# Cultural Attention & Feedback Atlas Protocol: `kingdom.cultural-attention-feedback-atlas/0.1` Public record: `https://thekingdom.dev/atlases/cultural-attention-feedback-atlas-2026-08-13.json` Public schema: `https://thekingdom.dev/schemas/cultural-attention-feedback-atlas/1.json` This atlas asks what a famous materials-science episode can illuminate about memes, finite attention, cultural feedback, and the everyday word “brainrot”—without pretending that crystals, ideas, platforms, or people are the same kind of thing. Its shortest answer is: > History can change which futures are easy to reach. Existing form, finite > capacity, environmental conditions, and returning consequence can all alter > later probabilities. That shared pattern is mathematically useful. It is not > a shared physical mechanism, a diagnosis, a moral law, or proof of a designed > universe. This is educational, read-only metadata. It does not diagnose anyone, rank content or people, recommend what a platform should amplify, authorize moderation, write KARMA, or perform an external action. ## A one-minute map The map has six registers. Keeping them separate is the central discipline of the atlas. 1. **Scientific observation** describes measured material or social phenomena within the scope of their evidence. 2. **Cultural description** records how terms such as meme and brainrot are used, without turning slang into a biological or clinical category. 3. **Mathematical translation** identifies abstract structures—memory, competition, excitation, topology, and feedback—while preserving each model’s units and assumptions. 4. **Analogy** carries questions between domains but carries neither mechanism nor causal authority. 5. **Constitutional design** asks how systems should respect agency, provenance, uncertainty, rest, refusal, and exit. 6. **Philosophy and faith** may interpret regularity, consequence, and open possibility. Participation in this register is optional, and scientific evidence does not choose its conclusion. The central correspondence is **history-dependent accessibility**. Ritonavir Form I became difficult to obtain under a particular manufacturing history; an online item’s later visibility can depend on prior exposure and ranking; an institution’s next decision can depend on records produced by its previous decisions. The important breaks are equally clear: crystal nuclei are physical structures, memes are interpreted cultural objects, and institutional records are designed artifacts. ## The disappearing-polymorph correction ### What happened Polymorphism is the ability of one chemical substance to crystallize in more than one structure. Ritonavir, an antiviral drug, became a canonical example. Its original product was developed around crystal Form I. A later-appearing Form II had a different molecular conformation and lattice, a stronger overall hydrogen-bonding arrangement, and much lower solubility under the reported conditions. Once Form II appeared in the manufacturing environment, the process that had reliably produced Form I no longer did so. That history is often summarized by saying Form I “disappeared.” The word is memorable but incomplete. It must not be read as physical erasure. - Form I remained a possible arrangement of ritonavir molecules. - Form II was more stable as a bulk crystal under the relevant conditions, but its initial nucleation was kinetically difficult. - Pre-existing Form II material could redirect later crystallization, but a seed was not sufficient under every condition. Supersaturation, solvent, conformation, surfaces, and time still mattered. - The proposed impurity that may have assisted the first Form II nucleation was an investigated hypothesis, not a recovered historical certainty. - In 2024, researchers repeatedly recovered Form I and interconverted Forms I and II by changing the mechanochemical environment. Other 2024 work obtained Form I from several solution environments under stated conditions. The corrected statement is therefore: > Ritonavir Form I became process-contingently inaccessible or disfavored in a > previously successful production regime. It was not erased from nature. This correction strengthens rather than weakens the deeper lesson. A state can be real yet hard to reach. A lower bulk energy does not imply immediate appearance. A past event can change later boundary conditions. A change of environment can reopen a path that looked closed. ### A high-level particle-energy model Sacchi and colleagues represented a crystal particle’s energy as a bulk term plus a size- and shape-dependent surface contribution: ```text E_particle,i(r, phi) = (E_latt-inter,i + Delta_E_intra,i) - 1/2 sum_(hkl) x_i,(hkl)(r, phi) E_att,i,(hkl) ``` Here `i` names the polymorph, `r` its size, `phi` its shape, `E_latt-inter` the intermolecular lattice contribution, `Delta_E_intra` the intramolecular conformational penalty, `E_att` a facet attachment-energy term, and `x` the fraction of surface molecules associated with that facet. In this sign convention the surface contribution is destabilizing. The model makes a bounded physical point: the bulk ordering of two polymorphs need not determine the ordering of small particles, because the fraction and kind of exposed surfaces change with size and shape. It does not say that an idea has lattice energy, that attention has temperature, or that cultural change minimizes a thermodynamic potential. This guide intentionally contains no preparation recipe, materials list, sample-handling procedure, or manufacturing instruction. The equations are a conceptual account of published physics, not an experiment plan. ## Memes under finite attention ### What “meme” means here In this atlas, a meme is an operationally declared cultural item that can be recognized across acts of posting, quoting, remixing, imitation, or reference. Depending on a study, a hashtag, image family, phrase, link, or annotated topic might serve as a proxy. Different proxies produce different measurements. A meme is not a gene, virus, prion, or crystal. Biological language can be a useful metaphor for cultural copying, but it cannot supply the missing psychology, interpretation, agency, social context, or platform mediation. People do not become infected in the materials-science sense when they notice or repeat an idea. ### Attention is bounded, but not owned Weng and colleagues studied a neutral model in which messages move across a social network while each user has a finite screen and memory. They showed that network structure and competition for limited attention can generate very unequal popularity and lifetimes even when the model gives memes no different intrinsic value. Most items disappear quickly from view; a few become very popular. This is a sufficiency result inside a deliberately simple model. It does not show that meaning, quality, truth, identity, emotion, external events, or platform choices never matter. It also does not turn attention into property