Marginal Cost

What this page is / is not
· This page is a “conditions preparation tool”: it works out for you which start criterion is missing, how to fill the gap and what to do first. It is not investment advice, and not business-decision advice; whether you make money also depends on Z4 (someone wants it) and Z5 (the channel) — the document says it in §9 in these words: “low marginal cost ≠ definitely profitable”.
· The thresholds come from an internal methodology document, “Economic Profit Algorithms · 07 Marginal Cost Algorithm v1.0” (domestic machine /opt/inventory/econ_algorithms/07-边际成本算法.md, 157 lines, version fixed 2026-09-14, read-only), with no external empirical testing at all. §1 of the document positions itself as concepts with provenance (marginal cost / economies of scale Marshall 1890 / information-goods cost structure Shapiro & Varian 1999 / learning curve Wright 1936), but those thresholds (20% ／ 5% ／ 80% ／ R ≥ 20 ／ 10%) are practical lines this methodology set for itself, not authoritative standards.

Evidence chain (the md document has no line numbers, so it is cited to the “section”)
· v = material + commission + (minutes ÷ 60) × hourly rate → from 07-边际成本算法.md §4 Step1
· R = F ÷ v (R ≥ 20 is treated as “replicable”), m = p − v, Q = F ÷ m → §4 Step2
· the four tiers (the two lines R and v/p) → §4 Step3
· Q ≤ 10% of the reachable audience → §4 Step4; expansion order Z5 → Z1 → Z6 → §4 Step5
· the six start criteria Z1–Z6 and their thresholds (v/p ≤ 20%, 10th copy ≥ 80%, ≥ 20 people, ≥ 1 channel, F ≤ 3 months of surplus) → §3
· the conditions preparation checklist and the “first step” text → §5.2 / §5.3

Unaligned / simplified (stated plainly)
· Unaligned ①: every row of the §4 Step3 table gives the range of R and the range of v/p at the same time, but it does not say “which one wins when the two columns do not line up”. This page implements it as “a row only counts as a hit when both columns line up” (R and v/p are mathematically independent, so not lining up is normal); cells that do not line up are marked straight away as “⚠ Not given in the document”, and it lists which row R alone falls into and which row v/p alone falls into. This is a gap in the document, not an omission by this page — this page does not pick a row for it.
· Simplified ①: §4 Step1 of the document gives only the way v is composed and has no door for “enter v directly”, so if one of the four components (material / commission / minutes / hourly rate) is missing, v cannot be computed — this page follows that convention and offers no “enter v only” shortcut.
· Simplified ②: Z1 “there is a replicable carrier” is a one-line judgement in the document; this page turned it into a “yes / no” dropdown; left blank it is treated as “not met”.
· Not done: §3 of the document mentions jumps such as “switch to No. 08 arbitrage / switch to No. 07”, and §8 mentions the links to machine seven's value chain / the octopus / No. 10 compound-interest curve / No. 11 pricing power; this page only links No. 09 (network effect · economic algorithm edition) and No. 11 (pricing power), and the other corresponding pages are not built yet.
· Simplified ③: every threshold comparison carries a 1e-9 floating-point tolerance — for example, with F=1000 and v=50, R is exactly 20 and v/p exactly 5%, which in binary may come out slightly above or slightly below; without the tolerance, an input “exactly on the line” would be judged into the neighbouring tier. The tolerance is only there to line up with the whole-number lines the document gives (20 / 5% / 20% / 50%), and it has not changed any threshold itself.
· Not confirmed: §7 of the document suggests “have an LLM distil your repeated actions into a template catalogue” — this page is pure front-end and is not connected to any model, so that enhancement was not done.

[Evidence section · where to look up every number]
① Source: internal methodology document 07-边际成本算法.md v1.0 (domestic machine /opt/inventory/econ_algorithms/07-边际成本算法.md, 157 lines, read-only): start criteria Z1–Z6 = §3; cost breakdown / R and Q / the four tiers / break-even quantity / expansion order = §4 Step1–Step5; preparation checklist and first step = §5.2 / §5.3. §1 of the document gives the concept provenance (Marshall 1890; Shapiro & Varian 1999, DOI 10.2307/1183273; Wright 1936, DOI 10.2514/8.155).
② Three categories
· [Verifiable] v = 材料 + 抽成 + 分钟数÷60×时薪, R = F÷v, m = p−v, Q = F÷m, the four tiers (🔵 R≥20 and v/p≤5% / 🟢 5~20 / 🟡 1~5 / 🔴 <1), Q ≤ 触达人群 10%, the expansion order Z5→Z1→Z6 — all of them line up with §4 Step1–Step5.
· [This page's display convention] The “← your input” row in the four-tier table is assembled by the page; and the 1e-9 floating-point tolerance that every threshold comparison carries is a page implementation (it does not change the thresholds themselves).
· [No source found] The numbers R≥20 / v/p≤5% / 20% / 50% / 10% / ≥80% / ≥20 people / ≤ 3 months of surplus have no external provenance in the md (§1 gives only the concept provenance, not the numeric provenance) → no source found (internal practical lines).
③ How to check it yourself: F=1000, material 50 (all other components 0) → v=50; unit price p=1000 → R = 1000/50 = 20, v/p = 5% → both columns hit row 1 → it should be judged 🔵 asset type. Then F=5000, material 600, p=400 → R=8.33 falls in row 2 while v/p=150% falls in row 4 → it should show “⚠ Not given in the document” (this page does not pick a row for the document).