TRAVELLERSIM : SETTLEMENTS, TERRITORIES, AND SOCIAL NETWORKS

Shawn Graham
University of Manitoba
grahams@cc.umanitoba.ca

James Steiner
turtlezero.com
james@turtlezero.com

Is it possible to move from 'dots-on-a-map' to 'fuzzy territories' of socially interconnected settlements? .....yes!

Model presented at the Computer Applications and Quantitative Methods in Archaeology 2006 conference, Fargo North Dakota.

Introduction Basic Hypotheses Re-implementation Outputs Importing New Maps  Other Models

(if there is a problem with the applet, you may need the latest java plug-in for your browser.)

 

created with NetLogo

view/download model file: Travellersim1.0.nlogo (this file includes map import procedure and social networks output procedures not shown in the applet above). A minor bug has been corrected in this version of the simulation; happily, the buggy version wasn't the version that I presented at Fargo!

The slideshow demonstrating the model, presented at the Computer Applications in Archaeology Conference at Fargo may be viewed here


INTRODUCTION

Archaeologists are concerned with, amongst other things, trying to understand whether sites discovered through field survey can be used as evidence for ancient territories, and by extension, understanding which sites are most likely to be important locations in the landscape. Various methods exist, but most are problematic because they do not take into account the actions of individuals, the way different individuals can react to 'tangible' and 'intangible' boundaries, and the fact that the same place may belong in different 'orbits' depending on how we look at it. We have created this model to deal with this problem of the individual and socially-constructed landscapes. We are inspired by the work of Tracey Rihll and Andrew Wilson, who write:

"The emergence and evolution of the Greek polis has a strong spatial aspect. It involved the formation of a community in a territorial unit encompassing a number of settlements, and the development of a 'captial' city. To have things in common, and in particular to share a common identity, presupposes a relatively intense level of interaction amongst those who constitute a community vis-a-vis those who are excluded from it. When the poleis were coming into existence, did discrete communities align themselves with those with whom they had most in common - those with whom they experienced the most intense interaction? Did location vis-a-vis other settlements have a significant effect on their affiliation and union?"
(Rihll and Wilson 1991: 60)

Rather than exploring the traditional archaeological concerns with site (geographic and environmental factors of the location), their model, and our simulation, is concerned with situation. In this way, we move from 'dots-on-a-map' to something of the human interrelationships between sites.

BASIC HYPOTHESES

The simulation presented here represents an attempt to reimplement Rihll and Wilson's model, or at the very least, create an allied agent-based model. In their model, a distribution of sites from an earlier period represents a starting point for simulating 'credits' and 'debits' of interaction from site to site. Mathematically, the model attempts to solve a series of differential equations, eventually settling on the 'best' answer. Two parameters, aside from the 2-dimensional scatter of settlements, are also modelled, to simulate difficulties in communications and the benefit of concentrated resources (hence attractiveness of a site for interaction).

Rihll and Wilson's basic hypotheses are that:

1) interaction between any two places is proportional to the size of the origin zone and the importance and distance from the origin zone of all other sites in the survey are, which compete as destination zones;

2) the importance of a place is proportional to the interaction it attracts from other places

3) the size of a place is proportional to its importance
(Rihll and Wilson 1991: 63, 60)

It is worth noting that we are discussing travel by foot.

RE-IMPLIMENTATION, RE-IMAGINATION

Their model does appear to predict eventual settlements of some importance, as well as indicating the hierarchy of lesser sites that 'look' to the main one. Our reimplimentation does reproduce their results. Coupled with social-network analysis of the resulting social network of settlement interconnections, sites of particular importance, as well as 'fuzzy territories' of overlapping factions (as defined through social networks analysis) can be determined.

We have chosen to represent the interaction between sites as the result of actual travel by agents. Our model has two 'breeds' of agents: settlements, and travellers. Each traveller has a limited vision, or knowledge of its neighbourhood. The 'vision' is set variable around 20 km, or roughly the distance covered in a day's travel by foot. Travellers decide which settlement to visit by calculating a score for each settlement based on its importance, number of visitors to it, the benefit of concentrated resources, the distance to the settlement, and the difficulty of communications. The traveller visits the settlement with the highest score. It leaves a coloured trace behind it, indicating where it has travelled.

