Nodes in non-social networks
One challenge of using social network analysis to help describe and assess the accessibility of environments is the nature of nodes. In traditional social network analysis, a node or vertex represents an entity that, when removed, reduces the connectivity of the network. In a communication network, if someone leaves, any connection to that person disappears. However, in a physical network such as a commercial district, an inaccessible business does not preclude someone from continuining onto the next business.
The two options for dealing with this have their strengths and weaknesses. The first option is to add more nodes to the network and represent a street with businesses along the sidewalk similar to that of workstations in a computer room using the bus network typology. Basically this would be like having a straight line (the sidewalk) with a node in front of every business and another node at the entrance of the business. This means that the sidewalk can can continue to be accessible (unless some barrier was encountered) even if businesses weren’t. They would be dead end branches along a main trunk. The junction would require properties such as 1500 mm X 1500 mm turning space on a level, firms surface free of barriers and hazards. The node for the business would depend on the standards for that type of business. With this approach, traditional measures of graph theory and social network analysis can be used but graphical representation can become rather messy.
The second approach is to simply connect the businesses to each other and identify two types of nodes – critical juncture nodes (nodes where two paths cross or a barrier exists like a curb cut or construction) and utility nodes (nodes that don’t impede physical access but may cause someone to change their behaviour due to cumulative effects of lack of access). In other words, if I’m going down the street and I encounter 4 or 5 inaccessible businesses in a row, nothing is physically stopping me from continuing (a critical junction node) but I may encounter a psychological barrier (utility node). The feedback I’m getting from this experience is – there’s nothing for me here so I should try somewhere else (dimunition of utility function), effectively resulting in a disconnected network.
The challenge with this is twofold. Firstly, traditional measures are not available for this type of data although Markov Chains may approximate its effects. Secondly, a utility function needs to be subjectively defined to determine at what point someone is going to give up on a particular path. This could be a heuristic such as, any time I have 3 consecutive businesses that are inaccessible, I will change direction or backtrack. Another possible rule of thumb is if there isn’t an accessible business for 50 m at a time, I will change direction. The rule has to be simple enough to visualize but complex enough to represent how a person navigates in their environment. Variables such as distance, frequency, type of node (relative to its environment), subjective utility based on the individual, rates of decay, etc. could all influence the utility function devised.
The benefit of this approach is that it is cleaner to represent on a network map. The dimunition of utility function may be necessary for both options because the same principle emerges regardless if the business is represented as being a node on the immediate path or a branch off a main path. The fact remains that if I’m rolling down the street and I encounter business after business that is not accessible, I am not enjoying a seamless, total experience and I’m likely to seek out a new destination (and subsequently spend my money elsewhere).
While this sounds like esoteric and theoretical, it may have great impact to any “connected experience” and application to other environments might prove useful.