This post is the fourth of our temporal observatory project (Tempobs) and related series focused on the future of the relationships between Saudi Arabia and Iran and aiming at improving the handling of time-related issues in strategic foresight and warning, risk, or more broadly anticipatory analysis. It answers and builds upon Dr Warren Fishbein’s (hereinafter Warren) previous article (Mindfully Mapping a Middle Eastern Morass – Saudi Arabia and Iran), as we designed the series as a dialogue where we progressively build the understanding related to the foresight issue by mapping the corresponding conceptual network, continuously scan the relevant literature and news, which will allow us, finally, to assess the future, to use Warren’s apt presentation of the work involved.

Here we shall present the new tool (best on desktops and tablets, but not on cell phones*) we are developing and using, which will provide readers, analysts and decision-makers with an interactive experience of the model or conceptual map for our question: “Within which timeframe could we see full cooperation or, on the contrary, war occur between Saudi Arabia and Iran?”, as shown in the screenshot below, on which you can click to access the full-page interactive network. We shall progressively explain the features of the interactive network and how to use them, while continuing building upon the initial map Warren started previously. We shall notably add three new cluster nodes for “categories” that were not yet considered, and then further develop the nodes (representing the factors) related to the central question to make sure dynamics and underlying processes can fully be considered.

screenshot graph whole 13 11 15 2
Click on image to access the interactive graph. Warning – it is difficult to use the graph on cell phones as the screen is too small.

From ink and paper to an interactive mapping network

To present properly the advantage of being able to use an interactive mapping network, let us first recall briefly what, most of the time, we have been used to when thinking about a problem or when analysing a question.

In general, when we think about something, we come up with an answer that will be more or less detailed according to our job or to the setting into which we are making the thinking. If this thinking happens during a conversation or discussion, thus under extreme time constraint (we cannot tell out interlocutor, “hold on a few hours/days, I’ll come back after I have read a few books and articles and thought about this”), we shall come up with an answer that will be more or less well-reasoned and informed. If our interlocutor opposes our statement, then this may prompt us in developing further arguments, preferably grounded in facts we have retrieved from memory.

Arab_israeli_memo sc
CIA Analysis of the 1967 Six-Day War between the Israel and the UAR (Egypt). The first page only of the draft of the estimate that predicted the outcome of the war – Original document created by CIA. Ian Pitchford at en.wikipedia, Public domain, from Wikimedia Commons

If we are analysts and the thinking thus takes place in the framework of our main job, the time constraints are less extreme – although most of the time still there. We shall work upon the “research question” and then write our analysis, in a more or less detailed way, according to various factors, from the time available to make the work to the space given to express the result of the analysis. This is what we would call a “classical analysis”, moving from the cognitive analytical process that took place in our head to sentences phrased on paper. Sometimes, a certain amount of brainstorming is included in the process, or other heuristic devices are used to improve our thinking, but not always. Even with these thinking devices and brainstorming, most of the time one always goes from our “head” to sentences and paragraphs on paper, be it virtual or real. In terms of analysis dealing with the future, this is what you will most of the time find when you read a political risk analysis as provided, for example, by the Eurasia group – e.g. Ian Bremmer and Cliff Kupchan, “Top Risks 2015” – or yearly by the Economist – The World If series, here using a “what if” approach to induce thinking about the future – to take only two cases among a flourish of similar documents.

Tapa_Potenciar_la_Memoria_1 cropped video red sc
Reworked image from original image “Tapa Potenciar la Memoria” by Alejandra Burzac, 2013, CC BY-SA 3.0 via Wikimedia Commons.

Most of the time, analysts when they create their analyses, people when they think, or, more generally, all of us, human beings, when we think, even if we are unaware of it, use an implicit model, which is there, “in our head”, to make the reasoning necessary to obtain argument and analysis. We thus always, and we must insist on the always, use a model when we think about a problem. It is this model that actually guides what we write on paper, thanks also to our knowledge of a language.

If, in many circumstance, to rely on an implicit model is perfectly fine, when we deal with foreseeing matters of vital importance, which will then lead to decisions and actions being taken and impacting large groups of people, it becomes at best sub-optimal and at worst dangerous. Renowned systems theorist Joshua Epstein, in his masterful article “Why model? (Journal of Artificial Societies and Social Simulation vol. 11, no. 4 12, 2008), besides explaining that we are all modellers, lists 16 reasons, from being able to explain, guidance of data collection, illumination of core dynamics to education of the general public and capacity to move beyond appearance, that should convince us that we need to make our thinking, our implicit cognitive models explicit.