When it arrives at its destination, it checks to see if the place where it has arrived is 'more important' than the place from which it left. If so, it changes its colour to that of the new settlement. If not, it changes the colour of the settlement to its own colour. In this way, the influence of sites over one another is demonstrated, yet mediated through individuals. The 'Territories' histogram counts the number of individual colours, and the number of settlements displaying those colours, as a rough way of showing the number and size of emergent territories.

The settlements are both two-dimensional points in space, and active agents aware of their environment. It is helpful to think of a settlement agent as a 'genius locui', or spirit of the place. Their primary function is accounting, keeping track of interaction. When a traveller arrives at a settlement, the settlement's importance increases. Attracting interaction increases a site's importance. But if a settlement does not attract any visitors in a given turn, its importance declines.

The 'benefit of concentrated resources' slider allows the user to simulate something of the macro characteristics of the economy of the region. The 'difficulty of communications' slider lets the user simulate difficult travelling situations (e.g., winter, heavy vegetation, what have you).

Increasing the number of travellers who are each differentiated by their ability to perceive their environment, makes for richer interactions, but also a slower simulation and a more cluttered display (the 'Clear Traces' button erases the traces left by travellers at the end of the run; the 'auto-clear' switch will clear traces every few cycles of the model run). When exploring the behaviour of the model, it might be best to have the minimum number of travellers, since this will create a clearer display, and also be comparable to the original simulation's method whereby each settlement calculated its 'debits' and 'credits' compared to every other settlement.

MODEL OUTPUTS

This model produces various data which can be considered on their own or exported into another programme for analyses. The 'territories' histogram in the top right of the interface window merely counts the number of settlements by color. The number of unique colors (as reported by the histogram) corresponds with the number of unique territories (which may also be seen on the map). This is the lowest level of complexity visible in this model.

The 'write network' button asks all of the settlements to list the settlements-of-origin for visitors to that settlement. This allows the user to study the social network of interactions between sites. THIS IS NOT THE SAME as the pattern of interconnections displayed in the view window. All travellers remember their home settlement; by visiting a new site, they create a social connection between it and their home site (compare with patterns of euergetism in the Roman world). This social network itself, since it is created through the interaction of agents, is a rich source of data in itself. UCINET (www.analytictech.com) is used to analyse these networks to identify the most central, the most powerful, the 'weak links' between local clusters of settlements.

To get the output into a format that UCINET or KEYPLAYER can analyse, you should edit the text file so that eg "turtle 1" becomes "turtle_1" (the id's of the settlements in netlogo-speak). Also, you might need to remove ( ) from around the turtle id's.. Then, place the following five lines of information at the top of the file created by the 'write network' button:

dl
n = x
format = nodelist
labels embedded
data:
turtle_0 turtle_23 turtle_46 turtle_56

.....etc. X = number of settlements, or 'turtles' as Netlogo calls agents. That first line of sample data above tells the social network analysis software that turtle 0 is connected to turtle 23, turtle 46, and turtle 56. It does not imply that turtle 46 and turtle 56 are directly connected.

Other outputs:

The two switches, 'identify-most-important-sites' and 'resize-settlements' provide two alternative visualisations of site importance. The first runs at the end of a simulation, identifying the sites with the largest importance by a red halo; then green, then yellow, then blue. Resize-settlements operates during the model run, and causes the size of the settlement to scale importance to maximum site importance on a scale of size 1 to 20 (so the most important site will always be displayed as size twenty; this is merely a display variable, and not used in any other calculations).
Finally, the button 'list origins...' looks at the most important sites based on the variable 'importance' and lists their social connections.

ASSESSING MODEL OUTPUTS

The social network of interconnected settlements, generated by travelling individuals (and output by the 'write networks' button), can be studied from multiple viewpoints to meet Thomas’ (2001) idea of the ‘productive tension’, the resolution of his two understandings of the word ‘landscape’, of a ‘territory which can be apprehended visually’, and a ‘set of relationships between people and places which provide the context for everyday conduct’. Social network analysis allows us to consider both local and global positioning of a settlement vis-à-vis every other settlement.

