So Much Ado About... Feedback Loops

So much talk about systems, so little attempt to understand them

In my world, from my use of game theory to my use of complexity theory, I think primarily about feedback loops. When I am talking about political ideologies, like when I am asked to explain the difference between Left Communism and the Comintern in 1928 versus the Left Opposition in the USSR in 1926, I don’t use this language. It’s too formal and seemingly bloodless, but I think about it in this language. Each step moves a system back into a dynamic equilibrium (a relative homeostasis that is not a closed loop which moves into entropic disorder), static equilibram (a homeostasis that is a closed loop and will start to move into entropic disorder), or into outright disequilibrium (where there theloops are broken and the system cannot reach a new status quo).

So let’s look at an old-school definition of the phenomenon:

Simple causal reasoning about a feedback system is difficult because the first system influences the second and second system influences the first, leading to a circular argument. This makes reasoning based upon cause and effect tricky, and it is necessary to analyze the system as a whole. As provided by Webster, feedback in business is the transmission of evaluative or corrective information about an action, event, or process to the original or controlling source.

— Karl Johan Åström and Richard M.Murray, Feedback Systems: An Introduction for Scientists and Engineers¹

Feedback loops are hard to parse in terms of causation. Feedback may be causal, it may be reactive, or it may even be purely correlational. Did the movement of United Left Opposition in the USSR, despite the disagreements between Trotsky’s Left Opposition and Zinoviev and Kamenev’s new opposition, harden Stalin and Buhkarin’s positions and solidify the continuation of the New Economic Policy and “Socialism in One Country” in the Bolshevik party or is the result of hardening positions? Is the liquidation of the Left Opposition within the Warsaw Pact why Stalin moved against Bukharin’s faction, or would it have happened regardless? These aren’t even counterfactuals, but they are damn near impossible to know despite the fact nothing about it was purely contingent.

What we can say is that by the time of the murder of Kirov in 1932, a set of positive feedback loops kicked in on both sides, which made reconciliation impossible. Kirov’s murder was contingent, as a singular event, it is where the lack of variables ironically means the most possibility for a different outcome. If Kirov had not been murdered, would the purges have happened?

You can see the problems with this, and yet we can map these out. The factions in the Bolshevik party in Russia were in conflict with one another, which was a stimulus that led to various events made disagreements about the possibility of re-establishment of capitalism in the USSR if socialism could not be spread beyond the Warsaw pact move from a hypothetical theoretical question to a hypothetical strategic question to a reason for purges and bannings through out the world. Furthermore, these systems aren’t discreet—in what ways it change the status quo of labor relations within North America and Europe?

It’s difficult to answer and when you think about this in turns of pure feedback, you can map the responses but it isn’t all that predictive to the outcomes until the dust is settled and NKVD has filled the grounds of Medvedev Forest with the blood of all the Oryol prisoners, including a ton of old Bolsheviks.

So, to get to the issues about why I think feedloops are helpful, even if they can’t establish discrete “cause-and-effect” chains, you need to understand what a feedback loop maps, particularly in a social system. Social systems are living, and thus not entirely closed systems from the perspective of physics. Social systems seem to have a tendency towards dynamic equilibrium—a state where opposed processes are happening and changes in these processes bring in new energy and stimulus, but negative feedback loops keep the system balanced as opposed processes do not change the status quo. This is an analogy from chemistry, but it applies to these systems.

Dynamic equilibrium relies on negative feedback. From the standpoint of the system, although not components of the system, the stimulus causes a change in the system, the system responds in the form of output, which counteracts or incorporates that stimulus, thus maintaining a dynamic equilibrium. The third party emerges, it siphons off voters from one of the main parties, but as long as is in a negative feedback loop, the third party leads to its ideas being incorporated into the area of politics where it was siphoning off voters from a major party, or it is nuetralized by agitprop from the mainstream party, or laws are changed to limit the effect. In the second half of the 20th century, all three of these things were done when Reform party under Ross Perot mounted the first real threat to the party duopoly of the United States since both the Progressive (Bullmoose) Party and the Socialist Party stood as real challengers in 1918.

