From Commodity Fetishism to Digital Fetishism: Metrics, platforms, tokens, and the LLM investment cycle, mapped onto Marx and the NiCE framework

From Commodity Fetishism to Digital Fetishism Metrics, platforms, tokens, and the LLM investment cycle, mapped onto Marx and the NiCE framework R. D. Kitcey · Version 1.0.0 Overview Digital economies increasingly coordinate human activity through quantities that acquire practical authority: engagement scores, rankings, ratings, platform balances, and issuer-controlled tokens. This essay develops “digital fetishism” as an extension of Marx’s commodity-form analysis: social relations take the form of apparently autonomous quantities, and participants must organize their behavior around those quantities even when they understand their limitations. The central question concerns how a representation becomes a governing incentive and how its authority can be tested against the activity, human needs, and physical conditions it is supposed to serve. Argument and mapping criteria The paper accepts a mapping from Marx’s Capital, Volume I, to a digital phenomenon only when three conditions are specified: R (relation), formally independent activities are coordinated through a common form; A (authority), the form constrains behavior regardless of participants’ beliefs; and O (observable), a measurement or experimental design can distinguish the proposed effect from its absence. Nine mappings connect these criteria to engagement metrics, rankings and recommendation, algorithmic management, platform payment forms, game currencies, gambling and prediction markets, commodified needs, the material infrastructure concealed by “the cloud,” and investment in large language models. Seven digital forms are developed alongside a separate discussion of the value-versus-rent debate. Formal model: substance, gaming, and metric inflation A minimal incentive model divides effort between substance q and gaming g. With reward coefficient a > 0, effort-cost coefficient c > 0, and relative gaming yield k ≥ 0, the coordinating metric and utility are: m = q + kgU(q, g) = a(q + kg) − (c/2)(q² + g²)q* = a/c; g* = ka/cGaming share = g*/(q* + g*) = k/(1 + k)m* = a(1 + k²)/cMetric inflation = m*/q* = 1 + k² These are model-derived results under the stated assumptions. A separate fixed-budget, linear-reward variant assumes q + g = T and predicts full allocation to gaming when k > 1, full allocation to substance when k < 1, and indifference at k = 1. The paper connects the mechanism to Goodhart’s law, Campbell’s law, and metric reactivity. Applications and the NiCE framework Applications include creator platforms, gig work, crowdfunding, software and product obsolescence, game currencies, randomized rewards, housing, healthcare, food, and education. The Nature–Consciousness–Environment (NiCE) framework adds an account of habituation, shifting baselines, conditioned incentives, and feedback among institutional conditions, experience, and biological and ecological consequences. The essay explicitly examines disagreements with Marxian analysis over value and rent, the subject of exploitation, ownership and control, reform capture, and the possibility of metric reactivity outside commercial production. Worked example: the LLM investment cycle Using the supplied 2026 data vintage, the paper reports that US private fixed investment in information processing equipment and software reached 5.35% of nominal GDP in Q3 2026, compared with 4.46% in Q4 2000. It reports combined latest-year capital expenditure of US$408.905 billion for Alphabet, Microsoft, Amazon, and Meta, up 72.3%, alongside total-company revenue of US$1,652.565 billion, up 15.2%. The ratio of the increase in capital expenditure to the increase in revenue rises from 0.47 to 0.79. Investment share: sₜ = 100 Iₜ/YₜCapital intensity: K/RMarginal ratio: m = ΔK/ΔRAnnual repayment requirement: A = Kr/[1 − (1 + r)^(−L)]Zero-return case: A = K/L when r = 0 At an assumed asset life L = 6 years and required return r = 8%, the annual repayment requirement is approximately US$88.5 billion. Asset lives, returns, contribution margins, house edges, and gaming yields are scenario inputs. The national-accounts category and company totals include activity beyond AI; the four-firm aggregate combines Microsoft’s fiscal year with the other firms’ calendar years. These measures motivate a conditional investment-risk argument rather than an estimate of AI-only returns. The energy discussion distinguishes weather-attributed load records from the separately unresolved contribution of data centers. Predictions and methods Eleven predictions specify measurements and potential falsifiers involving metric visibility, value/rent business models, independent verification, noncommercial platforms, habituation, footprint disclosure, household cost pressures, medical crowdfunding, randomized rewards, issuer discretion, and investment repayment. The package contains LaTeX source, a BibTeX bibliography, PowerShell scripts, archived source material, derived CSV files, generated tables, and a provenance manifest. The source records 53 quotation matches, 16 web-phrase matches, and 41 DOI resolutions. These are the supplied package’s recorded verification outputs; phrase matching and DOI resolution establish neither causal validity nor the truth of every associated claim. The FRED input is a documented page-read snapshot; the original run did not execute the optional FRED API branch. Research contribution and archival contents The contribution is an operational account of how symbolic coordination can redirect activity away from the substantive ends it represents, together with an explicit model, a worked investment example, and proposed empirical tests. This archival edition preserves the supplied version 1.0.0 source and adds a comprehensive Markdown transcription, Unicode description, formatted HTML description, field-by-field deposit guidance, keywords, citation files, and packaging validation notes. The manuscript’s numerical vintage and argument are retained.