Social network analysis is a suite of methods from the mathematics graph theory (which considers sets of connected objects). In social terms, it is predicated on the idea that agency can be mediated through both local and global positioning within a network. I.e., individuals who are well-connected have a greater range of opportunities since there are more channels through which information may flow to them.

With SNA, we can analyse the ties between settlements: determine which settlement is best connected to the others = social power, which settlement forms a link b/w otherwise disconnected clumps of settlements (forming a social bridge), which settlements can wield the most influence over others.

We can analyse which settlements are allied in their patterning of interconnections, and then use that patterning to determine likely ‘fuzzy territories’, and to understand the interrelationships of those territories.

Because this model depends on site and traveller interactions, and a number of outputs may be produced in a model run, we suggest using the following network metrics (KEYPLAYER and UCINET) to identify the most important sites:

1. Fragmentation criterion, group size 10. 10 starts. 20 iterations
2. Dependency (Bonacich Power with a -0.5 attenuation) (top 10)
3. Flow betweeness (top 10)
4. Degree
5. Factions (sets of overlapping patternings; 5 is a good number to start with)

The fragmentation criterion identifies nodes in a network whose removal cause the network to collapse into isolated fragments.

Bonacich power index identifies 'powerful' nodes in a network based on their relative positioning with regard to every other node. When negatively attenuated, it identifies nodes that are well connected to poorly connected others, which would seem to correspond with Rihll and Wilson's 'terminal' sites.

Flow betweeness identifies nodes which are on the majority of shortest paths between every pair of nodes in a network.

Degree is the simple number of connections.

Factions are sets of settlements with similar patterns of interconnections. Identifying factions also allows us to identify patterns of ‘alliances’ between factions (where there are overlaps in the patterning of interconnections).

SUCCESS?

A successful reimplementation will make similar predictions of site importance, hierachy and territory. Accordingly, the same map of sites from Geometric Greece is used here. When run at the same parameter values as in the original 1991 model, very similar results emerge.

The various metrics did indicate the importance of settlements such as Athens, Corinth, and Megara; for cities like Thebes it indicated extremely near neighbours, which is in fact what Rihll and Wilson found. The most important site, according to our experiments with this model, did not evolve into a city at all, but rather is the extra-urban sanctuary of the Argive Heraion, which is an intriguing result.

In any event, ‘situation’ rather than ‘site’ is a clearly very important part of the story in the evolution of the later city-states, but not the only factor. When we look at ‘factions’, to understand those fuzzy territories, the model seems to accurately predict the location and extent of allied groupings. The patterning of densities of overlaps within the factions also points to a heightened importance for Corinth and the Isthmia (the patterning of ‘alliances’ seems to lead to this faction in particular). This accords well with the evidence of pottery, where corinthian-wares are found across the eastern mediterranean and into Etruria in this period.

When we ran the model on the protohistoric period (8th – 6th BC) of Central Italy, we achieved results which also would seem to validate the model. The various network metrics indicated settlements such as Falerii Veteres, Fidenae, and Veii being extremely important. This agrees with the early history of Rome, when these cities were seen as being Rome's major competitors in the region. It is interesting also that these early settlements – all conquests of Rome – ranked higher than Rome did itself in our network analysis. This would recast Rome's early wars of expansion in part as a re-jigging of the networks to improve Rome's 'situation'.

This simulation then seems to have the possibility to be a useful tool for understanding the interrelationships between sites, the likely territories in a given area, and sites likely to be of some archaeological importance.

IMPORT NEW MAPS

The simulation looks at a map for two crucial things: a scale bar in red, two pixels thick, and single violet pixels marking the location of site. If your map has these (and the scale should be as long as an expected day's journey), then you can simply add the map name to the map chooser.

The 'process new map' button is for maps digitised by hand or exported from other programmes, where the sites are indicated by a red spot (but not necessarily a one-pixel spot). It processes the map by converting the red spots into single violet pixels (1 violet pixel = 1 site), and saving the output. These semi-processed maps may then be touched up using MS Paint or similar program to add the scale bar, and to check the location of processed sites.

