Press "Enter" to skip to content

Tag: manipulation

The way of the connoisseur observer, or, how to deal with propaganda

The Vision of Tondale – by a follower of Hieronymus Bosch, c. 1485 [fragment]

Preface: It took me way longer than expected to finish this essay. Apologies, dear reader! I had to cut out a lot, and it still is both too long and too short. If you are in a hurry, feel free to scroll down to the end and focus on the modes and strategies of propaganda.


Why is it important to think about propaganda? Because at a macro level propaganda is noise, and in times of turbulence there is nothing more valuable than the ability to see through the noise.

Nassim Taleb once said that the facts are true, but the news is fake. I always thought this a rather optimistically naive view of reality, full of faith in the inviolate good nature of facts. Facts are far more devious and complicated than Taleb gives them credit for. In a previous post I talked about what it means to recognize trends, how not to be confused with facts, even the “true” kind, and how to go about examining your assumptions. Here, I will consider how even if the facts are fake, and the news is fake, their effects can teach us a lot about what might be true.

In my experience, most people are quite confident they can recognize propaganda and, what is more, consider themselves more or less inoculated against it. Of course, this is the ideal environment for propaganda to thrive in, as most people willingly make the mistake of defending their opinions as if they were essential elements of their identity, and would rather suffer the consequences of their mistakes than admit they were wrong. Everyone has encountered people who offhandedly cancel large chunks of data with ‘that’s X propaganda’, where X is the designated enemy.

I recently witnessed the following conversation between a millennial (M) and a boomer (B).

M: What do you think about [something politician X said] about the war in Ukraine?

B: Why do you ask me that?

M: I think it’s important to raise awareness.

B: What’s so important about it? You are so brainwashed! Where did you hear this, the internet?

M: Yes, I saw it on YouTube.

B: Ah! That’s the problem with you young people. You are all so brainwashed with your internet. You should watch TV!

M: TV?

B: Yes, the truth is on the TV!

Of course, the problem is not just with the internet or the TV, and it’s nowhere near as simple as the cartoon stencil battle of fake vs true news. To begin with, propaganda is a consistent and enduring process targeting perception, and so to understand propaganda and how to defend against it you have to understand perception.

Perception and action

Contrary to the meme that we are now bombarded with more information than ever before, we humans have always drifted in an ocean of data. Everything known to us through perception is pure data, this is how our minds encounter the world. Reading a newspaper, listening to a bird, smelling crushed pine needles, feeling the wind on your face, or all of the above, these are all just sense-derived data inputs which our mind uses to dynamically build a perception frame of reality.

The white noise chaos of the real is comprised of infinite data streams out of which our minds frame the incomparably poorer picture of reality we can perceive. We sense an infinitesimally small sliver of the real, and for a good reason. We need to not only perceive, but also operate in this reality. We need to be able to make choices, predict potential outcomes, and act based on them in a fast changing environment. Given the physical limitations of how much brain processing power you can cram into a body evolved to survive in a scarce and hostile environment, a limited perception frame makes it easier to quickly eliminate non-essential data, make choices and act.

The limited frame works as part of our perception system even when our actions are misaligned with what was necessary to do, because we can quickly observe the results of our actions and correct them. Having less initial data in a complex environment speeds up and improves the decision feedback loop, given that the loop is iterative and we can access real data. This point is very important for what follows, so I will explain it in more detail.

Having the correct data on whether a mushroom is edible or poisonous can lead to a dramatic variance in the results of your decisions and actions regarding said mushroom. Obviously, there is a high premium on correct data if wrong data leads to death. Having the correct data in a rapidly changing environment is like having a shortcut, or a cheat sheet. In fact, that is exactly how our mind stores and processes data repeatedly proven as correct – as shortcut blocks [schema]. They can be very simple – red mushrooms are bad, and quite complex – if the government puts price controls on diesel better buy as much diesel as possible.

A mind operating based on such shortcuts has tremendous competitive advantage as it can act very fast and dedicate the freed processing power to other tasks. This is why every long-lasting and functioning culture has a long list of taboos and cultural norms derived from many centuries of experience. They are all just mental shortcuts packaging long data – decision – action – consequences loops into a simple cheat.

Where things get really interesting is in those cases where we have the wrong data, resulting in the wrong assumptions, decisions, and actions. Why? Because presuming you survived your wrong assumption-decision-action chain (congratulations!), you now have a choice:

1] You can decide to stop assuming, deciding and acting altogether. [Many such cases!]

