Covid-19 coronavirus is bound to have significant impact on global supply chains as it impacts all sectors of the economy. See graphs from the Spectator for illustration. It is also somewhat unprecedented as it in many area both supply and demand are impacted. Both our ability to consume and produce has in many areas been significantly impacted by many countries adopting long periods of lock-down. Covid-19 is therefore a systemic shock to the economy, and a systemic shock to supply chains.
Systemic Impact on Supply Chains
The danger with supply chains is that Covid-19 significantly disrupts what are now frequently called supply chain networks. Where the term network recognises the complex inbound and outbound interdependencies between organisations. Supply chain networks are difficult to fully map and understand. Couple to this a move in the economy post WWII towards lean supply chains. Where there are low levels of redundancy in the supply chain, and resources (in all senses of the word) are delivered just in time. Disruption in one part of the network can quickly spread (cascade) to other areas. This could cause the network to fail in unpredictable ways. Where the ability of an organisation to produce something being halted by it being unable to acquire an essential input. Materials, parts, service, resources in general.
Supply Chain Deadlock
The danger with a large scale systemic shocks is that they sufficiently disrupt something like a supply chains to the point where it can no longer function. In the case of supply chains and Covid-19, one possibility is that they become deadlocked. The term deadlock is perhaps most frequently used to describe a situation where a series of interdependent processes all stop because they are all waiting for each other to do something. Process A is waiting for process B, B is waiting for process C, and C is waiting for process A. A situation that computer programmers have long had to deal with an mitigate.
This situation of deadlock could arise in lean supply chains. The operations of one organisation may be halted due to it being unable to acquire some essential resource for its operations. The halting of this organisation’s operations could have effects elsewhere in the supply chain. The difficulty is that this might not be a simple, easily mapped, supply chain. Similar to that shown above. Instead it might be a complex network of interdependent relationships, and therefore understanding how to release the deadlock might be extremely difficult.
If Covid-19 causes deadlock, releasing this deadlock in global supply chains will require international coordination. For example, it might require organisations to temporarily produce resources that they normally wouldn’t. Or resources might have to be moved between organisations. Both to allow processes in other organisations to get underway. There will need to be some way to both map resource requirements, and move them. This will need to be done internationally. Perhaps through the production of an international register of needs are resources.
My new paper, “Total Systemic Failure?”, is out (this link should get you a free copy for a limited time). I wanted to write something that was more a big picture look at how the world is working. Or perhaps how it isn’t working. I think there is a problem that the world is stuck arguing about whether climate change exists when it is a least possible that multiple global-systems are failing. I also wanted to write something that is very clearly about complexity theory and systems analysis.
So what is systemic failure? The paper goes through how complex systems theory describes how the world works. It then moves onto the idea that if a system is put under enough pressure, and this starts to affect the relationships between the parts of the system, then that system could collapse. This might either be a change in the nature of the system, so that the global system behaviour changing significantly. Or it might be that the system collapses completely.
We don’t have particularly good methods for understanding if a system is likely to collapse. How close it is, or if it is in the process of failing? We don’t know what we need to know. If we did then we might be better able to understand which systems are likely to fail, and perhaps what we might do to change that.
Total Systemic Failure
So what is different about total systemic failure? Here we are asking the question of if a number of systems start to fail, will this result in all systems failing? Systems are connected, and the degree to which they are connected and the significance of these connections is difficult to understand. Therefore, could the failure of one system precipitate the failure of another system? If this happens could it cause a cascade of failures? We do not understand how individual systems failure, we are even further away from understanding how a system-of-systems might fail.
What Can We Do?
We need to work out what state we are in, perhaps there are things we can do. Perhaps there is an opportunity to build a connected, distributed, sensor network that can provide data on global systems? It will not be easy to build this network, or analyse the data. Artificial intelligence could help, perhaps we can build AIs that can help develop the sensor network and also analyse the results. A global distributed sensor network. The AI could learn about what data is needed and what interventions could be made.
We need to start thinking more about the big picture. Otherwise we are going to find ourselves in a real mess with little hope of getting ourselves out of it.
I have processed more of the Daniel Morgan data, and thus have an updated network of the data. Below is a visualisation of the data produced by extracting the network structure from Neo4J using R and iGraph, then saving the network as a gexf file and importing into Gephi. The network is more complete but also has edge labels.
The released Panama data comes in the form of a Neo4J database, or the files that you can make one with, seems to me a little tricky to do much with. There is no detail beyond attributes of the different entities, so that limits us to looking at the relationships alone and it is hard to judge the significance of the relationships without the context… that said its a fun data set to play with.
I decided to draw out some graphs of how things are connected via other things. Below is one from Officers connected to other Officers via *something* else, generated via R using iGraph from the Neo4J data set. This produces a few clusters containing a relatively small number of nodes connected to others. The query that produces the graph is, “MATCH (n:Officers)-[:`officer of`]->(o)<-[:`officer of`]-(m:Officers) WHERE NOT id(n)=id(m) AND id(n)<id(m) RETURN n.name AS Officer1, m.name AS Officer2, count(o) AS Weight”
After listening to the Daniel Morgan podcast, Untold, I became really interested in the murder investigation. To help me follow it I started building a network of all the key people, organisations, and events in the case. The networks this produces can be seen here,and you can keep up-to-date with the progress on the network here.
The story is a compelling one, I suggest you either listen to the podcast or read the book. Very briefly it looks into the murder of Daniel Morgan, and the subsequent investigations into the murder and the police handling of the murder. The book builds a compelling story of decades of struggle by the Morgan family to get justice, and the difficultly they have had in discovering the truth.
The network is not complete, at the time of writing I have only put in the ‘easy’ bits. The network stores objects as the nodes, so people, companies, organisations. The lines, or edges, store the relationship between the objects, e.g. Alistair Morgan is ‘brother_of’ Daniel Morgan. The visualisation is produced using Alchemy, and the data is stored in Neo4J. I intend to continue to develop the network further, and the visualisation which needs things like edge labels. Once the network is more complete it would be interesting to see if there is any useful analysis that can be done on the network. It would also be interesting to expand the data to include other related and interesting cases. Such as the Stephen Lawrence murder, and the Leveson Inquiry will likely form a part of Algorithmic Indexing in the future.
The people over at The International Consortium of Investigative Journalists have updated the released panama data. Its not clear to me if that is more data than they had already released, or that this time it is a ready made Neo4J database. They provide two versions of the database, Windows and Mac. Its easy to get it to work in Linux, just copy the graph.db file from out of the archive into the databases directory of your Neo4J install.
I made a quick query to look for officers with the same address. Seems there some, it would need something more sophisticated to did any deeper.
MATCH (n:Officer)–(a:Address)–(m:Officer) RETURN n,a,m LIMIT 25