Machines of Proliferation
An Argument for the Preservative Potential of Algorithmic Applications
Joost Dofferhoff (10574204)
Dual Master’s: Preservation and Presentation of the Moving Image University of Amsterdam
23rd June 2020
Supervised by: mw. dr. A. (Annet) Dekker Second Reader: dhr B. (Bas) Agterberg
Table of Contents
1) Machines of Proliferation ...1
2) Mediated Memories and the Algorithm ...8
2.1 Analogue ...8
2.2 Digital ...11
2.3 Algorithm ...16
3) The Automatic Algorithm as Preservation Strategy ...23
3.1 Reinterpretation and variability ...23
3.2 Preservation through Interactive Generation ...29
4) From Transmission to Preservation ...36
4.1 The Dawn of the Algorithm ...36
5) Conclusion ...40
6) Bibliography ...47
1
1) Machines of Proliferation
Every now and then my phone pops up with the following message: “Clear up 200 memes from your phone. Low resolution media and memes found”. When looking through the images that the Google Photos algorithm has chosen to delete, I find that it has done an admirable job of deciding what a meme is and what is not. However, upon closer inspection of this collection my phone has also chosen to delete pictures that are not memes or low resolution at all. Even though these images might be considered ‘unnecessary’ or ‘disposable’ it shows that the algorithm does not work perfectly. However, the interesting part of all of this is that smartphones and photo gallery apps decide some material deserves deletion in the first place. Who or what makes the decision that memes and funny pictures are worth destroying but that pictures of the burned down Notre Dame from my Paris trip are worth keeping?
Deletion is not the only thing that the Google Photos algorithmic application proposes I can do with the pictures in my gallery. On some occasions, usually after my phone realizes I have been away from the location I have designated as home, my phone prompts a
notification saying it has created a compilation or that it has created a new automatically generated creation (see Figure 1). The compilations are usually geographically based. They can be in video or slideshow format, made from a selection of pictures of the four day Paris trip or just one or two pictures from a short afternoon in the Dutch dunes. Geographic
location, however, is not the only thing that the smartphone chooses to make compilations off; they are also people, animals, cars and objects such as posters, bottles of beer and skyscrapers.
Interestingly these categorizations are often not categorized as what the picture was intended to be. Especially in the category of ‘cars’, the subject of the compilation seems to be a by-product of the picture. Often located in the background or in the corners or sides of the
photograph, it seems that the algorithm is particularly good in picking out cars, wherever they may be. This makes that your memory can be recalled differently than you would have imagined recalling it when you took that picture.
The ability of algorithms to make sure that content gets deleted and that content can be created in order for it to proliferate will be the subject of this thesis. Before explaining further what this thesis wants to achieve it needs to be made clear what I mean by algorithm. One famous definition of an algorithm comes from Yiannis Moschovakis who tried to define what an algorithm is exactly in his aptly named paper What Is an Algorithm? (Moschovakis, 2001). For Moschovakis the algorithm hinges on the concept of recursiveness, or as it is known in
2 computer science as the “method of solving a problem where the solution depends on
solutions to smaller instances of the same problem” (Graham, Knuth & Patashnik, 1994). Moschovakis ends his efforts on defining algorithms with the followings slogan: “The theory of algorithms is the theory of recursive equations” (Moschovakis, 2001). This means if one generalizes this statement for the purposes of this thesis is that algorithms are tools meant for repetitive uses that can be called upon or acted upon automatically. Algorithms therefore can be seen at iterative recursive tools, meaning something that is repetitive in order to generate the same or a similar outcome. However, as Yuri Gurevich writes, defining algorithms is not as simple as this. He writes that recursion is only one aspect of algorithms. Gurevich explains that the recursive algorithm theory does not include linear-time algorithms or distributed algorithms that do not hinge on the recursive approach that Moschovakis speaks off (Gurevich, 2012). With this in mind the definition taken for algorithms, or the algorithmic applications that I speak off in this thesis, will be that of iterative recursive tools. The algorithms that are mentioned in this thesis are algorithms that perform the same task on different objects and different content that produce a similar output on all of the content it performs the task on.
Algorithmic applications on smartphones and other ‘smart’-devices are a common sight these days. The examples above are taken from the Google Photos app and there is a similar application on Apple Iphone’s with iOS 8 or higher called Apple Memories. Older than these are Facebook’s automatically generated collections which celebrate friendships and relationships by using photos people have posted on Facebook together (McLennan, 2018). One resemblance between all those algorithmic applications that create collections or edited content for the user is that they do this without your active input. These recursive algorithms are called upon by the programming in someone’s smartphone. Even though the option to disable this feature is there, the default setting on a new smartphone is that this function is activated. This is a clear distinction between the ways people used to explore their memories. José van Dijck and Tara McLennan (Van Dijck, 2007, McLennan, 2018) both reminisce about family photographs found in shoeboxes and in old catalogues in their own or their parents’ attic. The way in which these memories present themselves to people is by chance, during the clean-up of your attic for example, or by going to purposefully looking for them because you want to look at them or show them to someone else. This is different to an algorithm that sends you a push notification about a new slideshow that it has created. It is as if you would
3 send a robot to your attic to pick out the nicest photographs in a common theme that it can find.
The problem or noticeable change that this creates is that the algorithms get to decide which pictures, videos or other content that will survive. Where in analogue times the content that survived was usually handpicked by the people who took the pictures, in these
algorithmic times our smart-machines do this work for us based on work of the programmer who created it and the machine-learning that the machine has done since. One can therefore ask the question: what proliferates in times of algorithmic curation? Contrary to this, one can then also ask the question of what does not proliferate. These types of questions lead into the preservation aspects of personal memories and other content such as (digital) art that I will address in this thesis. I will argue in this thesis for the progressive potential of the algorithmic application, such as the ones I have already mentioned. However, before making this
argument I want to go into the conservative nature of nostalgia, memories and the
conservative nature of archiving and preservation. The main research question of this thesis will therefore be “How have automated algorithmic collection applications changed the way in which people interact with personal memories? This question will be approached in chapter two through the history of analogue, digital and algorithmic memory making. The question that will be answered in chapter three will be: “How can automatic algorithmic applications be used, by people and institutions, for the preservation of personal memories and artworks”? This will be approached through analysis of the ‘alternative’ preservation methods. Chapter four of this thesis will take the preservative potential of algorithmic applications further and will make the argument for the proliferative potential of algorithmic applications.