2] You can ignore the entire episode and keep on as if it didn’t happen. [A crowd favorite]

3] You can decide to change your assumption-decision-action chain. [The rare choice]

The first two choices are always good value when observed from a safe distance, and can also be quite educational for the connoisseur observer, but let’s focus on the rare choice. You thought you are making the correct assumption-decision-action chain but it ended up being misaligned with what you expected as a result, and through some deductive reasoning you established that you’ve made a mistake. Something has to be changed.

Where do you start if you want to change your assumption-decision-action chain? Correct, you have to go back to the beginning of the chain and examine your data. Sure, the mistake may have been in your assumptions, or the decisions you took, or even in your execution, but since you already made the rare choice you might as well assume that the data you started with is wrong. This means you have to examine your environment again, and gather new data.

Interlude: the connoisseur observer

As an aside, since you like making rare choices you might also be a connoisseur observer, in which case there are many scenarios where you can observe when others make the wrong assumptions, decisions and actions based on the wrong data. As a connoisseur observer you can learn from that as if the experience was your own. What is more, often the connoisseur observer likes to speculatively imagine what having the wrong data, and therefore the wrong assumption-decision-action chain, would mean in a given scenario, and collects these imaginary speculative scenarios for fun and future reference.

The problem with data: what if all the facts are fake?

In any case, as long as you can access your environment directly and gather fresh new data we can say that you have access to real data. As long as you can keep examining your environment in order to refresh your assumption-decision-action we can say that your decision feedback loop is iterative.

What happens when you don’t have iterative access to real data? Quite simply, this means you can’t correct any wrong schema shortcuts you have in your mind. The schema cheat codes might have been true and tested at some point in the past, but are not true anymore. If you are inclined to make either of the first two popular choices above – stop bothering with decisions and delegate them to authority figures instead, or ignore your mistakes – it is likely you won’t even notice the change until the effects of the misalignment become unavoidable. If you are a connoisseur observer however, you still have the option of finding real data in the actions of others, even if you can’t collect data directly. You can also revise your assumptions by speculating on various data scenarios that are not yet true but might be. All fairly straightforward.

Where it gets really interesting is when you have iterative access, but the data you have access to is itself corrupted. This is a counter-intuitive situation, because we are evolutionary wired to trust our senses. You see stuff happen, and therefore it must be true. You saw it! Our narrow perception frame may be a competitive advantage in a dynamic environment, but it is also a bottleneck that can be targeted and exploited. If someone is consistently fed corrupted data over a meaningful period of time, most of their mental shortcuts will be aligned with the corrupted data and misaligned with external conditions. Moreover, they will be convinced their assumption-decision-action chain is correct because of the iterative access to data they think is real. Simply put:

Manipulating the data inputs will always result in control of the perception frames built from those inputs.

In other words, even if you took option 3, the rare choice, and tried very hard to examine your assumptions and get new data from your environment, if the data itself is consistently corrupted you won’t be able to make the necessary course corrections. Even if you make corrections they will be the wrong ones, since your starting data is wrong. Furthermore, you won’t have any way of knowing this is happening, unless, as a connoisseur observer, you are consistently entertaining various outlandishly speculative data scenarios and comparing their output, and the actions of those around you, to what you perceive.

If you’ve read carefully so far you ought to have noticed that everything I’ve discussed up to this point is at the individual scale. Scale matters, because humans are social animals and like to compare and cohere with each other, drawing cues about the surrounding reality from the behaviors of those around. The problem is that while the phenomenon of the connoisseur observer exists only at the individual level and cannot be scaled up, modern propaganda is a phenomenon of mass scale, and its effects only improve as it scales up further.

Scale, the masses, and the current thing

Our ancestors knew well the power of injecting targeted data into someone’s perception frame, hence why “thou shalt not bear false witness”. Even so, for most of known history Lincoln’s saying held true:

You can fool all the people some of the time and some of the people all the time, but you cannot fool all the people all the time. Until recently, it was physically impossible to manipulate data inputs at scale large enough to affect entire populations. You could generate a completely synthetic picture of reality and drive the population of a given area to hysteria – as in the Salem witch trials or periodic pogroms against this or that minority – but it could only be done locally. Things changed dramatically with the industrial revolution, mass literacy, and the arrival of broadcast media.