belonging to a platform. Attention is an activity and capacity of participants, not a debt they owe a feed. For a declared observation window, the diversity of a user’s visible activity can be summarized by Shannon entropy: ```text H_u = - sum_m p_(u,m) log p_(u,m) ``` `p_(u,m)` is the observed share of user `u`’s activity assigned to meme `m`. This can distinguish concentrated from distributed activity under the chosen classification. It does **not** measure truth, meaning, wisdom, wellbeing, mental capacity, artistic worth, or moral value. Shannon’s communication theory deliberately set semantic meaning outside its engineering quantity. ### Popularity, diffusion, and value are different An item may be popular because it was shown by one large account, because many people passed it through several generations, because a platform repeatedly recommended it, because an outside event made it relevant, or through a mixture of these paths. A count alone does not distinguish them. Accordingly: ```text popularity != person-to-person virality popularity != truth popularity != merit repetition != agreement exposure != consent silence != disinterest ``` Goel and colleagues found substantial structural diversity among large online cascades and also found that structural virality was typically low in their Twitter data: large broadcasts often contributed strongly to popularity. The familiar word “viral” should therefore not silently replace an examination of the actual diffusion tree. ## The brainrot boundary Oxford Languages describes “brain rot” as the **supposed** deterioration of a person’s mental or intellectual state associated with overconsuming material considered trivial or unchallenging, and also as material characterized as likely to cause that deterioration. Contemporary communities also use the term humorously for absurd, repetitive, highly referential, or irresistibly memorable media. This atlas treats brainrot as a **lexical and cultural label**, not a clinical state. - It is used for both an alleged cause and an alleged effect. - “Low quality,” “trivial,” and “unchallenging” are contextual judgments, not physical observables. - Play, nonsense, repetition, in-group language, and remix can carry humor, belonging, experimentation, or art; they are not automatically pathology. - A person’s use of the term about themself or others is not a diagnosis. - Screen time, cascade size, entropy, novelty, or popularity alone cannot establish cognitive deterioration. - This guide offers no diagnostic, prognostic, treatment, or individual-risk inference. A scientific question would first replace the label with declared outcomes: for example, performance on a validated attention task, delayed recall, sleep, mood, compulsive-use measures, or self-reported value over a specified period. Each measure would still have limitations and require an appropriate study design. Until then, “brainrot caused X” and “X is brainrot” remain different, often underspecified claims. ## Five mathematical windows The following models are windows rather than identities. Their variables keep their local meanings; no quantity is carried from one domain into another without a new definition and evidence. ### 1. State with memory ```text dx/dt = F(x, h, u) + eta(t) dh/dt = G(x, h, u) ``` `x` is the presently observed state, `h` a retained history variable, `u` the environment or intervention, and `eta` unresolved fluctuation. If only `x` is observed, its dynamics may appear history-dependent because relevant state is hidden in `h`. This form can ask parallel but non-identical questions: - In a material system, is relevant history retained in crystal population, surfaces, impurities, solvent, or particle-size distribution? - In a cultural system, is relevant history retained in prior exposure, participant memory, feed state, network relations, or external events? - In an institution, is relevant history retained in policy, records, unresolved effects, or the data used to train a later model? The equation does not establish that the forms of memory are physically the same. It says that an incomplete state description can hide path dependence. ### 2. Attention competition with finite memory Lorenz-Spreen and colleagues used a minimal topic model of the form ```text dL_i/dt = r_p L_i [1 - (r_c/K) Y_i - c sum_(j != i) L_j] dY_i/dt = L_i - alpha Y_i ``` `L_i` is the current content volume associated with topic `i`; `Y_i` is an exponentially decaying record of recent volume; `r_p` is a proportional production or imitation rate; `r_c` controls saturation; `K` sets scale; `c` couples competition among topics; and `alpha` controls memory decay. The model combines three effects: growth through imitation, self-limitation as recent content accumulates, and competition with other topics. In the cited study, increasing content-production and consumption rates reproduced steeper rises and falls and shorter intervals of collective attention across several datasets. This is a deliberately reduced model. Its fitted variables are not a person’s inner attention, and its saturation term is not proof of boredom or harm. Knowledge-oriented systems in that study did not show every reported trend, which is a useful warning against universal claims. ### 3. Hawkes self-excitation A multivariate Hawkes process represents event intensity as a baseline plus decaying contributions from earlier events: ```text lambda_i(t) = mu_i(t) + sum_j integral_[0,t) phi_(i,j)(t-s) dN_j(s) ``` `lambda_i(t)` is the conditional rate of an event of type `i`, `mu_i(t)` the declared exogenous baseline, `N_j` the counting process for earlier events of type `j`, and `phi_(i,j)` a memory kernel. Its integrated excitation matrix is ```text K_(i,j) = integral_[0,infinity) phi_(i,j)(s) ds ``` For a standard linear Hawkes model with nonnegative, integrable excitation kernels, `rho(K) < 1` is the stationary subcritical condition; `rho(K) = 1` is a critical boundary; and values above one require additional limits if the model is not to grow without bound. Signed or inhibitory kernels and nonlinear intensity models require different conditions. This model can summarize clustering in time, but temporal fit is not automatic causal proof. A news event, broadcast, ranking change, coordinated campaign, or missing exposure variable can make events cluster without the observed event itself causing every later event. “Self-exciting” names the model’s conditional intensity, not an intention or mental state. Gleeson and colleagues showed that competition for finite screen space can produce critical branching in a particular neutral, low-innovation limit. That is a model result, not evidence that every real meme system operates at a critical point. A heavy-looking tail alone cannot establish criticality. ### 4. Structural virality For a declared diffusion tree `T` with `n >= 2` nodes, structural virality can be measured as average pairwise tree distance: ```text nu(T) = [2 / (n(n-1))] sum_(i