Notable Quotes

Saturday, 27 July 2024

Science, Strategy and War: The Strategic Theory of John Boyd by Frans Osinga:

This quote emphasizes the human element in scientific research, where much of scientific work is based on guesswork and intuition. This challenges the notion of research is supposed to be purely objective and mechanical process of gathering and analyzing data.

Subsequently Polanyi criticized the ideal of objectivity, asserting that much of science stems from guesswork and intuition and that although, in theory, science is continually modifiable, in practice it doesn’t work like that.

‘The part played by new observations and experiment in the process of discovery is usually over-estimated’, he noted. It is not so much new facts that advance science but new interpretations of known facts, or the discovery of new mechanisms or systems that account for known facts. Moreover, advances often have the character of ‘gestalt’, as when people suddenly ‘see’ something that had been meaningless before.

His point was that scientists actually behave far more intuitively than they think, and that, rather than being absolutely neutral or disengaged in their research, they start with a conscience, a scientific conscience. This conscience operates in more than one way. It guides the scientist in choosing the path of discovery, but also guides him in accepting which results are ’true’ and which are not, or need further study.

This quote tangentially reminds me of the reductionist culture that is often pervasive in enterprises. It seems more common for consumers of a system to care about the novelty of isolated features (i.e. parts). What seems to rarely enter consciousness is the complex web of relationships between parts, and how the system determines the parts.

Shifting the attention from macroscopic objects to atoms and subatomic particles, nature does not show us any isolated building blocks, but rather appears as a complex web of relationship among the various parts of a unified whole. These relationships are expressed in quantum theory as probabilities, which are determined by the dynamics of the whole system. Whereas in classical mechanics the properties and behavior of the parts determine those of the whole, the situation is reversed in quantum mechanics. It is the whole that determines the parts.

This quote reminds me that it is impossible to have a divine point of view that will make all opponent activity visible (despite what vendors may intuitively imply). What we observe is not opponent activity itself, but their activity “exposed to our method of questioning”. Our ability to observe a particular opponent’s activity is proportional to how close our method of questioning aligns with how they perceive their own activity.

The irreducible plurality of perspectives on the same reality expresses the impossibility of a divine point of view from which the whole of reality is visible. It implied that reality studied by physics is a mental construct. Heisenberg noted that ‘what we observe is not nature itself, but nature exposed to our method of questioning’.

This quote reflects Boyd’s view of human nature:

The goal is to survive, and to survive on one’s own terms, or improve one’s capacity for independent action. Due to forced competition for limited resources, one is probably compelled to diminish adversary’s capacity for independent action, or deny him the opportunity to survive on his own terms, or make it impossible to survive at all.

Life is conflict, survival, and conquest.

This quote introduces “phase space” and “attractors” as a way to understand the complex behavior of chaotic systems and reveal order within what might otherwise seem like disorder. It would be interesting to understand how this can be quantitatively applied to human behavior at both individual and group levels.

To describe non-linear behavior of systems, the idea of ‘phase space’ was coined, which describes the range of positions a system can occupy. The problem with chaotic systems is that, unlike a clock’s pendulum (which will slowly return to a standstill), they never pass through the same point, i.e. the system never repeats itself, so that each cycle of a pendulum (to continue the example) covers a new region of phase space.

It will not be possible to predict which point in phase space the system will pass through at a certain time, but will be possible to map the phase space, for in spite of the seemingly erratic motion, the points in phase space are not randomly distributed. Together, they form a complex, highly organized pattern (aptly named attractor) which computers are able to visualize.

Sunday, 21 July 2024

Deep Learning With Python by François Chollet (2021):

This quote defines intelligence as sensitivity to abstract analogies:

The kaleidoscope hypothesis: our experience of the world seems to feature incredible complexity and never-ending novelty, but everything in this sea of complexity is similar to everything else. The number of unique atoms of meaning that you need to describe the universe you live in is relatively small, and everything around you is a recombination of these atoms. A few seeds, endless variation—much like what goes on inside a kaleidoscope, where a few glass beads are reflected by a system of mirrors to produce rich, seemingly ever-changing patterns.

Generalization power—intelligence—is the ability to mine your experience to identify these atoms of meaning that can seemingly be reused across many different situations. Once extracted, they’re called abstractions. Whenever you encounter a new situation, you make sense of it via your accumulated collection of abstractions. How do you identify reusable atoms of meaning? Simply by noticing when two things are similar—by noticing analogies. If something is repeated twice, then both instances must have a single origin, like in a kaleidoscope. Abstraction is the engine of intelligence, and analogy-making is the engine that produces abstraction.

In short, intelligence is literally sensitivity to abstract analogies, and that’s in fact all there is to it. If you have a high sensitivity to analogies, you will extract powerful abstractions from little experience, and you will be able to use these abstractions to operate in a maximally large area of future experience space. You will be maximally efficient in converting past experience into the ability to handle future novelty.

The UnCODE system: A neurocentric systems approach for classifying the goals and methods of Cognitive Warfare by Ask et. al. (2023):

This quote emphasizes the fractal nature of neural systems, which enable targeted cognitive interventions at multiple scales (from neurons to nations) to produce similar outcomes. Local changes at lower levels (e.g. neurons) have the potential to produce effects at higher levels (e.g. nations).

The neural systems approach explains why cognition can be targeted at different levels (e.g. at the level of the neuron, the individual, or a nation) to produce the same outcomes. In other words, it captures what aspects of cognition that are self-similar at different levels of analysis, and that constitute the mechanisms that can be targeted with advancing science and technology.

Neural systems capture this self-similarity in part due to their fractal nature. Neural systems are fractal in the sense that they start at the subcellular level for animals (sub-transistor/logic gates for machine intelligence) but can be combined at increased hierarchical complexity to constitute neural systems at the cellular, organismal, and up to super-organismal level such as nations.

For instance, a single neuron is a neural system because it consists of neural components that allows it to process inputs and produce outputs. Similarly, a social network is a neural system because it consists of humans that, by virtue of being made up of neural components, are themselves neural components. Consequently, social networks can process inputs and produce outputs (group behavior, culture) because they emerge from the combination of components that processes inputs and produces outputs.

In order to perturb a neural system (e.g. cause social incohesion in a society), there has to be some physical change at the lowest level of the neural system. In the case of a human social network, this change is at the level of single neurons in the brain of the humans that make up the social network. Because neural systems are fractal, local changes at the level of single neurons have the potential of permeating at increased levels of complexity up to the level of nations and even the global population. This is why CogWar goals can include sub-cellular effects as well as social-level effects. No matter what approach is used, the goal is always to influence activity at the lowest level of the neural system.