In the chapter that follows this introduction I will talk about the difference in
analogue, digital and algorithmic memory making, especially regarding the family album and the individual collections. My argument there will be that nostalgia, memory making and reminiscing all have a conservative nature. Not just in the fact that people are conserving, or preserving items, but also that the collections that people make halt progressive progress (Grainge, 1998). The family album is a constant reminder of what used to be, and is used as a utopian point of time where people often fondly look back upon (McLennan, 2018). People often credit nostalgia to things that are old, such as analogue photography, but in the chapter I will argue that digital memory making, the act of actively deciding what gets to be memories and what does not, still builds upon this conservative way of memory making and that
4 algorithms have not changed this either and that they might have even enlarged the nostalgia boom that is going on.
In the chapter two, however, I will make a 180 degree turn and argue for the
progressive potential of algorithms for preservation purposes. Where analogue media and the objects and content that come from their use, can be kept for a long time through storage, for digital media storage is often the shortest term approach (Ippolito & Rinehart, 2014). By looking at different ‘alternative’ methods of digital media archiving, using Martine Neddam’s concept of ‘generative preservation’ and Annet Dekker’s writing on the concept of
‘proliferative preservation’ in particular, I will argue that people and institutions can use algorithmic applications such as Facebook’s memories and Google Photos to distribute their creations over the web and ‘preserve’ them through having their creations multiplied in the hundreds or thousands and thereby giving them a large chance of survival by dispersing them to as many places as possible.
What adds to dispersal and proliferation of digital content is that these algorithms usually all give people the opportunity, with only a few clicks, to post their newly generated creations on social media platforms. This, of course, is not that different from analogue photographs, because people also use those to share their photos and the stories that connect to these photographs with other people, be it on social media or face-to face. However, it is interesting that these generated creations are made to be shared; in all the application the first option to click or to press is not to save the pictures to your own device but to share it to your friends or your social media of choice. This difference that Mette Sandbye, a professor of photography studies who writes on the connection between memory and photography, also recognizes is cause for a “revise to Barthes’, Sandbye Claims. The statement that every photograph posits “this-has-been”, the connection a photograph has to the time and space in which it was taken, which Barthes saw as the essence of every photograph (Barthes, 1980). Mette Sandbye’s revision suggests that the live stream of images is a constant replenishment of “this-is-now” (McLennan, 2018). This applies to the content that these applications generate, because they are created quite quickly after one has taken a photograph. Often it takes only a few minutes after someone has taken a picture that the smartphone will display a notification saying that the algorithmic applications has made ‘a new automatic generated creation’ for the user. This, to me, indicates that these pictures are meant to be posted on social media within the same day. Therefore, these pictures no longer display something that someone was doing but it is displaying something that this person is doing. Mette Sandbye
5 shows that these new digital practices, which utter the “this-is-now” mentality, fulfil “the user’s existential desire to “feel time” (Sandbye, 2012). These arguments have often been since the arrival of social media and especially the web 2.0. People no longer wait for the evening news to known what is happening in the world but take to Twitter to see what is happening right now. In recent years this trend of ‘live-blogging’ has also been taken over by mainstream news outlets, usually the websites of newspaper companies or online-only news sites, to also gather information as quickly as possible during an important event or
impromptu serious situation (Thurman & Walters, 2013; Matheson & Wahl-Jorgensen, 2020).This phenomenon is also reflected in personal memories and especially those that are made and shared on social media. This idea is strengthened by the fact that Instagram used to only allow people to post pictures on their story that were no more than 24 hours old (Loren, 2017). Even though they changed this and gave people to option to post older pictures, now when someone posts a picture that is older than 24 hours old it automatically puts the date of the picture in the story (Garun, 2017). Even though this can be easily be avoided, for example by taking a screenshot of a photo in your gallery which causes the phone to register this screenshot with the date on which one takes the screenshot, people often keep this date in their pictures as another form of a ‘liveblog’ or diary so that, again as Sandbye claimed, people can ‘feel the time’ in someone’s social media feed. This new way of sharing and creating memories is not something that is archived or preserved, except for people who are in important government positions. However, as I will argue in this thesis the sole existence of algorithmic applications that generate these creations for people and allow them to share them to social media can be seen as a preservation method for the proliferation of this material.
A possible reason that algorithmic applications have increased in numbers and popularity might be because smartphones are part of a phenomenon of ubiquitous and abundant storage. If one takes multiple pictures per day and even move their images over from their old phone to their new one, this will eventually lead to a smartphone filled with thousands even tens of thousands of pictures. This phenomenon, which Giselle Beiguelman calls the ‘age of documentary overload’, is one where there seems to be “memory fetish” that causes that companies try to seduce people with products that keep having more and more memory to store things (Beiguelman, 2015). The phenomenon of pocket-sized smart devices that have larger and larger storage capacities has created another phenomenon which I call the expiration date of these pictures. Because of this ubiquitous amount of storage there is a small chance that someone will have the time or the will to look through all these photos in order to
6 use them. However, due to the fact that there is such a large space of storage older photos will get buried under newer photos. In practice people even need to ‘dig’ or scroll down very far through their phone to be able to find older pictures, given there is no easy way to search for them. Studies even show that this amount of pictures taken for memory purposes, and the fact that those memories are entrusted to the cloud, causes changes in how people perceive
memories. Because people rely on pictures to tell them their memories, there is a chance that people are creating false memories from these pictures, in the sense that they build their memories from the pictures instead of their experience (Mazzoni, 2019). Another problem is that people do not relive their memories often enough, because people trust on their memories that they have in storage, which causes people to forget certain events quicker than people used to in the past (Mazzoni, 2019).
However, what algorithms do is analyse all pictures in the phone and place them in aforementioned collections. Through this people might see pictures again that they haven’t seen for a number of years. If one visits Paris in 2017 and 2020 for example and takes pictures of the Eiffel Tower in both instances on the same phone, the algorithm will put these pictures together because it recognises the building in the photo. The pictures taken in the year 2020 will then cause the pictures from 2017 to resurface due to the algorithmic application that the pictures are uploaded through.
