IEEE Technology and Society Magazine - March 2015 - 70

which enable them to understand how the other person
is situated. When the other person in a communications transaction is situated doing different things that
the observer is not aware of, due to being in another
location, the observer needs to learn to make more
conservative estimates from their inferences, or they
will be at risk of making judgments that will not achieve
desired outcomes. As a rule, more general inferences
are less tailored to specific individuals (and situations)
and are not necessarily the most accurate or the most
efficient. Generalized inferences do not work as well as
more tuned coordinated social interchanges. This kind
of impact of fragmentation in PoSR could easily happen
in an engineering sense: messages, the foundation of
sociability, require observations and other data to produce accurate inferences and judgments for successful
communication and in turn, successful cooperation.
When observational cues are absent, more conservative general estimates must be made. It is not a hardware problem; the hardware can process whatever it
needs to in a vehicle in terms of data. But software is difficult to write because, unless there is some type of corrective for contextual interpretation, more conservative
judgments will need to be made, which in turn means
less efficient/accurate/appropriate judgments, which in
turn reduces the scope of what can be accomplished. In
a car that is monitoring many different sensor inputs plus
potential multiple, multiplexed, social messages that
contribute to interpretations of PoSR context, plus its
own agency, the event of one message interpreted poorly
could have disastrous results. This problem also makes it
difficult to certify such a system because in order to certify it, nearly all of the local inferences will need to be as
close to 100% reliable as possible. Due to variability of
interpretations in PoSR with respect to multiplexed messaging from a hypothetically huge number of vehicles on
the roads, this becomes nearly impossible.
This complexity problem emerges from a combination of agency, volume of messages, and the context of
the messages that are both coming in and being sent
out to other people and vehicles. Not all messages are
available to all segments at all times. An incomplete
distribution of the messages creates more confusion,
as it is impractical to send everyone every message and
expect them to process the data.
Because there only needs to be one message that is
not properly contextually transmitted or interpreted for a
disastrous result, particularly in an automobile, accounting for a broader range of activities that are happening
at any given time will have to be designed into the system. In other words, PoSR contexts might be alleviated
somewhat - if each network member knew something
about the other networks of each other member. Solutions might include agents that summarize different

things for different parts of the communications in
order to create more accurate interpretations of messages or to design less efficient systems. A possible
solution, in part, may be to produce some kind of subsystem that manages context within PoSR. A system of
contextual "tokens" could accumulate for each context
and be transmitted with messages as a kind of summary of the other's originating and interpretive context.
These could be collected as they flow across more and
more networks. This would produce a contextual history
that is attached to the communications, allowing for the
development of more agency on the part of the system.

Thing Theory and the Connected Car
The realization of the connected car might be possible, or at least improved, by incorporating internal
systemic decision making with reference to agency and
bi-directional interaction.
Thing Theory [18] is an agent-based network model
that conforms knowledge of the local context of each
participating Thing-agent to a common understanding by all Thing-agents, thus granting agents within
a network a shared contextual understanding. Local
knowledge for each Thing-agent is gained through its
relationship to people and other agents in the environment, as well as system-subagents and meta-agents in
other systems. A Thing-agent is informed of the local
knowledge, which is then fit to the objectives and circumstances of all participating agents.
Thing Theory was inspired by the notion of the character "Thing" from "The Addams Family," a 1960s television
show based on a comic of the same name created by
Charles Addams. Thing is characterized as a disembodied hand (and forearm) that has been with the family for
many years and is described as both a "family retainer"
and "friend." Thing inhabits a series of tabletop boxes
in different rooms of the house [19] that could be compared to a type of roughly cobbled physical network.
Thing communicates with the family by gestures, sign
language, writing out notes, or tapping out messages in
Morse code. Thing serves the family by accessing a portal
in contextual proximity to what is needed or desired at
the precise moment it is required, in the precise room or
context needed. Thing is not only a ubiquitous agent, but
also an anticipatory one that migrates within the environment. The sensing, response, and location-awareness of
Thing is a useful aspirational model for a network agent in
a static or moving location-aware smart environment.
Thing functions as a communal agent. It develops
relationships and shapes knowledge and responses for
different actors within the system that it serves. The
Thing-agent collects information on all systems. While this
presents issues with protecting the privacy and security of
all parties in different connected cars, the development

IEEE TEchnology and SocIETy MagazInE

∕

MARCH 2015

Table of Contents for the Digital Edition of IEEE Technology and Society Magazine - March 2015