The industrial revolution needed workers who can be disciplined to operate complex machinery, follow commands, and perform repetitive tasks as if they themselves were machine extensions. To achieve this, mass schooling was introduced in the 19th century to supply these workers, first in Prussia and then everywhere else. Gradually, the new paradigm raised hundreds of millions of people regimented to perceive and act similarly, obey authority [teacher-factory manager-political leader], and read abstract symbols encoding data. As everyone who has gone through the full cycle of schooling knows, you need to raise your hand in a silent plea for the teacher’s permission to speak. Mass schooling’s first and primary lesson – authority has to be obeyed and individual agency delegated to it.

The masses emerged, and with them the glories and wonders of the current thing and the long 20th century. The ability to target the perception of the masses at scale aligned just in time with the arrival of the first global current thing – WWI – known at the time as “the war to end all wars”. Broadcast media – first newspapers, then radio and television – gave the masses an unending stream of regimented data [it’s called a program for a reason] targeting the primary communication senses and already framed as coming from an authority. The masses were raised and trained to align with and follow whatever the authorities say is the current thing. The mass scale effect amplified every message, as everyone around seemed to be in agreement. All these people agreeing on this current thing can’t all be wrong, can they? Suddenly, you could fool not some, but most of the people all the time.

The regimented mind

The first person to think systemically about perception, and what can be done with it given the power of manipulating data inputs at scale, was Edward Bernays [Sigmund Freud’s nephew], known as the inventor of propaganda and public relations. This is how he describes propaganda – a term he coined – in his 1928 book [emphasis mine]:

Modern propaganda is a consistent, enduring effort to create or shape events to influence the relations of the public to an enterprise, idea or group.

Propaganda is creating circumstances and pictures in the minds of millions.

In its sum total propaganda is regimenting the public mind every bit as much as an army regiments the bodies of its soldiers.

Notice that according to this definition propaganda is all about the creation of entirely synthetic realities [events, circumstances, pictures], which, in turn, will generate entirely synthetic and controlled perception frames. As I pointed out above, if the initial data is synthetic so will be the perception frames built from this data. Again, manipulating the data inputs by “creating or shaping events” will always result in control of the perception frames built from those inputs.

The main issue here is that Bernays described the creation of synthetic data in the minds of millions at once. Suddenly, it is not a localized minority that finds itself temporarily misaligned from reality, but the majority or entirety of a given population all at the same time. A completely synthetic event is generated across all major media, everyone sees it, and the media keeps returning to it [iterative access]. The scale guarantees the ‘everyone agrees’ mass effect. It doesn’t matter if the event really happened in some actual physical form, or if it ‘happened’ but in a fully synthetic setting, or if it didn’t happen at all. What matters is that the masses saw it happen, and more importantly, saw each other see it happen.

To achieve this level of regimentation with minimum friction, good propaganda messaging is kept simple and coherent with little to no internal contradictions. After all, the aim of the process is to manipulate and regiment the perception of the masses at scale, and to therefore influence and to control their future actions informed by that shared perception frame. The frame has to be internally coherent. The ideal propaganda content is visceral and polarizing, without doubt or uncertainty. A well designed propaganda process leaves no place for distinctions or qualifications, everything is binary and stencil-clear. An important side effect of the mass-deployment of simple and internally coherent perception frames is that it now becomes much easier to detect individuals who think differently.

Interlude: the connoisseur observer

A careful reader would immediately realize that it is much harder to be a connoisseur observer when surrounded by a population enacting a synthetic perception frame. Due to the mass scale effect there are little to no spaces left unaffected by the synthetic frame one can retreat to. The connoisseur observer has to not only focus on maintaining iterative access to real data but also on camouflaging their actions in such a way so as not to reveal they operate with a non-synthetic perception frame. In effect, in such a scenario the connoisseur observer has to maintain an internal non-synthetic perception frame coupled with an external synthetic pseudo-frame.


What do we have so far in terms of propaganda characteristics? It creates synthetic realities, exploiting our narrow perception frame and the importance of schema. It works best at mass scale, the grander the better. It is internally coherent, visceral, simple and polarizing. It has a regimenting effect, generating predictable and controllable lockstep action chains in the masses. These are all useful, but I can unpack them further into a very crude but actionable schema of modes and strategies for recognizing the process on first encounter.


No matter how complex the various propaganda techniques, at its root the process can be described as having two primary modes, with the complexity emerging from the way the process oscillates between them. The primary modes of propaganda can be roughly described as mobilizing and integrating.

