one day we might look back at the era of piloted drones with the same sense that we now regard the early pilots of the First World War, as something which is quaintly romantic in it’s crudeness and it’s dirty violence, in contrast to the cold, distant killing of today, or tomorrow.

To return to Paul Virilio, in an interview given the same year as the publication of The Vision Machine, he spoke of reading a science fiction novel about a world where cameras had become so ubiquitous that they were now even being inseminated into flakes of snow, which were released on the world, seeing everything there was to see, and leaving no blind spots. When asked what he believed we would dream of in a world so saturated with imagery and the machines that produce them he responded that we will likely dream of being blind. What I would like to suggest to you is that perhaps we are starting to reach that point, where images dominate our world and confound our understanding so much that the thought of blindness might even start to feel like a relief. But we have also perhaps begun to move past it, and perhaps we are responding to that overwhelming feeling by starting to relinquish the task of to interpret and understand, and passing this burden on to the machines. We are allowing these algorithms a part some of the most important, powerful roles our societies have, and I’d suggest we are also starting the process of locking ourselves out, as machine vision develops in forms which are beyond our perception, as machines are built to see with technologies designed primarily for the understanding of other machines, not human eyes. Vision is no longer just mediated through technology as with traditional photography, technology is now overtaking and replacing our vision, with our partial our assent. So what I’d like to leave you with is the idea that we maybe now face a choice, between on the one hand the desire to shut down our senses to this incomprehensible storm of imagery, to delegate the role of interpretation and judgement, and on the other hand the need, the responsibility, and the burden, to see.

One thought on “The Blind Eye and the Vision Machine

  1. I suspect that you overestimate the powers of the machine. The spooks will certainly spend billions attempting to build various sorts of Big Data Cloud computational things to try to dredge out of the blizzard of pictures whatever sorts of bizarre information suits their demented fantasies of Files On Everyone.

    It won’t work nearly as well as they hope, nearly as well as they will claim.

    The machines will still surprise and dazzle us, but never in the ways we expect. It might still work out badly, but it’s as likely to be funny, beneficial, or neutral, as it is to be dangerous or evil.

Leave a Reply

Your email address will not be published. Required fields are marked *