This can be linked to a phenomenon called post-value recall. This phenomenon is described as “the true recall value of information is not completely known until after the information is archived” (Czerwinsky et al., 2005). In their article Czerwinsky et al., describe it using an example of people hoarding filing cabinets full of papers just because one cannot predict which papers they may actually need later. I argue that this is similar to the hoarding of pictures on smartphones because people might keep them because they do not know their value either. In a 2018 study by Vitale, Janzen & McGrenere it was shown that there is a large number of people who hoard everything on their phones what they can, even if they
considered it to have little value. Some of the reasons given for this was in line with post-value recall mentality because they considered that some content on their phones might be worth something, or be useful, in the future (Vitale, Janzen & McGrenere, 2018). Whereas post-value recall speaks of complete knowledge of information, once the information in question has been archived, with modern day uses of photographs this may need to be
changed. In current day photography it is not the aspect of information but the aspect of value that needs completion. Value in this sense can mean a number of things. There can be value of
7 memory attached to a photograph, which can later turn into nostalgia, there can be social value to a photograph that can increase a person’s social and human capital and lastly there can be financial value to a photograph if someone earns money by posting to social media. Algorithmic applications that dig up and let resurface pictures from a long time ago can, by putting them in collections or in the form of memories, reimburse an old photograph with new nostalgic, social or financial value for the user of the application.
8
2) Mediated Memories and the Algorithm
To start talking about how algorithms have changed the way in which interaction with mediated personal memories works, this literature review will first go into a short history about how mediated personal memories and memory making has changed. This chapter will look at the analogue and the digital history and will argue for the conservative nature, as mentioned in the introduction, of the family album and everything related to those kinds of personal memories. This argument will address the ideological and political form of conservativism and the ‘conserving’ or preservative nature of the word. Not just technical differences will be stated but literature will be referenced to signal how scholarly thinking about mediated memories has changed through the years. This is especially interesting when looking at the moment in time when the shift from analogue to digital took place, because that is the moment where the difference between analogue and digital are less clear than they are today.
2.1) Analogue
Personal memories have a long history of items that have been preserved. From family heirlooms, trinkets and other materials or stories and tales about certain people (Ippolito, 2016). This is important because it has shown that sometimes these personal memories and artefacts, as opposed to things that are archived on a national or institutional way, are the only thing to survive from a certain point in history. Jon Ippolito gives the example of the story of the mapinguary, an animal described by the native Machiguenga people of Peru, whose existence is only known today through countless retellings of the stories about this animal throughout history. The stories have proven to contain more details and specific details about this animal than modern archaeology and palaeontology have been able to uncover (Ippolito, 2016). It is therefore that personal memories and collective memories that are generally not included in national institutional archives are taken into account when speaking about preservation because they can in the long run prove to be very important for making history.
To focus this chapter and not analyse this history too broadly, I will mainly focus on mediated memories such as photography, film and other memory making visual media because these have more directly, and in a way that is more traceable, influenced the way in which the algorithmic applications operate today. I do this because it is helpful to look at the ways visual memory making has changed and how the objects and storage media on which
9 these memories are stored have changed. Photography already signalled a change, in the way memories were made, from times before where people would get portraits painted. Painted portraits were expensive and therefore were mainly made by the economic elite (McLennan, 2018). Even though photography portraits were also expensive at the time of the mediums conception, photography quickly became more and more available to people who had less money to spend. This caused for a new form of memory making to come into existence namely the family album (Reading, 2008). Pictures from holidays, family events and gatherings would be collected in a book to keep them from decaying too quickly and to reminisce about or show to others. The family album was a change from portraits because the use of the portrayals changed from denoting stature to the use of nostalgia and memory, or as Beiguelman states: “Until the end of the 18th century, talking about the aesthetics of memory means talking about the strategies that were used to perpetuate through the arts the image of some individual for posterity” (Beiguelman, 2018). This perpetuation of posterity hinged upon the fact that these portrayals were unique and therefore valuable which signalled a person’s wealth, this changed with photography because of the fact that one could make almost endless copies of a photograph. Joanne Garde-Hansen, in quoting Annette Kuhn, “argues that family photographs are simultaneously the same as the millions of other baby and child photos that exist across the world in millions of albums, and, due to the personal
resonances of each one, very different” (Garde-Hansen, 2009). Kuhn herself writes that “on the surface the family photograph functions primarily as record (…) visible evidence that this family exists”. The family photo, she says, “looks toward a future time when things will be different, anticipating a need to remember what will soon be past” (Kuhn, 2000). Even though one can argue that painted portraits were also not unique and that were also made for memory purposes, these portraits were not as readily available as photographs came to be. One thing that did stay the same was the utopian function of the portrait by which I mean that portraits were made to make someone look the best, the wealthiest or give someone the most stature and impressiveness on this portrait. This can also be found in the family album, only with different aspect than what a painted portrait wanted to convey. A curation of family photos in a photo album creates an image of the perfect family. While the perfectly sharp and happy photos are kept inside the family albums, the ones that were not selected for putting into the book are often kept in dusty shoeboxes under the bed (Garde-Hansen, 2009). This
perpetuation of the perfect utopian family though the curation of family photos in the family album is the first conservative aspect of the family album, which I explain in the coming paragraph.
10 Looking at this concept of the family album as a utopian view of the family, one can analyse what stays in the family album and what will move towards the dusty shoeboxes. This has to do with Freudian concept of the death drive which states that every person has a drive for aggression, destruction and self-destruction within them (Derrida, 1996). This concept was later analysed further upon by Jacques Derrida in his famous work Archive Fever. The death drive applies to the family album because of this conservative nature. In this I deviate from the argument of Annette Kuhn where she says that the family photo “looks toward a future time when things will be different, anticipating a need to remember what will soon be past” (Kuhn, 2000). Although I do not wish to argue that things won’t be different, I want to argue that the family album, and the content within, exists as a utopian point in the family’s history, curated to look as perfect as possible, to fall back upon as a family. Future family albums or photographs will be measured against previous ones to keep the image of the family as perfect as possible. This is why the death drive, or the dusty shoeboxes with the not-so-perfect
photos, is interesting because it not only shows the conservative nature of the family album; it also shows the conservative nature of conservation in general. It is of course no coincidence that these words derive from each other. Conservation is not progressive because it holds on the past, often the very distant past, as something to build a current situation towards or to keep a current situation in check. Archives, in particular very traditional state archives, are there to collect data, information and objects about certain situations at different points in time. By collecting this data the archives create a certain moment in time themselves about a certain situation. Archives have the power to shape history by deciding what they do and what they do not accept in their collections. This very traditional archival approach caused more and more alternative archives to jump into existence to fill the gaps that they thought national state operated archives were missing (De Kosnik, 2016). Examples are the International Institute of Social History, which focusses on archiving from the point of view of social injustice and the group in search for social justice. Other archives are ATRIA, the institute archiving the history of the women’s emancipation history, and The Black Archives, which focusses on the social history and neglect of black people in the Netherlands.
