Thursday, January 8, 2009

Exercises in Identity pt.2: machine entities

The taxonomy of identity currently under discussion can be freely extended to non-human machine entities: computers, software agents, and automatons. While such machine entities do not to date enjoy what we might accept as sapience, their participation in the elements of identity can be easily described.

Although such machine entities do not fall under what might be readily described as biological life, they do have physical, material extension. Let us call these Hardware identity. Many machine entities can be described in terms of Physical-hardware: they are made up of collection of components assembled into a single chassis. At their creation, they are comprised of a single set of physical components. Over the course of their existence, components may be removed and others added. However, such additions and subtractions are made to a collective whole that continues as a single set of components over a certain lifetime. This is a form of unique identity.
At the moment, machines do not enjoy any sort of unique code or specification. So the concept of genetic identity does not seem to apply as it does in biological entities.
Most mass-produced hardware is more or less indistinguishable on its exterior, so biometric identity doesn't normally apply. However, it is not unheard of for individual users to customise their external appearance; this could be interpreted as a form of what we might call physiometric identity.

Modern computers frequently write down certain of their internal states onto permanent digital media. This allows them to, in a certain sense, have a continuity of memory. Indeed, it is possible to remove the physical memory of one computer (it's hard drive or similar) and place it into a different computer. This allows a kind of transference of identity as the continuity of memory preserves a form of psychological identity across a change of hardware-physical identity.
Modern machine entities have never been known to show what might be proved to be sentience or an individual ego identity. Indeed, possession of a singular ego identity may be a sufficient condition for sentience in a human-style intelligence.

Informational identity is in many ways the most significant element of machine identity, as informational identity as a realm has been developed for the convienence of machines. Any machine connected to a computer network has one or more network addresses, Machine Access Codes, Universal Resource Locations, and so on. These are all forms of network identity although in many cases they are assigned to the machine identity from outside, and are can be highly variable. Many machine implementations in exisistence have encoded encryption keys that they require to identify themselves in order to access certain resources. This is a form of codestring identity.

Social identity is a complex issue when applied to machines. Some machines are named, and so can be said to have a nominal identity - I myself am writing on a machine named 'gcn-traffic' - although this is often (but not always) tied into the machines network addressing and thus its network identity. Individual computers and other machine entities are frequently not creative or communicative in a social, human sense. They do not particpate clearly in stylistic or narrative identity. Although if individual machines experience non-uniform quirks in their operation, such that the identity of an individual machine can be determined from its output (as perhaps in the case of an inject printer with a misdirected jet), then they might be said to participate in a form of stylistic identity.
Datapoint identity for machines is largely indistinguishable from codestring identity. The distinction here is that codestring identity for machines exists in specific memory adresses within the machine, places that are queried specifically in order to access those codestrings. Datapoint identity would exist if an external machine quieried an address not reserved for a specific codestring to verify a machine's specific identity; a practice that to my knowledge is uncommon.
Document identity is also underdeveloped for machine entities, although would include warranty cards or repair contracts.


Anonymous said...

Stylistic identity in a machine would thus be labeled 'disease' and summarily corrected by whoever is in charge of that machine?

Lord Carnifex said...

Sure. Any large enough deviance from some idealized standard could be considered a defect to be corrected. This applies to both machine styles and to human styles as well.

In human writing, for instance, style (to a certain extent) is a deviance from the standard rules of diction, grammer, spelling, and so on. Some shift is permitted, depending on the author's percieved value, but large deviancies are labelled 'incorrect' and are corrected in the editorial process.

Similarily, small deviances in the performance of machines are usually unnoticed or accepted by their users (although they can be found by close enough examination). Large, consistent deviances are corrected. A slightly misaligned inkjet nozzle may produce readable printing, for example. It may be indistinguishable to the naked eye from perfectly operational printing. A clogged jet that mars the visible printing, though, will be corrected.

Again, this is dependent upon the percieved importance of the machine. Just as James Joyce's style in Ulysses was not "corrected" in Ulysses, a less important printer's style might not be. But a deviation of a few micrometers in one of the U.S. currency printing plates will be corrected as soon as it is observed.