Erkki Kurenniemi has documented his life, but has not archived it. He didn’t seem to bother working on a systematic model for what he calls a template for all human life. In his profound techno-enthusiasm, he relies on the quantum computers to make sense of it all. The machine will sort it out all right, and he concentrates on recording, capturing, filming, photographing, drawing, talking, and keeping every object he interacts with, from computers to grocery tickets.
We have no quantum computers to make sense of it all, and yet we are happy with the messy conglomeration that we have been commissioned to inspect and engage with. Usually when we begin to work on an archive, the material has been already processed, ordered, and a classification scheme is more or less decided. Our role as active archivists is often to negotiate between the classifying scheme and the resistance of the data to comply with it. As this time, the intelligence that will organize the data will only appear in 2048, we have the opportunity to sneak in early in the process.
Our interest is to explore what we can sense from the material when we cannot rely on an external set of metadata coming from a source of reference. Contrarily to what one may think we don’t lack information about the data, the images, the sounds or the texts. We may lack contextual information, social historical information, but the files themselves contain their own metadata, their own composition, their own structure, etc. We cannot directly access these structures just by looking with our eyes, we need intermediaries. These intermediaries, our senses to “read” the images will be algorithms.
What we will try to do is to show, invite others in the dialogue between these intermediaries and ourselves.
Doing so means to challenge dominant modes of representation of the data, namely ‘direct’ representation (Image display) or literary descriptions of the data (Grouping & Retrieval via metadata/tags). More often than not, image databases reach their climax when they display the image in high resolution: here comes the Image. With the appropriate description that underlines what one needs to ‘read’ in the image. And the invitation to access similar images based on their place in a taxonomy. Images databases are caught in a recursive conceptual loop: an image is an image is an image. And in the iteration of this loop, the data is reduced to its classification.
But the fate of an image is not the traditional illusionist representation. It can be a lot of things: an interpreted composition of pixels, a collection of statistics, lines of contours or directions, a music score, a legal reference or all of this together.
And the possibilities of connexion between images are by far richer than the mere parallel story of the taxonomy (or folksonomy).
An image is not a black hole whose meaninglessness has to be redeemed by a legend. Given the right intermediaries, what if we could turn pixels into interlocutors?
The question of images are only symptomatic when we talk about archives and especially about databases. In the coming months, we would like to pay attention to tasks which are usually treated as secondary and usually delegated to a subaltern workforce, historically sometimes to women and nowadays outsourced to developing countries.
The “practical” matters of working with the data links an archive project to the very materiality it hopes to transcend. Therefore questions like “How much time does it take to copy?”, the “weight” of the mass of data, the conflicting formats will be treated as integral part of the investigation and not as mere insignificant annoyances that we need to overcome.
By looking at the relationship between data and work, we need to recognize the dimension of sociality inherent in an archive project. All items in an archive are shared objects. They are produced in a transaction, collaboration with instruments (pen, paper, camera, recorder, etc) and people. Every object embeds a web of relationships. And these relationships are framed by the law.
By making an integral archiving of his life through different media, Erkki generated an impressive amount of shared objects. The different people, instruments will have their say in how the images, sound files may be communicated. At this early stage, the identification of these relationships has hardly begun. The law therefore acts as a filter through the intermediary of the lawyer and policies of the Finnish Archive. A shared object cannot be published until its different co-owners are identified and have reached an agreement. A dialogue that the lawyers will connect each item to different articles of the Finnish Law (and probably international law) regarding privacy, copyright, etc.
Here again, we don’t want to treat the gray literature of the database (the legal documents, the agreements between the different stakeholders, etc) as mere obstacles to overcome as quickly as possible. On the contrary, we want to integrate them in the work.
Intermediaries function as detectors of relationships. Algorithms in dialog with the law. If we won’t use a face detector to show the face it detects, we can still use the information that there is a face (or more) in a picture. If a color analysis detects the presence of skin, it may calculate the proportion of skin in an image without showing the skin or simply highlight the skin zone or the heat zones of an image. If all the pixels combined can lead to unwanted identification of a person, the reduction of this image to its averaged pixels may as well re-present the image. Even one pixel can stand for an image. Or a Gaussian filter can output a blurred image that will give some impression of the image without giving out any detail.
And detecting the presence of a face may well connect the estimation of an algorithm to a legal filter. Detecting a face is a first step in identifying the possibility of a shared ownership of the document and therefore to a legal dimension (such as privacy right).
If the images cannot be ‘shown’, – and perhaps this is a blessing rather than a tragedy – what can be shown are the relationships.