The Rules of Guess Who? From Daguerrotypes to Algorythms
For those who are young enough to be called ‘millennials’, it was probably a favourite pastime: the Guess Who board game, in which strategic questions and sharp eliminations should leave one face standing on a crowded board. How much more easy would this game be if one of the competitors had digital means at their disposal, like a tool that could compare all characters on the board to the card in the opponent’s hand? The Dutch police now have this nifty option readily available. Since December last year, when a new biometrical search system was launched, officers can complete their own rounds of Guess Who within a few seconds. Photographic collecting and the use of anthropometrics have been part of police practices for about a century and a half, but the scales have now been upped significantly. Nevertheless, there has hardly been any public or political debate over this recent expansion of photographic databases, which in the Netherlands alone already include over 800,000 searchable images. We need only go back to the earliest boom in police photography to realise that this lack of debate is deeply concerning and will have future repercussions.
The mugshots of the 19th century
The use of photography for judicial ends is almost as old as photography itself. The first photograph of a person and the world’s primordial selfie were both taken in 1839 – a decade later, the photo camera was turned onto border-hopping criminals in Belgian, Swiss, British and French prisons. In those early days, police authorities did not have their own photographic departments and the commercial photographers they hired made few efforts to distinguish between separate portrait genres. Things began to change in 1890, when the Parisian police clerk Alphonse Bertillon outlined exact standards and classifications for judicial photography in his La Photographie judiciaire – thereby providing the basis for the mugshot as it still exists today. The innovation quickly turned global as authorities all over the world adopted Bertillon’s photographic and biometrical methods. This worldwide adaptation contributed to international exchanges of suspect images and data, enabling the development of standardised databases.
Restraining civil liberties
Just like their digital counterparts today, these nineteenth-century innovations were accepted with little debate. The need for improved means of identification seemed obvious to contemporaries, fearful as they were of international criminals, anarchist terrorists, and the flocks of migrants in which crooks might hide. In the case of Rotterdam in 1895, the minutes of the municipal council show a complete lack of debate concerning police plans to introduce standardised photography and the collection of biometrical data. In fact, the council even flipped out its chequebook and claimed that money would not be an issue. The police reforms quickly resulted in officers copying hotel registers, detectives shadowing suspicious anarchists through the city, and commissars exchanging heaps of photographs with German colleagues.
As the example of Rotterdam shows, the lack of political and societal debate led to almost unbounded implementation of the new police tools. Accordingly, there was little to no consideration of how these techniques might possibly endanger civil liberties. In the Netherlands, laws stipulated that only captives could officially be photographed, but in states with less legally restrained and more repressive police services, such as Imperial Germany, non-criminals were subjected as well. Contemporary German reports show that being suspected of anarchist sympathies was enough reason to be dragged off the street to be photographed by the police. Because these photographs were collected in so-called ‘anarchist albums’, the depicted suspects were instantly framed as potential threats. When visual data collections were exchanged at different levels, a photographed person could become subject to immediate surveillance once entering other jurisdictions. The possibly very deep impact on the lives of innocents was generally neglected: political groups such as socialists and anarchists could become entrapped in arbitrary police registers, resulting in limitations to their freedom of movement and possibilities of keeping or finding employment.
The impact of misidentification on innocent individual has hardly reduced since the late nineteenth century. In fact, in the online, interconnected world of today, misidentification can have more lasting consequences than ever before. It is expected that the amount of unjust identifications will only increase in years to come. The dangers of blindly relying on new police technologies, are symbolised in the poignant history of Steve Talley. Talley, a resident of Denver, Colorado, was arrested in September 2014 for two bank robberies that he did not commit. The basis for his detainment rested on subpar surveillance footage, which was shown for identification to his ex-wife and two former roommates. During the two months that he was incarcerated, Talley could no longer pay his rent and lost his house. Afterwards, he was unable to go back to work due to his criminal record. In December 2015, he was arrested again due to facial-recognition technology – even if he was three inches taller and missed peculiar facial marks of the actual robber. This time it took four months before he was finally dismissed. His case underlined the 2009 conclusions of the US National Academy of Sciences: apart from DNA testing, no other forensic method is sufficiently reliable and consistent – especially if the evidence is based on grainy surveillance images.
Although facial recognition will never be flawless, data is being retrieved from ever-increasing proportions of the population. This often happens without much publicity or possibility of legislative and societal control. In the United States, already half of the country’s adult population is listed in the police’s comparative facial-recognition databases. Still, legislation that checks the police efforts is most often non-existent. Rather than fixing the system’s remaining flaws, an American homeland security agency recently opted for another leap into the digital realm. It signed a contract with the Isreali start-up company Faception, which claims to provide tools that can recognise crimes before they are committed, simply on the basis of facial expressions. The combination of facial-recognition methods with police body cameras looms large. In case of complete population surveillance, the step from Orwellian dystopia to everyday reality may be very small indeed. In the same vein as the nineteenth-century police photography, new techniques that were intended to search for criminals are being applied to entire social groups. The historical example shows that databases tend to become all-encompassing when technologies seem to allow it and legal boundaries are not clearly delineated. Societal debate is crucial when it comes to the implementation of new techniques. It should therefore be about time for lawmakers to limit the board size of their Guess Who sets once again.