Sometimes, however, the center cannot hold. This moves feedback loops into positive feedback loops. Positive feedback loops, from the standpoint of the system itself, are negative. This is when the product of a reaction (output) increases the response from the stimulus. Instead of stabilizing the system, the response destabilizes it, and again, this is not entirely contingent or stochastic. This increases momentum or the speed of change in a system.

Ironically, if a negative feedback loops work too well, it can make a system start to stagnate. Sometimes, one positive feedback spiral can move a system just enough into a new direction that it changes the situation. In both cases, relatively predictable determination in aggregate can lead to unpredictable results. A black swan event can emerge even if that event is not random. It is often a tipping point into either entropic disorder from stagnation or over-momentum when a system cannot adjust, and starts to fall into cascade failure. The latter will seem completely random in the moment, but after it is over, the movements and patterns start to be clear and even seem obvious. The former can lead to a social situation where “nothing ever changes,” which makes the people maintaining the system socially non-responsive, and institutions will be unable to maintain their mission since nothing is testing the competency of those maintaining the system. Ironically, both entropic disorder can turn into over-momentum where “a week will seem like a thousand years” or a major change in momentum made lead to elites and experts trenching where they become more and more resistant to criticism and stagnate.

Both can lead to a cascade failure and even system collapse, and not just in social systems either. A lot of ideological posturing seeks to avoid looking at these dynamics, and since an individual exists in multiple systems, it can propose change in one system in one’s life while ignoring the way stagnates others systems.

Another issue, and this is the one that makes causation hard to establish, is the tendency for feedback loops to generate mutually constitutive feedback loops. This describes a cyclical relationship where two or more aspects of a system influence each other, creating a self-reinforcing effect. The way, for example, the woke left and the anti-woke left move a lot of their rhetoric, not in response to the movement of general society, but to counter each other. The dynamic feeds into itself, and will most certainly turn into a positive feedback loop—the woke left moving increasingly towards symbolic capital and the anti-woke left increasingly making common cause with the anti-woke right, until both systems have altered so much in response to each other that they don’t resemble the groups they started out as. This can generate positive feedback loops where both parties have their ideologies co-opted or destroyed or lose touch with general society, and fail to often guidance to general society since they are responding to each other almost solely. Often, one can even see these patterns play out over time in mutual response to each other until an aphoria is reached, and the substance of the debate shifts—but perhaps the structure doesn’t. Think about politically correct battles on college campuses in the 1990s being interrupted by the suppression and then repression of elements of American society after 9-11, only to shift in the “call-out culture” debates from 2007-2014, and then shift again into the woke wars of the late aught-teens.

These are mutually constitutive feedback loops, but they can’t transcend the structure of the debate. It is new forces that try to utilize this feedback towards new purposes—an administration using attacks on a new framework like D.E.I. to go after socially accepted and established frameworks from the 1964 Civil Rights Act, which breaks the feedback loop as a new factor has emerged.

Now, these are just very abstract examples with some very specific illustrative anecdotes, and one must be careful of any abstract framework that attempts to explain everything. However, one feels like things are getting out of control from situations where nothing ever seemed to change priorly: ask yourself, did the mutually constitutive feedback loop finally break in response to positive feedback throwing a system into disequilibrium? Did the agents of these loops ignore that they were using more and more social energy on something that merely responded to another stimulus until it generated overcomplexity for parts of the system outside of a mutually constitutive loop?

This won’t solve all your problems of analysis, but once you see these systems dynamics, you can start to ask better questions.

1. Karl Johan Åström; Richard M. Murray (2008). "§1.1: What is feedback?". Feedback Systems: An Introduction for Scientists and Engineers. Princeton University Press. p. 1.