Even though these progressive archives exist, the act of archiving itself remains one of a conservative nature. Archives still make choices in selecting what enters their collections and which things do not. This halts progressive thoughts because archives keep promoting things from the past that have been repeated over and over without showing what has been built upon these thoughts. Paul Grainge argues that this type of nostalgia, the type of utopian
11 nostalgia that one sees in the family album and the type of nostalgia that makes its way into archives, stops people from accepting change (Grainge, 1999). However, in all this, also lies a paradox because archives and nostalgia might be called a necessary evil to be able to build upon thoughts from the past. Without conserving past knowledge, one cannot trace how certain changes have happened through society. The question now is how this conservative nature of personal memories carries on when photography and memory making changed from analogue to digital.
2.2) Digital
Digital memory making can be said to have begun with the invention of digital video cameras. Even though digital tools already existed, none of those tools had the commercial impact that digital video cameras and digital photo cameras had in daily life. In the first couple of years of video equipment being available it was still very expensive to buy and therefore not aimed at individual amateur use. However, as the technology improved the equipment developed into household equipment that was affordable and usable for almost everyone, or at least more affordable than the 8mm and 16mm home cameras that came before them. These early digital video cameras still stored the videos and photos on a digital tape which could then be played on television sets. This type of storage eventually ran out of use when hard disks and later flash drives were introduced. Not only were these types of storage smaller in size, which meant the camera itself could become smaller in size, the technological advances of the time also meant that these disks and drives could store more and more. Even though the digital tape, like the DV (digital video) and MiniDV tapes, had more storage in the late nineties than the first flash drives, this quickly changed due to technological advances. The SD cards, or Secure Digitals cards, that were introduced in the late nineties and are still being used today in video and photo cameras, quickly took on more and more storage. Where they started in 1999 with memory cards that only had 32 megabytes of storage, in 2006 this already changed to 32 gigabytes. Nowadays the microSD cards, which are even smaller than their predecessors, can house up to one terabyte in storage (Wikipedia, 2020). For comparison the (Mini) DV tapes had a storage of about 14 gigabytes, which is only 1, 4% of current day storage.
The reason why these numbers are so incredibly important to state is because the most important change from digital to analogue is the fact that people who take pictures can almost
12 instantaneously delete the photos they are not happy with and have them disappear into
nowhere with the click of one button. If one would go out with an analogue camera with one photo roll in the camera, one could shoot for example, 24 or 36 pictures. If someone, today, would go out with a 4K camera with a one terabyte of storage, one has the possibility of taking two hundred thousand pictures (assuming every picture has a 5 megabyte file size). If somebody has a smartphone with a storage card that houses one terabyte, which to be fair not many people will have, this number will be even greater as pictures taken with phones will be smaller in size. What this shift in storage capacity has done for memory making, the act of creating content made for remembering, is the fact that one can repeat a particular photo, or try to get it better, over and over again. There are almost no restrictions anymore to how many photos one can take. Another difference with analogue photography is the fact that digital photos and videos often have an instant replay function. No more waiting around for your negatives to develop but someone can immediately decide if they are happy with the result or not and decide to try again. Where analogue photographs and reels of film end up in the dusty shoeboxes instead of in the family photo album, these digital photos that get deleted are basically gone forever. People can of course run hard disk recovery programs, but usually when people delete photos they do it with intent and not accidentally. To strengthen this argument Anna Reading’s 2008 research showed that people didn’t even view their digital collections, which Reading likened to the family album or a modern form thereof, they had amassed on their cameras and phones as ‘worth keeping’ (Reading, 2008). This shows a change in perceived value of a photograph between analogue and digital. Where discarded analogue photographs were still worth keeping, Readings research shows that digital photos that are not deleted are not even worth keeping.
An important not to place to this statement is the fact of physical materiality of analogue photos against the digital materiality of digital content. Because analogue photographs can be touched and because they can physically be put into a photo album. Because of this physical materiality people might be less inclined to discard these
photographs. However, an interesting way to look at this is go back to the point of ubiquity. If one could take as many analogue photographs as they currently can take digitally with their smartphones, would people still be less likely to throw away their photographs? I would argue that if there was such a ubiquity in analogue photographs people would be more inclined to decide to throw their photographs away because of the amount of space these photograph would need to store them. Secondly, I would also argue that if someone could take an almost
13 infinite amount of analogue photographs, the value that someone then puts on a single
photograph would also decrease. This would then be because someone could, in such a situation, just make that particular photograph again just as is currently the case with digital photography. In such a situation there is no limit anymore so people do not have to be careful anymore with what they decide to photograph in case their photo roll became full. This is important to the way in which our interactions with personal memories have changed, because the current situation makes it that people can perfect the photographs they are taking on the spot and not have to wait for them to be developed. A certain aspect of surprise is therefore missing from digital photography that was there in analogue photography.
An interesting way to look at this is through the theory of Walter Benjamin on the ‘aura’ of mechanically reproduced objects and medium.Even though Benjamin said that photos, in opposition to paintings or other non-mechanically reproducible objects, did not have any ‘aura’ anymore because they were mechanically reproduced (Benjamin, 1936), one can say that digital photos would have even less ‘aura’ than their analogue counterparts because they are even easier to (mechanically) reproduce. Because of this, I would argue, photographs that are made using analogue camera and reproduction techniques have
(re)gained a certain ‘aura’, which in contemporary culture is usually regarded to as ‘authentic’ qualities. As a newer medium becomes the norm and takes with it parts of medium before it, through the process of remediation (Bolter & Grusin, 2000), there is medium specific part of the older medium that stays behind. This medium specific part of an older medium has the chance to become something rare or ‘one-of-a-kind’. This rarity is what can give something ‘auratic’ qualities, which can come after the medium has been obsolete for a few years. An example of this is the scratch sounds that one hears on vinyl records. When Compact Discs (CD’s) were introduced they took the rotating disc aspect from vinyl records. What they did not take was the scratching sound that vinyl records could produce if they became damaged. At the time this scratching was seen as a negative quality of vinyl records that CD’s could help solving. However, years later when the vinyl record regained popularity, this sound of scratching was seen as one of the emblematic qualities of vinyl records because it showed the materiality and gave the music that came from it a warmer and more ‘real’ quality as opposed to the digital music players that had become the norm. Another interesting example that is often reference are the scratch marks on old damaged film material that was first seen as negative and is now seen as an emblematic quality of old film and often replicated digitally to create something that has the same atmosphere or ‘auratic’ qualities as film from early
14 cinema. A famous quote by musician Brian Eno summarizes it well: “Whatever you now find weird, ugly, uncomfortable and nasty about a new medium will surely become its signature” (Eno, 1996)
This is interesting because Benjamin rejected this notion when he wrote “From a photographic negative, for example, one can make any number of prints; to ask for the
‘authentic’ prints makes no sense” (Benjamin, 1936). Especially in relation to the archive one could argue that photographs and other analogue formats have been gaining ‘aura’ and
‘authenticity’ because of the value that is placed upon them by being institutionalized and preserved. Even though the concept of ‘authenticity’ has a different meaning in the archival sense, one cannot deny that items that institutionalized and archived get a certain ‘aura’. These institutionalizing of the object might strengthen an objects already existing ‘aura’ or makes it so that an object (re)gains it. One aspect of this that can be analysed is the “presence in time and space” which Benjamin speaks off. Archives can be said to place this element, which Benjamin says reproductions lack, onto photographs or other reproductions. This is the process of authentication. Archives give objects ‘aura’ because of their limited accessibility, one might need to travel to one particular archive to view a single photograph, and the
language surrounding such photographs and films such as them being ‘lost’, ‘discovered’ and ‘rare’ might add this ‘auratic’ effect on these items.
