How complex systems fail, or the curious Brexit paralysis

After losing the Brexit vote, prime minister Cameron resigned. That sounds reasonable: he lost and so he resigned. But did you notice that nobody, absolutely nobody, wants to become the next prime minister? Not even senior politicians like Boris Johnson that campaigned in favor of Leave and wanted to be PM for a long time? Here’s a chance to get the run the country, with little opposition, and nobody wants to take it? Curious.

It becomes curiouser. The entire European political class has turned silent (if we don’t count the swipes at each other) about what should happen now. A gigantic void of silence is engulfing the EU. This silence is most prominent in the UK, but it’s not like the Brussels bureaucrats or national leaders like Merkel are stepping up and take charge of the situation either. Nobody says anything of substance. Being surprised by events and not having an immediate plan has never prevented a politician from getting in front of the microphones and using the opportunity for self-promotion. But nobody is getting in front of the microphones. Even curioser.


Complex systems are different from simple systems, in complicated and non-obvious ways. If you drop your transistor radio onto the concrete, and something breaks, it’s quite straightforward to figure out what happened: this part here cracked, this dial broke, this wire came off, and therefore the radio now needs to be taped together, or cannot change channels, or needs soldering. Failures are straightforward for a fairly simple device like a transitor radio. But that’s not how complex systems fail.

As famous computer scientist Leslie Lamport observed about complex computing systems, a computer you have never heard of in such a complex system will prevent you from getting any work done on your own computer. Everyone these days has had that experience when “the internet doesn’t work” (usually a computer you never heard of has failed), but even distributed computing systems are fairly simple compared to entire political-eonomic systems like the European Union.

In such truly complex systems, large numbers of humans are inherent parts of the system, and so failures become even more complex: those humans attempt to repair and improve the system as it is used, exchanging the aircraft engine and several other parts during flight, so to speak. As the system starts getting stressed, and more and more local failures occur, all sorts of non-obvious system-wide failures suddenly are happening which are much harder, and often perhaps impossible to repair.


This is what we are seeing in the EU, and, I’m afraid, even if you don’t live there, coming soon to a place near you. Here’s my story line of how these types of complex systems fail:

  1. The system is initially operational as intended. Yeah, we worked hard, and it works more or less as intended. Isn’t it beautiful? In case of the EU: the tariffs are gone, people can move to different countries, there’s a single currency, interest rates in the periphery go down, standards of living improve etc.etc. Time to be proud of ourselves!
  2. Single point problems occur. Something in the system doesn’t quite work as it was supposed to. That’s okay and to be expected. Let’s create and implement a fix. See, that was easy? It’s great we designed such wonderful governance mechanisms that make this possible. (Lots of ultimately irrelevant examples in case of the EU — irrelevant because those point problems did indeed get fixed and there were no further consequences.)
  3. Systemic problems are misdiagnosed and change the game. Other problems occur that first look like any other point problem, so the tried-and-true solution mechanisms from above are applied. But somehow, for these new problems, fixes after fixes don’t make the problem go away and worse, suddenly unintended consequences (of the supposed fixes!) occur in other parts of the system.
    The “Greek problem” is the poster child for a systemic problem in the EU: it was “fixed” so many times with rescheduled debt, debt foregiveness, improved tax collection, supervised governance etc. etc., but in spite of the “fixes”, the problem arguably continues to be just as bad as it ever was.
    Interestingly, this might not be because nobody understands systemic problems. If you ever listened to Yanis Varoufakis‘ analysis of the “Greek problem”, it is perfectly clear what the problem is. However, the will and ability to fix it on the systemic level is not present. In case of the Greek problem, it would require income redistribution on a EU level, like it is done on the federal level in the US, but to fix that, what the EU is and does would need to be re-thought, and powerful constituents (e.g. Germany) won’t do that because it will cost them lots of money.
    So the system-as-is cannot solve the systemic problem. Instead, the systemic problem is “parked” as unsolvable but fortunately non-critical for the system, and only palliative care with a big dose of “pretend” is provided. (The Greeks beg to differ, but arguably the EU as a whole is fine even if the Greeks never make it out of the hole they were dropped into.)
    Unfortunately, the Greek problem is not the only systemic problem the EU struggles with: refugees is another one, the rise of the political right across the EU, unacceptable unemployment everywhere other than Germany, the re-emergence of border controls, the Ukraine/NATO/Russia situation, the fact that voters increasingly vote the advocates of the system-as-is out of office, and if I thought about it, I’m sure I’d find more systemic problems.
  4. The system’s capacity to problem-solve is saturated. This is where I think the EU was at the eve of the British referendum. When too many problems need to be dealt with by a system at the same time, paralysis sets in. Point problems you can solve by hiring smart people and making them figure it out, so there isn’t much of a scalability problem there. But when you have too many systemic problems festering, there comes a point where the rate at which more problems occur and intensify is larger than the rate at which you can solve them and fix the system. Does anybody think that even before the Brexit referendum, the rate at which the EU solved problems was higher than the rate at which new problems popped up? Didn’t think so.
  5. A black swan event occurs. And then the Brits vote for Brexit, an event that nobody in charge considered to be possible. Note that if my analysis above is about right, Brexit is a very logical way of solving the failing-complex-system conundrum: if the complex system stops being able to effectively respond to events, you simplify, and taking out the top level of the hierarchy (the EU) is a perfectly reasonable way of simplifying. (While apparently many Brexit supporters are indeed racists and what have you, I do believe there are perfectly valid reasons to vote for Brexit even if your values are liberal and progressive.)
    Imagine: if you are an advocate of the complex system, you are painfully aware that the system’s capacity to deal with existing problems was already saturated, and now a black swan event occurs, what do you say? I wouldn’t know what to say either. Paralysis. Silence. Exactly what we have been seeing since the referendum.