1. Export from your GIS (or digitise/scan in) your base map of settlements as a PNG. Make sure these are coloured red.

2. Add the name of this base map to the map chooser.

3. Select that map.

4. Click on 'Process New Map'. This will load the map, and begin to adjust it.

5. Open the resulting map in your image software. The name of the resulting map will be
'base map.png.xxxx.png'
where base map is the name you gave the map,
xxxxx is a number based on the internal clock of your computer

6. Examine the settlements. These should be violet SQUARES of pixels. If the settlements are rendered as a line of pixels, draw in the necessary pixels to make it square.

7. Add a two-pixel thick line, coloured red, which will be the scale bar. Make it the appropriate length of a single day's journey (we usually use 20 km). Save as a PNG with a new name.

8. Add the newly-named map to the map chooser in the model. Select the new map. Press the SETUP button. Examine the map. If all has gone well, your violet-coloured settlements will have spawned settlement agents. If not, you might see some settlement agents, and some settlements pixels. If this happens, open the newly-named map and fix the settlements at issue.

This process can be a bit awkward. We apologise; hopefully as more people explore this model better import routines will be developed. You might be able to skip some steps if your GIS allows you to colour your settlements with the correct shade of violet in the first place. Use the 'Tools - color swatches' menu in Netlogo to work out the correct shade (violet = 115).

Existing Maps
A number of maps are provided with the simulation. The numbering of them is a legacy of the development process, and need not concern the user.

"Greece_2.png" is a map of central Greece with settlements from the Geometric period plotted. This map was hand-digitised from Rihll and Wilson's original article.

"Tibervalley3.png" is a map of central Italy, with protohistoric sites marked on. The data was culled primarily from Chris Smith, 'Early Rome and Latium', Tim Potter, 'Changing Landscape of South Etruria', and other archaeological maps conserved at the British School at Rome.

"Upper Ottawa.png" is a hand-digitised map of the Upper Ottawa Valley in Canada, showing settlements founded in the 19th century before the advent of the railways (from a base map published in Canadian Geographic Magazine)

The two 'test pattern' maps allows the user to control the placement of settlements (lattice, ring) to explore the effects of altering population and the other sliders. It also allows the user to explore and test for 'edge-effects'.

WORKS CITED

Borgatti, S. P., M.G. Everett, and L.C. Freeman (1996). UCINET IV Version 1.64. Nantick, MA, Analytic Technologies URL: http:www.analytic.com.

Potter, T. (1979). The Changing Landscape of South Etruria. New York: St. Martin's Press.

Rihll, T.E. and A. G. Wilson (1991). “Modelling settlement structures in ancient Greece : new approaches to the polis”. In J. Rich and A. Wallace-Hadrill (eds). City and country in the ancient world. London : Routledge.

Smith, C. (1996). Early Rome and Latium Economy and Soceity c.1000 to 500 BC. Oxford: Clarendon Press.

Thomas, J. (2001). "Archaeologies of Place and Landscape". In I. Hodder (ed). Archaeological Theory Today. Cambridge: Polity Press.

ACKNOWLEDGEMENTS

Canadian Research Chair in Roman Archaeology, Department of Classics, University of Manitoba

The British School At Rome

Thanks also to the participants in the simulation modeling session at the 2006 Computer Applications in Archaeology Conference, Fargo, North Dakota

COPYRIGHT

(c) Shawn Graham, James Steiner 2006
Some Rights Reserved.
Creative Commons Attribution-NonCommercial-ShareAlike License v. 2.0.
Visit http://creativecommons.org/licenses/by-nc-sa/2.0/ for more information.

To cite this model and this information page, please use:
Graham, S. and J. Steiner (2006) "TRAVELLERSIM : SETTLEMENTS, TERRITORIES, AND SOCIAL NETWORKS" http://home.cc.umanitoba.ca/~grahams/Travellersim.html

This page was automatically generated by NetLogo 3.1. Questions, problems? Contact feedback@ccl.northwestern.edu.

The applet requires Java 1.4.1 or higher. It will not run on Windows 95 or Mac OS 8 or 9. Mac users must have OS X 10.2.6 or higher and use a browser that supports Java 1.4. (Safari works, IE does not. Mac OS X comes with Safari. Open Safari and set it as your default web browser under Safari/Preferences/General.) On other operating systems, you may obtain the latest Java plugin from Sun's Java site.