How then does this work with digital photographs and other ‘objects’ which are not housed in a specific building (even though they are still ‘housed’ in a server room or on a hard disk somewhere) but are ‘housed’ on websites or other digital-only spaces? More interesting even when one looks at digital spaces belonging to traditional archival institutions. Even though analogue photographs can be copied an infinite number of times, one still needs to track down the original negative which might entail moving to another physical space. In the digital realm one can copy an entire work of art, such as a photograph, with the simple click of the right mouse button (assuming the website in question has not disabled that option). Does the time it takes to create a copy of something weigh in on questions of authenticity? Copies of famous paintings that are copied by hand are far more expensive than copies that are made digitally or mechanically. Benjamin speaks of time and space, meaning the moment and the location that an artwork belongs to, but might time also mean the span of time that it took to create something? I would argue that this also adds to ‘auratic’ qualities of objects and because of this I would say that copying a digital file does not get ‘auratic’ qualities through
15 its copy process because it can be done almost instantly. However, I am not arguing that digital photographs or other digital content cannot get these ‘auratic’ qualities at all.
If one would follow Benjamin’s train of thought completely digital photos and media art do not have an ‘aura’, however, I wish to go against this. Going in on the ‘auratic’ qualities of digital content, Ippolito and Rinehart, in their book on new media preservation, use the terms ‘presence’ and ‘presentness’ from Michael Fried to decide how digital art presents itself (Ippolito & Rinehart, 2014). Fried used this term to defend late modernist paintings and used the term ‘presentness’ to describe an artwork that is “at all times wholly manifest” (Fried, 1998). Ippolito and Rinehart explain that this concept means “that a state of grace comes from this type of suspension of time where the past is hidden, the future is unknown and there is only an eternal present” (Ippolito & Rinehart, 2014). One could imagine this with painting that hang in museum that still use the ‘white cube’ idea which takes all the context that surrounds an artwork away and presents an artwork just as it is. If, in this way, the artwork makes it so that a person feels that “it is wholly manifest” then an artwork could be said to have ‘presentness’. Opposite this is the term ‘presence’ which Fried describes as ‘theatrical’ (Fried, 1998). This kind of theatricality means that “an artwork is not manifest in an eternal moment but rather unfolds, bit by bit, in real, mundane time” (Ippolito & Rinehart, 2014). This unfolding can mean that people have to walk around to artwork to see it completely, or interact with it to see its manifestation.
Even though Fried invented ‘presence’ and ‘presentness’ to argue for his love for modernist artwork and painting which signal ‘presentness’, nowadays we can try to take these concepts out of their hierarchy to see how artworks that are described with ‘presence’ can be seen as equally valid manifestations. Benjamin wrote that even when an artwork does not contain an ‘aura’ it can still be relevant. He wrote: “for the first time in world history,
mechanical reproduction emancipates the work of art from it parasitical dependence on ritual. To an even greater degree the work of art reproduced becomes to work of art designed for reproducibility” (Benjamin, 1936). Ippolito & Rinehart finish their thoughts on these concepts by saying that these old assumptions “that all artworks should retain an aura and the
modernist ideal that they exhibit presentness is based on notion of purity and fixity that no longer apply” (Ippolito & Rinehart, 2014). Even though Ippolito and Rinehart argue against ‘aura’ and ‘presentness’ in digital art, I would argue that these concepts do still seep through in currents debates on digital art. A good example is the way in which digital art and artists are written about. Artists like JODI and Nam June Paik are often described as ‘pioneers’ of
16 digital art. This signals that these artists starting working with digital material, or that they were one of the first ones to get recognition for doing so, in the ‘pioneering’ days of the art form, which fits into the cliché of the phenomenon that every that has to with digital media has to be ‘new’ and the artist has to be the ‘first’ in something (Chun, 2008). In doing so, people want to put a certain sticker of authenticity, in both time and space because often years and places are attached to their title of pioneer, on an art form that wanted nothing to do with authenticity in the first place. This is an example of the concept of recuperation, formulated by members of the Situationists International, which explains the process by which radical ideas, such as denouncing the need for authenticity, are co-opted into mainstream society and neutralized (Debord, 2003). In this case, the radical ideas of denouncing the need for
authenticity are seen as just moments in time for certain art form or certain artists. Through this, one can become a pioneer in a movement that stands against the idea of what a pioneer stands for. This was of forcibly making something ‘authentic’ is done in institutions often for financial gain because the marketing teams of such institutions know that words such as ‘pioneer’ and ‘visionary’ are easy buzzwords to attract an audience (Chun, 2008).
Now that I have laid bare the underlying notions of the characteristics are of analogue, digital memory making are, I will begin to speculate about how the way we see and interact with digital memories and digital artworks changes when algorithms get ‘into the picture’. What are the changes that people go through, in regards to them experiencing their personal memories, when it is no longer them that curate their pictures for them but it is an algorithmic application that decides what to edit and what should be shared with the world?