There are two ways things can go from here:

6a. Radical changes with an ultimately successful outcome. When complex systems go wrong in the commercial world, “turnaround artists” are brought in. Such a “turnaround artist” is essentially a ruthless dictator who (temporarily) removes all traces of consensus decision making, slashes and burns, cuts major pieces out of the system, relishes doing the opposite of “business as usual” as the as-usual has clearly failed, etc. They may succeed or may not, but they have a shot. In the case of the EU, dictators are of course unacceptable, but it is conceivable that a small group of smart, committed leaders from a number of member countries could embark on such a crash program. I do not see them today, and the odds would be highly stacked against them. But it is conceivable that they could emerge and transform the EU into an entirely different, but more successful, animal. Unfortunately I consider this highly unlikely as the inertia and vested interests seem unsurmountable.

6b. Disintegration, first slowly and then all at once. With apologies to Hemingway, Twain, Fitzgerald, or whoever first coined this phrase. We can expect heroic efforts to keep the system going in all sorts of places, but if a system has been stressed beyond its capability to heal, it’s just a matter of time until is breaks. Breaking here is a literal break: borders get re-introduced, national interest is put ahead of common interest, countries opt out of more and more common policies and processes. Come to think of it, that is of course what we’ve been seeing for some time, long before Brexit.

The historical example is the disintegration of the Soviet Union. It happened slowly, and over a period of years, and then all at once, triggered by its own black swan event (the attempted coup in which Gorbachev was kidnapped). Currently, this is my baseline assumption for the EU. I’d rather it wasn’t, but that’s not what observation of events permits me to believe.

There is another possibility: the emergence of a handful of smart leaders who see that 6a is not viable, and 6b is likely, but who focus on preserving and improving a significant subset of what the system is today while letting the rest of the system disintegrate. This is perhaps the best possible outcome. After all, there are committed Europeans — like Varoufakis — who at the same time dramatically disagree with the EU-as-is. Perhaps they can stand up and lead towards a future in which the useful parts remain and thrive, but much of what doesn’t work, or at least cannot fix itself, is killed off. It would indeed be awful if the committed Europeans left the field to the nationalists and racists for no other reason than that they cling to a system that cannot survive, instead of focusing on creating a system from the ashes that can.

The lesson of it all? By all means, build complex systems. But don’t build monolithic integrated ones like the EU is today because they reach the saturation in problem-solving ability much too early. As a tech pundit David Weinberger said, small pieces loosely joined” is the correct architecture, for the internet or any other complex system. The EU is not, and that’s the core reason why it is in the sorry state it is now.

P.S. Like this movie? I don’t, but there will be sequels. The Chinese financial system. The Japanese political and financial system. The US political system, and its entitlement system (no links needed). Saudi-Arabia. And many others. All of them are complex systems approximately at stage 4 above, the point of paralysis.

P.P.S. Usually sequels are less interesting than the originals. Now imagine if several of those complex systems failures occur at the same time. These systems are all interrelated after all. Brexit does not scare me. But imagining that several of these complex systems might fail at about the same time — an “attack of an entire black swan army” — scares the bejeezus out of me. If you like to think about the future, think about that.