Furthermore, not making our thinking model explicit, is to open the door to the many biases that may undermine our analysis (e.g. Heuer, Psychology of Intelligence Analysis, 1999). It will notably make difficult if not impossible the evolution of our thinking models when new information and knowledge is known (for example being unable to see that the Islamic State is a new form of state and NOT solely a terrorist group), what is called “Belief persistence after evidential discrediting” (Anderson, Craig A., Mark R. Lepper, and Lee Ross. “Perseverance of Social Theories: The Role of Explanation in the Persistence of Discredited Information.” Journal of Personality and Social Psychology 1980, Vol. 39, No.6, 1037-1049). Not transforming our implicit models into explicit ones will also make it very difficult to work collaboratively on an issue. Finally, if we were to build scenarios for the future, or to warn about something, without an explicit model we would lack a sound methodological basis for our scenario-building and our warning, which would diminish both the reliability and the use of both.

Now, to make a model explicit, we need to systematically identify factors (or better variables), and if we only rely on pen and paper (or word-processor), most of the time the best we can do is to get a laundry list of factors. Then we also need to systematically point out all the variables or factors which may relate to another. Again with only a pen and paper, we would have to rewrite many times the same sentence (Factor A will link to factor B, factor A will influence factor C, etc. ) and finally we would have loads and loads of quite boring pages. If we try to make it interesting – what we actually do when writing an analysis, then the systematic and neutral character is lost and thus the model is not explicit anymore. We thus have to find another device that takes us, at least as far as making the model explicit is concerned, out of the classical writing process. This device is the creation or mapping of a network or graph.

An analytical working graph done with the desktop software Gephi

Up until now, when we wanted to build or map a model, we had mainly a few desktop tools (the specificity of politics, security and international relations on the one hand, of foresight analysis on the other, restraining the choices available), as we used in the previous post. If we wanted to share this work, either for cooperative work or to present it, notably online through a website, for example, we were left with hardly any other choice than converting something that was meant to be evolving and interactive, a real tool to work on an analysis, in an image, which is static. There is virtue to this approach as it freezes steps and thus is didactically helpful, especially when the model is being built, i.e. when the implicit model we have in our head is made explicit, refined and detailed to make sure we do not miss anything important to understand not only the present but also the future. We shall thus continue using it. It is however also limited because it stops interactivity.

Now, we can benefit from the capacity to interact with models being built, thanks to the visualisation which has been developed with Gephi (an excellent open source visualisation software for social network analysis, perfectly adapted to our needs), Sigma.js (“a JavaScript library dedicated to graph drawing”), and the Sigmajs exporter (developed by the Oxford Internet Institute, University of Oxford).

Throughout the rest of the article, we shall walk through the new interactive graph, starting from where Warren stopped last time, showing its capabilities, meanwhile developing further the model. Indeed, this tool being quite new, we must all learn how to use it to take full advantage of it. After having learned to move from ink and paper to word processor, we now need to learn include one more tool in our panoply for analysis, using an interactive network .

Mapping further our network for Iran-Saudi-Arabia

The need for well-defined and labelled variables

When we moved from the previous map to the interactive one, it was not anymore presented according to the nice initial categorisations Warren had created to make it appear simpler to a WF post 1 image 2 2 scgraph 13 11 for comparisonreader. Indeed, in social network analysis, the way a network looks, its visual appearance, matters according to the relationships existing between the nodes (the factors or variables or even actors). This will be a plus for us as our network grows, as the shape and the visualisation will give us indications regarding nodes, i.e. the factors, drivers or actors, and the whole question.

However, as nodes moved around, when the translation from one graph to another took place, we had a few nodes that appeared as not sufficiently defined. For example, we had twice “strategy/diplomacy”, but we did not know anymore to what they related. The label of the nodes was thus changed to reflect what they meant exactly. You can check the information on this node by clicking on the node, in the interactive graph, for example for the re-labelled “SA strategy/diplomacy” here.

You can see, as information given for one node, a description, the post of reference and the author of the node. As the network will grow and evolve, as we shall move from the cluster nodes to more detailed networks, as Warren explained previously, we shall enter more information in the description (see also below for more detailed nodes). The user will thus be able, at first glance, to have all necessary information on the node, and, if s/he wants more, to click on reference links. This capacity is very important as it allows explaining why this node was created and what are the grounds for this creation. This should allow us to make sure that as much knowledge as possible is incorporated in our network, including respecting the contributions of each. For example, SA Public Spending was created thanks to the comment of a reader, but you will see other types of references, such as books, or articles, below.

Later on, as we shall scan and monitor the issue, we shall also be able to easily enter changes and reference evolutions taking place in real life for a specific node (of course, this demands that the node be specific enough to allow for monitoring, and we shall not reach this stage immediately).