2.3) Algorithms
Just as with digital photography, algorithms that curate one’s personal photos came out of technological advances. However, these advances caused our interaction with storage to change which in turn caused that these algorithms were becoming necessary (Rainie & Anderson, 2017). Because people are constantly inviting and inventing new technologies into their lives it means that even newer and, as people will say, ‘better’ technologies are
necessary to deal with these changes (Chun, 2008). This is the case with the kind of
algorithmic application this thesis speaks off. This is because our increasing ‘memory fetish’, that I have mentioned in the introduction, causes us to have more and more storage which in turn can cause people to be overloaded with the possibility to store millions or even billions of
17 pictures (Beiguilman, 2018). Not surprisingly in a world dealing with an obsession of storage is that data has become one of the most valuable things on the planet (Garde-Hansen, 2014). This has caused a societal ‘fear of information loss’. This is why we, as people living in this age, need tools for “keeping track, recording, retrieving, stockpiling, archiving, backing-up and saving” to defer us from having to deal with this fear (Garde-Hansen, 2014). This fear exists because information has become one of the most valuable things in our current society (Brown, 2019). It is no coincidence that the current societal status is sometimes described as the ‘information age’ because information is what brings in the most money for some of the biggest companies of today. While companies such as Google are under no obligation to announce how much each and everyone’s data is worth, the company does make a daily profit of 92 million dollars (Brown, 2019). While Google also does not show how much of this profit comes from their user’s data, it is estimated that a single account can make up to 15 dollars for the company every month. These notions of the value and worth of personal information has seeped through into society and created a situation where information is regarded one of the most valuable things in society (Garde-Hansen, 2014)
This is where algorithmic applications step in and promise to deliver these tools to us, usually in exchange for our data. This causes that social media practices are becoming a proxy self-archiving platforms in the sense that people use their social media pages to create an archive about themselves (Garde-Hansen, 2009). Here one can also see that different social media platforms serve different ‘archival functions’. Facebook can be seen as the ‘Friends and Family archive’, Instagram as the ‘individual identity archive’ and LinkedIn as the ‘individual professional archive’ and so on. The fact that these different social media platforms serve different functions, such as the social connection functions in the case of Facebook and Instagram and the professional job related connecting that happens on LinkedIn, is not something that is clearly visible within these platforms but come to exist through cultural frameworks of collective memory (Garde-Hansen, 2009). This can be seen in Halbwachs definition of collective memory who says that when a group “considers its own past, the group feels strongly that is has remained the same and become conscious of its identity through time” (Halbwach, 2011). The group in the case of this thesis are the users of a
particular social media. This group also consists of multiple groups because there are different subgenres of functions within the major functions on any social media which carry their own specific uses of the particular platform. Garde-Hansen writes about this “while historians focus on changes, ruptures, discontinuities, challenges and differences (…) collective memory
18 produced life essentially unaltered”(Garde-Hansen, 2014). Here one can see a resemblance with the analogue family album mentioned before in which I mentioned that families create a ‘utopian image of the family’ in order to have something to look back at and use to know what family image to keep pursuing and, more importantly, keep creating.
It can be said that algorithms applications perpetuate this conservative, essentially unaltered, way of creating memories as well through their working of ‘iterative recursivity’ (Moschovakis, 2001). Algorithmic applications are invented to constantly and continuously do their task, which in the case of Google Photos can be editing recently taken pictures or put certain pictures together based on time and geological location, without fail or change. An algorithm is basically a script that can be activated on every object or picture it is given and then interact with these objects the same way for every new object, almost indefinitely (Moschovakis, 2001).
One can see this kind of behaviour best in the algorithms that decide the
advertisements we see on the side of websites or that we see scrolling through our social media feeds. By analysing, through machine learning, the sites and pages someone interacts with these algorithms learn what we are interested in, in order to target us with advertisements that the algorithms have learned will best suit our needs. The algorithm therefore has the potential to predict what a consumer might want, by presenting you with the products, websites and other items that you have possibly seen before or something that will fulfil a need that someone is possibly looking for.
Algorithmic applications such as Apple Memories, Google Photos and some functions of Facebook do the same thing. Take Facebook’s algorithm for example that makes video’s using your pictures or videos based on anniversaries with friends or to make a slideshow of all the photos someone has taken in the last year, or at least the posts that were most interacted with. What these algorithms do is present the user with photos they had already uploaded before and give them the opportunity to post them again in a different narrative, namely that of anniversary or reminiscence. McLennan adopts Henri Bergson’s notion of duration to say that:
“These photographic assemblages pierce the mediated present with the past, exposing the beholder to time as the co-existence of what has been, with what is. As such, these algorithmic arrangements illuminate the ways in which photography has historically grappled with a paradox: a departed moment is experienced with present immediacy,
19 as though the captured instant were both alive and dead in the one shot” (McLennan, 2018)
This, again, goes back to the conservative nature of these albums, slideshows or
‘photographic assemblages’, as McLennan calls them, because they show what someone was doing or experiencing earlier in time and shows what someone is doing or experiencing at the current time in relation to the original pictures that are used in the algorithmically generated photographic assemblages. The algorithms learn from our behaviour to, which can for example be the interaction people have with those generated assemblages, and from every new instance create things for users that fall in to the same line as the assemblages that the user has shown interest in. Especially on social media platforms such as Instagram the application recognizes which effects, filter or GIF (Graphics Interchange Format, usually a moving animation) the user has used before in order to recommended this first thing when the user is creating a new post. Arguments for this have been made because once the Instagram algorithms notices that a certain post format works and gets reach, the algorithm will
recommended the same or similar format options to the user again (McCarthy, 2019). This is because when a post gets reach Instagram gets money. For the company it is therefore useful to see what generates reach for them and what does not, so that they can push certain formats forward while burying those which do not make them money further down in people’s social media feeds.
The last part of McLennan’s quote, about the “captured instant being both alive and dead” also goes back to Barthes’ writing on photography. He wrote that this capturing of life and death was the crux of photography, it captures life (or liveness) by killing it making “every photograph a catastrophe” (Barthes, 1981). McLennan continues on the argument to state that “This duality of absence and presence is the poignancy of the medium, which in its stillness appears to keep departed moments alive as unchanged images, and yet serves as a reminder that time has continued to enfold everything in time as becoming” (McLennan, 2018). Miriam Ghani, reiterating the words Jacques Derrida on Freud’s death drive, also utters that “to archive something, it must be fixed in time, like a butterfly pinned in a glass case, and thus to archive is also kill the very thing you feared to lose” (Ghani, p 13, 2013).
This train of thought, to kill in order to save, is used often in archival studies (Derrida, 1996; Ghani, 2013). However, it is mostly used to denote objects that are archived in storage or objects that are being exhibited in museums. Barthes states that every photograph shows “this-has-been” and thus shows what was, it shows the liveness that is no longer there (or in
20 Barthes’ writing that was never there in the first place). As mentioned before Mette Sandbye already revised this “this-has-been” statement to suggest that todays “live stream of images” turns this statement into “this-is-now”, meaning that photography isn’t used solely for
showing what one has done, but what one is doing (Sandbye, 2012). However, what I wish to argue is that algorithm applications that re-use uploaded content again, or present the user with edited version of their pictures like Google Photos does, present another dimension within this phenomenon.
