a list of high-value explanatory ideas, frameworks, models, and thinkers for general purpose analysis, ideation, and problem solving. feel free to import as assumed knowledge when helpful.
- Bayesian reasoning
- Start with what you think you know (priors), state a credence in (0,1), then update with new information. Repeat. I used to think Bayes was overhyped, but then I changed my mind.
- buffers
- Adding a buffer to a system doesn't just smooth out variation - it changes the system's fundamental properties. Decoupling producers from consumers, supply from demand, or generation from use transforms what's possible. Batteries are about to prove this at civilizational scale (like grain silos did before).
- Shannon / information theory
- Communication has a speed limit set by the channel. Redundancy, compression, and noise aren't just metaphors - they're the actual structure of how information moves between systems.
- Latour / actor-network theory
- Agency is spread across human and nonhuman actors. If you strip out the material, institutional, and technical mediators, you're leaving out most of the story.
- catastrophe theory
- Smooth, continuous changes in conditions can produce sudden jumps in outcomes - a cousin of phase transitions. The discontinuity is baked into the geometry of the system, not an anomaly. Thom
- attractor basins
- Systems settle into stable configurations. Knowing which basin you're in, and what it takes to escape it, matters more than analyzing the current position alone.
- rugged fitness landscapes
- Optimization surfaces are full of local peaks. Hill climbing alone gets you stuck; see annealing. The shape of the problem determines which search strategies actually work. Wright
- systems theory
- Behavior comes from the structure of feedback loops, stocks, flows, and delays - not from the intentions of individual parts. Leverage points are often counterintuitive.
- complex adaptive systems
- Agents following simple local rules generate macro-level patterns nobody designed. Emergence, adaptation, and path dependence are features, not bugs.
- self-organized criticality
- Many systems naturally drift toward a critical state where small nudges can cascade at any scale. The sandpile doesn't need tuning to produce avalanches. Bak
- deep time
- Thinking across geological and evolutionary timescales changes what counts as fast, slow, stable, or transient - see also pace layers. Most of what looks permanent is just slow-moving.
- Herbert Simon / satisficing
- Optimal solutions are often computationally intractable or just unknowable - especially on rugged landscapes. Bounded rationality means choosing "good enough" on purpose, not as a failure mode.
- Goodhart's law / proxy failure
- When a measure becomes a target, it stops being a good measure. But proxy cascade offers a corrective: higher-level proxies can constrain and correct lower-level proxy failure.
- pace layers
- Different layers of a system move at different speeds, and that's load-bearing. Fast layers innovate, slow layers stabilize. Problems come from trying to force them into the same tempo. Brand
- boundary objects
- Things that sit between communities, plastic enough to mean something different to each group but robust enough to hold the collaboration together. Maps, standards, prototypes. Star & Griesemer
- spin glass models
- Systems with competing, frustrated constraints that can't all be satisfied at once. There's no clean global optimum, just a sprawl of near-equivalent messy compromises. Parisi
- phase transitions
- At critical thresholds, more of the same input produces something qualitatively different. The interesting behavior lives right at the boundary between regimes.
- affordances
- The environment doesn't just exist passively - it offers possibilities for action. What you can do is a relationship between the agent and the world, not a property of either alone. Gibson
- coarse-graining / multiple realizability
- The right level of description throws away detail on purpose. Many different micro-configurations produce the same macro-behavior, and that's what makes higher-level explanations legitimate, not just convenient. Related: carving at the joints.
- carving nature at the joints / scientific realism
- Some category boundaries track real structure in the world and others are just conventions. Figuring out which is which is most of the work in getting a model right. Plato, via Quine
- Markov blankets
- The statistical boundary between a system and its environment. What a thing "is" can be defined by what it screens off. Identity as conditional independence. via Friston
- Wright's law
- Cost drops as a function of cumulative production, not time. Every doubling of units produced brings a consistent percentage reduction. Learning is baked into doing.
- umwelt
- Every organism inhabits its own perceptual world. The tick's universe has three relevant signals. Yours has more, but it's still a subset, not the whole picture. See also unicept. von Uexküll
- unicept
- A concept isn't a shared universal with a definition - it's a particular nest of accumulated information belonging to one person. Your "weight" and mine refer to whatever each of us has been collecting information about, not to some agreed-upon paradigm. A shared "concept" is really a fuzzy overlap of many unicepts. Millikan
- STPA (systems-theoretic process analysis)
- An analytical framework that models safety as a control problem rather than a component reliability problem. Start from hazards, map the control structure, find where constraints go unenforced.
- stock and flow
- Stocks accumulate, flows change them. Most confusion about system behavior comes from not tracking which is which. Bathtub dynamics are surprisingly hard to intuit. sensu Meadows, Forrester
- positive / negative feedback
- Negative feedback stabilizes, positive feedback amplifies. Most interesting system behavior comes from the interplay between the two, and knowing which one is dominant right now.
- desire paths
- The trail worn into the grass tells you where people actually walk, not where the planner wanted them to. Observed behavior is a more honest signal than stated intent or designed flow.
- annealing
- You need enough disorder early on to explore the space, then gradually tighten to settle into a good solution. Too much order too soon locks in bad configurations.
- path dependence / lock-in
- Early choices constrain later ones, often irreversibly. The current state of a system tells you as much about the sequence of past decisions as it does about present fitness. Arthur
- motivated reasoning / talking one's book
- People find reasons to believe what benefits them. Vested interests, entrenchment, and position-defending aren't separate from analysis - they shape which evidence gets taken seriously and which gets explained away. A failure mode of updating.
- Red Queen effect
- You have to keep running just to stay in place. In co-evolutionary dynamics, standing still means falling behind, because everything around you is adapting too.
- Landauer's principle
- Erasing a bit of information has a minimum thermodynamic cost. Information and energy are convertible - computation is physical, and abstraction is never actually free.
- dissipative structures
- Energy flow through a system organizes the system. Order doesn't just persist despite thermodynamics - in open systems far from equilibrium, it emerges because of it. Prigogine
- underdetermination
- Evidence never fully determines theory. Any experiment tests a whole web of assumptions at once, and multiple incompatible models can fit the same data. Theory choice involves more than just the evidence. Quine-Duhem
- research taste
- Some people develop good sense for generative theory. Since evidence underdetermines theory, theory does more than fit data - it influences future data investment. See also path dependence.
- 4E cognition
- Cognition is embodied, embedded, enacted, and extended - not a program running on brain hardware. Thinking happens across bodies, tools, and environments, not just inside skulls. The agent and its world aren't separable.
- interbeing
- Nothing exists independently - everything is made of non-self elements. The paper is the rain, the sun, the logger. A corrective to foundationalism's habit of looking for the one thing at the bottom. via Thich Nhat Hanh