A few nodes remain in need of better definitions because their definitions will influence the way they are then developed into a more detailed network. For example, should we use “actors” or “powers” for “global powers” and  “regional powers“?  We should use as neutral as possible a definition, hence actors should be favoured actors. In this light, “regional powers” was changed for “regional actors”. However, as far as “global powers” is concerned, Warren could have meant actors with a global status, who are usually perceived as influencing the world. The node was thus not changed, waiting for the author of the node to decide.

The term used of “region”, as well as “regional”, will also need, in the future, to be better defined, as it will frame which actors will be integrated in the network. This will be done in the course of the mapping. Nonetheless, the importance of this issue must be already emphasised as it is strategically crucial. Indeed, if we narrow too much our analysis or bound it too tightly geographically, we may completely miss a vital actor, or take unnecessary risks in terms of response, for example in the case of the war against the Islamic State (H Lavoix, “At War against the Islamic State – From Syria to the Region“, RTAS, 2 Nov 2015). It is thus very important not to introduce such potential biases from the start in our analyses.

These exemplify the discussions that are generated by the mapping of networks or models, and which are crucial in allowing us mitigating our biases and developing proper dynamical models. To further ease and favour the discussions, the main interrogations identified are right now integrated in the text for each node in the interactive graph.

As far as the information added to a node is concerned, we shall need to find a balance between giving too much information, which will clutter understanding, and not enough, which would impoverish it. This balance will also need to be practically replicable by analysts, and at once conducive of better analyses, while being neither too time-consuming nor too constraining to achieve practically.

Adding three crucial cluster nodes

While working on Warren’s initial network, it appeared that three cluster nodes or large categories were still missing, which were thus added: “climate change”, “systemic level (International Relations) change”, and  “monetary and currency system”.

Climate change (click to access node on interactive network) must be added considering first a growing body of evidence according to which it impacts security situations (e.g. JM Valantin, “Environment, Climate Change, War and State“, RTAS, 16 March 2015; Fred Pearce, “MidEast Water Wars: In Iraq, a Battle for Control of water”, Yale 360°, 25 August 2014; Colin P. Kelley et al. “Climate change in the Fertile Crescent and implications of the recent Syrian drought“, PNAS, January 30, 2015). yemen_mrg_2015308 scSecond, recent events show that the two countries analysed are also highly likely to be directly or indirectly impacted by climate change and related extreme weather events. These events are the heat dome which settled over the region over the summer 2015 (e.g JM Valantin, “Climate Nightmare in the Middle East“, RTAS, 14 Sept 2015), and, starting end of October 2015, low pressure concentrating on the Persian Gulf and generating flash floods falling over Iran, Iraq and Saudi Arabia. as well as two cyclones, Chapala and Megh, which, within one week wreaked havoc on Yemen (e.g. BBC News, 9 Nov 2015;  “Deadly Floods and Storms Hit Iran, Iraq and Saudi Arabia“, 30 Oct 2015, and “Floods in Iran and Iraq Leave almost 70 Dead – Fear of Cholera Outbreak in Iraq“,  9 Nov 2015, Floodlist; Elena Ugrin, “Seasonal low brings heavy floods to Iran, Iraq and parts of Saudi Arabia“,, 30 Oct 2015).

From a methodological point of view, note how including those last extreme weather events nicely illustrates Warren’s earlier point regarding how scanning will be integrated in the map, while the references used can be accessed directly from the information pane for the node concerned in the interactive map.

By systemic level changes, we refer to all these changes taking place at the systemic level in world politics and which will impact state’s behaviour. We use here the third level of analysis as classically defined by Kenneth Waltz in Man, the State and War (1959 – here for a review). For example, the characteristics of the international system in terms of poles – unipolar, bipolar, multipolar – and the effect of each type on state’s behaviour is an explanation that is given at the systemic level. Part of the work of the English school of international relations theory, as exemplified by Hedley Bull The Anarchical Society, and focusing on international norms and their evolution is also located at the systemic level.

Monetary and currency system is added to our mapping as a full cluster node to be detailed, apart from global economics. Indeed, first, money and the implied power to mint money are directly related to the regalian power of a state. Second, we must consider the crucial importance of the dominating currency in the international monetary system, which, since 1945 at least, has been the US Dollar, but which might currently be questioned (“Dollar, US” in Routledge Encyclopedia of International Political Economy, R. J. Barry Jones, ed., 2001, 364-367; “Dominant and dangerous“, The Economist, 3 Oct 2015). This is all the more important in our case, considering the existence of the so-called petrodollars. It will thus be necessary to understand and map as well as possible these elements.