Whereas the “this-is-now” way that the live feeds of social media platforms present a flow of things happening at that moment, the algorithms that rehash old content usually under the pretence of some sort of anniversary or a (recently) passed happening, present a
combination of “this-is-now” and “this-has-been”. It takes old material, such as all the photos a user has taken together with another user in the case of Facebook’s ‘Year in Review’ video. This is different from other algorithmic applications because it does not just give an option to re-post an already existing photo, but it creates new content from already existing content and even does this from different sources and not just one account’s content. What it also creates, as McLennan said, is a departed moment with present immediacy (McLennan, 2018). This is because the algorithmic application tries to reinstate liveness in to the older content, by trying to make the memory happen at the specific moment in time that it offers such generated content to the user. Therefore it presents a contingency with what Sandbye writes, the ‘rehash’ creation that these algorithms offer to us for posting are presenting us with a “this-has-been” which, when it was originally posted was a “this-is-now”, with the opportunity to bring the post back to life, to stick with the Barthesian notion of death and liveness, and turn the “this-has-been” into a new “this-is-now”. One could see this as a form of a ‘zombie medium’ (Hertz & Parrika, 2012) which states a medium, or interaction with a medium in this case, that is artificially being brought back to live in order to regain value from it again. However, just as with a zombie, it is not brought back to live exactly how it used to be but in a newly constructed way. To take a metaphor from another monster the algorithmic
applications can be said to create Frankenstein-content, because it tries to instil liveness in new content which exist of already existing older content
Even though the concept of ‘zombie media’ was invented to denote non-commercial and artistic (re)uses of old media (Hertz & Parrika, 2012), I would argue that commercial platforms and companies have taken over this notion when they noticed that they could gain monetary value out of reusing old material, because generated content such as the one’s
21 Facebook’s memories algorithms serve to people also generate more data for the company which in turn generates financial gain for them. This went together with a ‘promise’ for users whom can use the algorithms that reuse these materials to share something with just the click of a button and create social and cultural for themselves, as well as monetary value when used by an influencer who earns money through these platforms.
Remixes and reposts of previously used materials causes this co-existence of what has been and what is, as McLennan stated (McLennan, 2018). This is changing our interaction with these memories because it reactivates memories and starts another countdown every time the post in particular is reposted. A personal experience with this is a particular picture from the second year of high school, when I was around 14 years old, when we visited the town of Maastricht. This particular photo keeps getting reposted and shared, by the people in the picture, at the moment when Facebook sees that the pictures has an ‘anniversary’, meaning that the date of when the pictures was posted is the same except a year or multiple years later. This particular picture get shared over and over again often because it reminiscent of early high school days, but it mostly gets shared because of the unfortunate, to put it lightly, fashion senses and hair styles. For these fun and nostalgic reasons this picture has been reposted and (re)shared multiple times, making it difficult to find out when the original post was made, because different instances of the same photograph keep appearing. A few times per year Facebook algorithm’s that presents you with ‘memory anniversaries’ will present this picture to me, it being a different iteration of all the times that picture has been posted again. This makes the interaction with this particular memory complex because it signals multiple
moments in time that can be allocated to this picture. It can signal the original post, signifying the actual time and place that this picture was taken, or in some cases it is the memory of five years beyond that moment when we, as a group, looked back on this picture and discussed it on Facebook. The next reminder of the pictures may be when we came across it three years later again and this will go on and on as long as anybody will repost and share this particular picture again and again.
I believe that this way of algorithms presenting and pushing memories to people has changed the way in which we interact with our memories. Even though this form of
reminiscing about reminiscing isn’t new, even with analogue photographs people could already reminisce over and over again and I do not wish to deny that people already
reminisced about reminiscing before. However, what I think has changed, is that the pushing of memories upon us is something new and especially the fact that these memories are
22 immediately presented as ‘sharable’, as is not just the case with Facebook Memories but with the other applications as well. This shows that memories have gained a monetary and
commercially exploitable value for both platform and user. I would argue that this is another example of the conservative nature of nostalgia, Grainge also argued, and the family album (Grainge, 1998). By presenting users with the same memories, or with different iterations regarding to these memories, in a recursive manner it might stop the user for producing newer and perhaps more progressive content. Even though these algorithmic applications can also generate new content, this content is still made up of old content and presented as a recursion of this content.
However, as I want to argue in the next chapter, I think this way of re-using and especially re-sharing can also be used in a preservative manner in order for the content that these algorithmic application help generate to proliferate and help them to keep existing on the web. Since digital preservation is difficult because of the constant changes in formats, hardware and software objects on the web need to be re-activated every now and then to keep them from disappearing. In the next chapter I will look at different (alternative) ways of digital preservation, often applied and theorized in the field of media art preservation, in order to argue for the preservative potential of automatic algorithmic applications.
23
3) The Automatic Algorithm as Preservation Strategy
This chapter of the thesis will look at different “alternative” preservation strategies that differ from classical ways of archiving which usually focus on the storage of an object in a vault or digitally on a hard disk or other storage device. In contrast to this the preservation strategies that this chapter will speak off and apply to the case of automated algorithms often depend more on the (re)usage and (re)sharing of the thing that one wants to preserve. The methods that will be taken as examples is this chapter will be the concept of reinterpretation and variability as described by Ippolito and Rinehart and the concept of generative preservation as talked about by Martine Neddam, because they show different ways in which algorithmic applications can be useful for preservation. I will start by discussing the concepts by Ippolito and Rinehart because these represent the most “classical” alternative methods, meaning that these are the ones that other alternative methods often go against.
3.1) Reinterpretation and Variability
The reason new media or digital media need different preservation methods than the more traditional forms of media such as photography or film is because these digital media operate completely differently than analogue media. When analogue media, such as film or
photography, is printed it is always ‘readable’ until it fades, disintegrates or gets destroyed, however while it still perfectly visible and usable one could use it for a very long time. Especially if a photograph is printed on paper it can remain in good condition for a long time if properly stored. For film this is a little more difficult since there must still be playback material, such as a projector or viewing table, for it to be viewable or readable. This is why, for the types of media that are printed on paper such as photographs or paintings on canvas, storage is the longest-term strategy since it is an outside influence, such as “sunlight, air, water (…) that are most damaging to the stuff we store, be it parchment or pigment” (Sterling, 2003). However, for digital media, storage is the shortage-term solution, as Ippolito &
Rinehart argue. “Equipment left in a crate eventually becomes unusable as voltage standards change, cathode ray tubes blow, and floppy disk drives disappear” (Ippolito & Rinehart, 2014). From the time that printed analogue media leaves the printers or the devices that produce the object, this object is no longer connected to the camera or other device that created the object in the first place. Therefore, even if the technology that creates a
24 though it too has a limited shelf life and can be destroyed it is no longer depended on the machine that created it. For analogue media such as film and video and also digital media, in this case, it is completely the opposite. Even though a digital file can still be completely fine, there might no longer be a machine or software program readily available to show it or to reproduce it. In most cases these are the machines or the software on which the digital files were created in the first place. Another crucial aspect of the preservation these types of media is that they need to be used in order from them to still work. When not used, the hardware degrades, just as a car or a bike can become rusty or brittle when the moving parts have been stagnant for a long time.