Mobilizing propaganda

Mobilizing propaganda [MP] is the most basic form of the process. It aims to agitate its subjects towards a certain perception frame by shocking, provoking and agitating. It works by appealing to basic instincts such as fear, or basic emotions such as love and hatred. It always simplifies reality to stencil binary statements. Due to its crudeness it is most effective over the short term, but is also the most intensive form of propaganda. Most people fall victim to it due to its brutal intensity. Every time you see reality framed along the lines of ‘X sank our innocent passenger ship!’ [think Lusitania], ‘Y is killing babies’ [think first Gulf war], or ‘the children are drowning – we must do something!’ [think Europe refugee wave], you are dealing with MP. You recognize it best by observing your own emotions, and, if you are a connoisseur observer, the emotions and reactions of others.

Integrating propaganda

Integrating propaganda [IP] on the other hand aims to sedate and co-opt its subjects towards a perception frame. It works by appealing to our socializing instincts, the desire to belong and be a part of something bigger than us, to participate in something good and meaningful. Where MP is intense and emotionally harrowing, IP is mellow and positive. It works through simple mantras such as ‘the right thing’, ‘one of us’, ‘yes we can’. While MP aims to shock into a polarized frame generating a course of action, IP aims to reinforce an idea, object, or action positioned within a frame of belonging. You recognize it best by looking for the central element of the frame generating the feeling of belonging. Whatever is positioned in the center of the frame is the element being sold through sedative integration.

The simplest example of these modes is footage of soldiers in the middle of violent combat, followed by soldiers lovingly embracing their loved ones. A well designed campaign oscillates fluidly between the MP and IP modes, first raising the emotional agitation and then reinforcing it through sedative integration.


Either of the P modes can be positioned in the minds of the masses through various strategies. The strategies can be, again very roughly, distilled into two primary types, with the variety emerging from the oscillation between them. The two primary strategies of propaganda can be described as vertical and horizontal.

Vertical propaganda

Vertical propaganda [VP] appeals to the views of a hierarchy or an authority figure – ‘experts have decided’, ‘a source in the government said’, ‘scientists agree’, ‘celebrities take a stance’, ‘the Party points the way’, ‘most doctors smoke Camels’. The VP strategy is an exploit of the fact the masses are already trained by mass schooling to obey and delegate agency to authority. This is why VP is the easiest strategy to deploy in conjunction with either of the MP or IP modes. Notice that corporate propaganda [advertising] uses this exploit, in conjunction with the IP mode, to regiment brand loyalty. The brand is an authority and a frame one can belong to. This is how we end up with the phenomenon of millions of people professing anti-capitalist views while simultaneously acting as fanatical foot soldiers for the same few corporate brands. VP is easy to recognize, as it has to invariably appeal to an authority figure or a hierarchy. That is why more sophisticated P strategies work horizontally.

Horizontal propaganda

Horizontal propaganda [HP], as the name suggests, appeals to the views of one’s peers rather than an authority or a hierarchy. It works by generating the appearance of a consensus – ‘everyone agrees’, ‘my friends told me’, ‘people saw it’. While VP is a tried and tested strategy for broadcast media, HP became a truly powerful weapon only with the arrival of social media. The deluge of content on social media establishes a radical horizontal battle space for all content, where every piece of data competes for limited user attention. Most of these data comes from others ‘just like you’, not from authority figures. This is why the primary filter for quality of content on social media switches from authority to authenticity. The more ‘real’ the person, the more valuable their content. This is why HP thrives on exploiting the ‘authentic person’ point of view. Now, you can mask any synthetic data as coming from ‘an authentic person just like you’. Unlike in the VP strategy, here the propaganda message is delivered as if it was coming from the point of view of someone really there. HP can be very confusing to isolate at first, but it can be recognized by observing whether it comes packaged with an MP or IP mode. Non-synthetic data will rarely if ever try to agitate or sedate on its own.

Similarly to the deployment of the primary propaganda modes, a well designed campaign oscillates fluidly between the VP and HP strategies, appealing in turn to the authentic figure from the masses and the authority figure from higher up in the social hierarchy, Think of the use of influencers and celebrities to sell the current thing.

This is how the two primary propaganda modes and strategies, and their combinations, align together:

Propaganda: the mobilizing [agitation] and integrating [sedation] modes, aligned with the vertical [authority] and horizontal [peers] strategies.

It’s a fun exercise to observe various data streams and position them along this framework. If they can’t fit anywhere chances are they are not propaganda.

Finale: the connoisseur observer

What can an individual connoisseur observer do in the midst of all of this? Observe, analyze, compare, speculate. Generate a synthetic external pseudo-frame where necessary. Repeat. This is the way.