For now, we continued using colour coding according to the categories of nodes as initially defined by Warren.

Keeping in mind the tension outlined previously by Warren between a network that is readable enough visually yet detailed enough analytically, when we shall develop cluster nodes, we shall most probably create new interactive graphs for each cluster being developed. We would then have the initial map, quite similar to what we have now, which somehow represents the outline of the problem. We would then have sub-networks, one per each cluster node. Finally we would have a fully detailed map, integrating all networks, which would be the basis for further analytical and monitoring work and visually useful to those users who have started training in visual network analysis.

Integrating war and cooperation

Finally, it is crucial if we want to make sure we can foresee future dynamics that these very dynamics and processes are integrated from the start in our analysis.  Going back to the initial question, we need to understand it better, wondering notably about the dynamics we are trying to identify. We are thus, here, starting to further detail the original “outcome” node which was “Iran-SA: Conflict/Cooperation”.


Let us recall our foresight question: “Within which timeframe could we see full cooperation or, on the contrary, war occur between Saudi Arabia and Iran?”

One of the terms of our question is thus war, and if we want to be able to foresee properly war, then we need first to define what we mean by war. The best definition, for us, as it incorporates a reference to actors and is dynamic, is the one given by the famous German strategist, von Clausewitz:

“Each strives by physical force to compel the other to submit to his will: his first object is to throw his adversary, and thus to render him incapable of further resistance.

War therefore is an act of violence to compel our opponent to fulfil our will.” (Clausewitz, On War, 1832 [tr 1873], Book 1 – On the Nature of War – Chapter I What is War?)

Desert Storm 1991
A column of M-113 armored personnel carriers and other military vehicles of the Royal Saudi Land Force travels along a channel cleared of mines during Operation Desert Storm, 1991. By staff Sgt. Dean M. Fox [Public domain], via Wikimedia Commons.

We thus need to add, instead of war as a node, two (mirror) nodes: “SA starts forcefully to try compelling Iran to fulfil its will” and “Iran starts forcefully to try compelling SA to fulfil its will“. The happenstance of such an act, by either country, would depend from two large types of factors, intention and capabilities, to take a classical approach often used in intelligence analysis, which are thus linked with edges or arrows (see here for Iran on the interactive graph). As we are still in the first steps of elaboration of the mapping, we can, for now, consider that intention will depend upon the difficult to define yet practical notion of national interest, that we shall detail – and probably question – further in following posts (by clicking on, for example  Iran’s intentions on the interactive graph, one sees the children (descendant) and parent (antecedent) nodes displayed). Two new “sub-graphs” have thus been added to our network, one for Saudi Arabia, and one for Iran.


Before to close, we need to wonder about the other term of our potential outcome, cooperation. Cooperation, indeed, is much more than the simple absence of war. It is defined by the Oxford Dictionary as the “action or process of working together to the same end”. However difficult it may be to currently imagine Iran and Saudi Arabia cooperating, such changes over time are possible. Thinking in terms of cooperation will force us to think out of the box – which is crucial for good strategic foresight and warning – to imagine possible other ways through which Iran and Saudi Arabia could relate.

The Oxford dictionary’s definition is very interesting for us, because, by emphasising “the same end”, it uses a concept that we should relatively easily relate to Clausewitz’s idea of “will”. We thus here anticipate that we should be able to link the various corresponding nodes as we further identify underlying dynamics. We shall thus add a last node, for the time being, upon which we shall build in the coming posts “Iran and SA start working together to the same end” which will represent cooperation.

To anticipate, we shall be able to develop the network around this node to use Robert Axelrod’s seminal work The Evolution of Cooperation (1984), where he identifies conditions under which cooperation rather than competition may thrive, in the situation of anarchy that characterises the international system.

We now have a new interactive network, where nodes are clickable with all the information necessary being displayed in an information pane, where sub-networks surrounding each nodes can be displayed and accessed easily by clicking on a node. Meanwhile, the network – our model for understanding and then monitoring the future relations between Iran and Saudi Arabia – has been further developed by adding three new cluster nodes to be detailed. The very question at the center of the network has been further explained and detailed to make sure dynamics, thus the future, can be fully integrated.

Helene Lavoix, PhD Lond (International Relations), is the Director of The Red (Team) Analysis Society. She is specialised in strategic foresight and warning for national and international security issues.

Featured image: Air strike in Sana’a 11-5-2015 by Ibrahem Qasim CC BY-SA 4.0 , via Wikimedia Commons.

*Some users have reported that on Mac with Yosemite as OS, the display looked less nice than on the picture above. We are currently working on solving the problem assuming it is not linked to Yosemite many bugs. We shall also see if a celle phone version can be developed.