Ippolito and Rinehart name the example of animating in HTML. On one of first versions of the web browser Netscape 1 there was a bug, or ‘mistake’, in the system that allowed people to build website that had multiple pages simultaneously. What this made possible was for people to make animations by making every page a little different in order to create a sense of motion in order to make it seem like objects were moving on the screen. This feature, or bug as the browser’s developers called it, was removed after eleven months when a new version of the browser, Netscape 2, was introduced. This made it that, without any form of emulation or software downgrading, it was no longer readily possible to create animations in this way. Ippolito and Rinehart argue that this way of making art might be the shortest lifespan of an art form there has ever been (Ippolito & Rinehart, 2014).
The fact that the machine, in this case a computer, is needed to create the artwork and also display the artwork is a big departure from artworks and types of media that do not rely on this. However, the particular engineering of the product, in this previous example the HTML animation stays the same while the engineering of the browser, Netscape, changes. This causes that the bits of the artwork cannot be read properly by the browser anymore and therefore the artwork ceases its function. It is the problem that the structure of the on which the artwork depends changes which causes the entire artwork or the idea of the artwork to change as well. In the example above is it even more extreme since the artwork fully depends on the structure meaning that the artwork disappears when the structure
disappears, leaving only the idea and the documentation. For this reason digital media art and other digital objects need to be refreshed and reactivated in order to ‘rearrange’, or update, the entire working surrounding a file, which can include code, software and even hardware, so that it can be used again on the new software or the new hardware it needs to run on. This is why storage is not a long term solution for digital media because over time it will become
25 harder and harder to find the hardware or software that is capable of refreshing the files. From this point on it will become harder and harder for the files to be read and they can begin to become corrupt which can mean that the file can be permanently lost. Ippolito and Rinehart discuss three other preservative methods, namely emulation, migration and reinterpretation which I will now shortly revisit in order to see their potential in the preservation of automated algorithms.
Emulation is a form of preservation that is the closest to the original product because it is in principle a remake of the original product. What have to be remade are the inner
workings of the product such as the coding but sometimes also the graphical parts of a product if these are no longer available. This remake has to be made in such a way that it “looks the same, feels the same and behaves the same as the original” so that a lot of the original experience of the object will be kept (Ippolito & Rinehart, 2014). The only difference is that, in the case of a digital object, it usually runs on a different medium. This entails that a reproduction of an older machine or the older machine’s software is built within the current contemporary machine. The functions of the older machine are simulated so that the digital file ‘thinks’ it is running on the original machine, which makes that it can be used again. This can be a software program that can be executed on current computers, this then creates what is often called a virtual computer within a computer, or it can be an external piece of hardware that is designed to only run this particular piece of software. Often this is a positive because it can mean that that medium is solely made for the purpose of that particular emulation. This is often the case with video games, as Jesse de Vos says, which get emulated onto a completely new device that is made for (retro) emulation. Because this device is made once it inner workings never have to be changed again and therefore it is a stable product, or ‘platform independent’ (De Vos, 2018). If such a device, or independent platform, is made for a bulk of objects that are to be emulated, it can be one of the cheaper options for video games, but as Ippolito and Reinhart argue, for media art, emulation is an expensive option because often “replicas are rarely useful outside of recreating a particular work” (Ippolito & Rinehart, 2014). However, this claim is mainly for analogue media, for digital media Ippolito & Reinhart also agree that emulation is a powerful preservation tool.
In thinking about the preservation of algorithms, or the preservation of the things that algorithms create, I would not argue for emulation. Even though in some cases the algorithm itself is the artwork, it is usually the productions it creates that will be remembered and not the system itself. Therefore emulating the algorithm wouldn’t be very useful, because new
26 algorithms can be made that do something similar or something new. However, as I will come back to later in this chapter, the preservation of the original algorithm itself is something that should be actively pursued as well as it can be used as a document for tracing back the history of a certain object or artwork and be used as the documentation from which to build new objects and artworks.
The next preservation method that Ippolito & Rinehart mention is migration.
Described as a method which entails “simply upgrading its (the work in question) technology to the current industry standard” (Ippolito & Rinehart, 2014). This method of preservation usually sticks closer the medium it was intended for, especially if the upgrading only entails software changes which means the hardware it runs on often can stay the same. However, a work might lose its look and feel if the work is migrated too often, especially if the work relies on the workings or aesthetics of a particular hardware or software. Some archives already work with migration for their ‘simpler’ works, as Jesse de Vos explains. For the Netherlands Institute for Sound and Vision (Sound and Vision from now on) migration is automated. This is “performed by a piece of software that can both ‘read’ the original file-format and translate it into a newer file-type” (De Vos, 2018). One can imagine this to work for films and videos, whose presentation do not necessarily rely on a specific medium, but it is harder to imagine this working for more complex pieces of media. For film- and video art however these migrations can cause minor aesthetic changes which can change the artwork in such a way that it no longer portrays what it wanted to portray. De Vos explains this in the case of video games, but one can replace these thoughts also with other complex interactive pieces of media art. Also for art migration can be complex as the artwork can rely on a very particular version of a graphical format or software version. Because of the migration of the artwork the whole idea as well as the look and feel of the artwork might be compromised When digital films or photographs usually exist as a single file they can be easily migrated, but since videos games are built from a large number of files, many of which are also
connected and linker to each other through the working of the software or the coding language they depend on. By migrating such large numbers of complex files there is an almost certain chance that the migrated product will not work after automatic migration. For this reason a human agent is necessary to migrate the files and something even to reprogram the game in its entirety. This method is therefore not very cost efficient, but De Vos also explains that it is not a long-term method because the whole process might have to be done again once newer technology makes the current standard obsolete (De Vos, 2018). The algorithmic applications
