Holding deleted data, I object

The Guardian reports that an Austrian law student has lodged complaints with the Irish data commissioner after finding out that Facebook had kept 1,200 pages of data of his activities on the site, data which he had deleted. These included:

“rejected friend requests, incidences where he “defriended” someone, as well as a log of all Facebook chats he had ever had. There was also a list of photos he had detagged of himself, the names of everyone he had ever “poked”, which events he had attended, which he hadn’t replied to, and much more besides. The information was broken down into 57 categories, including likes, log-ons (a list of when he logged on and which IP address he used) and emails, which included some email addresses Schrems had never personally uploaded to the site but which he assumes were discerned from another user’s profile”.

The story broke just as I have been working on a section of JPEG: the quadruple object, looking at Facebook through some of its patents and a technical paper outlining its Haystack photo storage system (the technical paper is available here). Haystack was designed to make the billions of images on Facebook easier to manage. One way they did this was to separate the metadata from the image data. That way all the metadata that enabled the images to be found and used within the social graph – the Facebook ‘relationship engine’ – could be held in memory meaning less read/write operations, more speed and efficiency. The system effectively kept all the image files lumped together in massive files with the ‘needles’ kept accessible being the metadata necessary to find them. A consequence of this ‘deleting an image’ in Haystack consists of turning a flag in the metadata to ‘deleted’. As James Hamilton summarises a seemingly deleted Facebook presentation on the system: Facebook does not “delete photos at all since delete rate is VERY low so it the resource that would be recovered are not worth the work to recover them in the Facebook usage.  Deletion just removes the entry from the index which makes the data unavailable but they don’t bother to actually remove it from the Haystack bulk storage system”.

The deletion or not of ‘an image’ is a matter of efficiency. But it is also a matter of objects Facebook does not delete images, just their addresses. Now leaving aside the legal, ethical and even ontological issues around ‘the image’, ‘deletion’, ‘ownership’, ‘data’ and ‘data protection’ etc. This seems to me to be a perfect example of object at work, and this ripe for a bit of object-oriented exploration.

When we consider the image data, the metadata, the flags, the material Haystack servers and networks, the engineers, Facebook brand and its governmental ‘relationship engine’ as objects connects within objects, this story makes sense. Haystack was designed as effect way of connecting objects so that new connections could be made by Facebook users and Facebook algorithms. The more cone tins, the more the Social Graph object within which they connect grows, generating new connection and so objects “Jo has tagged you…”, “Sam has added new images…” etc. From a quadruple object perspective, the real dimension of those objects, their complete totality remains withdrawn. What the objects in play conect with are particular instantiations, the sensual dimension of objects.

As has been noted the engineers think in object terms. The lawyers’ discourse too is object-oriented. The ‘modules’, ‘machines’, ‘databases’, ‘fields’, ‘needles’ and ‘files’, as well as the tagging, ordering and connecting associations detailed by both are objects on the pages but also on material servers. Where Harman’s framework adds value is firstly in opening up the scope of the objects that make up the engine and secondly in explaining how they connect as a matter of objects not some wider field. In short it enable us to see Facebook’s governmentality as a matter of objects not as the outcome or result of an external governmental rationality.

From Harman’s perspective what we have seen is objects connecting within objects. The objects discussed in the patents and technical documents and familiar to the engineers and lawyers are joined by the Facebook user-object, the photographer-object, the Facebook brand-object, the ‘friend’-object – a whole collection of objects within and without Facebook. All have a real dimension that withdraws. We can never access the totality of the “relationship storage module” or the “identifier”. We encounter its sensual dimension as we (human user, software algorithm, image data etc) expand energy on it. But those objects have real dimensions outside of our relation to them. The Haystack has a reality ‘beneath’ the shifting accidents as we walk around it, encounter its particular instantiations. But those wider objects in play have that dual character too. The Facebook brand, the Social Graph are not just ideologies floating in the abstract. We encounter their sensual dimension as they are realised in the world but they have a reality, an eidos and an essence that holds together the various characteristics, that connects with the sensual dimension of other objects.

It is when we come on to look at how objects connect that we realise that engineers actually have a lead on philosophers. For the developers of Haystack or the creators of the Timeline, objects connect within object. The ‘network’ is nothing more than nested objects: components connect within engines, metadata objects connect within indexes within servers within… For them this is a matter of engineering reality. Harman would see the same thing. Objects connect within objects. There is nothing outside objects. User-objects and governmental ‘self’-objects connect within profile-objects, connect within timeline-objects, connect within social graph-objects, connect within…

What Harman would add is that those object never fully connect. Objects always withdraw from each other. The sensual dimension of one object connects with the real dimension of another. The real ‘user’ connect with the sensual dimension of the profile-object, an instantiation, a particular accident apparent as long as the user pays attention – is quite literally logged in. The same real-sensual relation happens between unhuman object-actants. The real social graph, that unfathomable but actual reality in the world, connects with the sensual dimension of an image tag, that instantiation or quality accessible to it and necessary to its work. The reality of that tag (whether at the nested scales of the electrical charges, machine coed, or even semantic nature) remain withdrawn.

Facebook as relationship engine, as an issue of governmental rationality, surveillance, sousveillance, power, psycho-cultural practice or any other framework by which it is explored is a matter of objects not of some meta-field of becoming, processuality or determinacy.

The deleted/undeleted flag is an object that connects with the image file but also with the server load-object, the Haystack system-object, the PowerPoint presentation in the Facebook management team on costs of server space and bandwidth and now lawyers in Dublin.

The issue can be addressed as and through objects. There is no need to look at broad, ill-defined and inaccessible fields of ‘privacy’, ‘surveillance’, ‘data-mining’ or ‘technosocial capitalism’. All of those are matters of objects too. by starting with objects and tracing their real, materiality and presence as well as the way they connect within objects, those fields also become accessible to analysis and potentially change.