5D Library › Supercritical Viral Cascades, Information-Gap Hazard Rates and Multi-Tier Paywall Optimization · 中文
Preface: The Arithmetic of Growth and Charging
Note
This is not business, legal or investment advice; please read this first. The book studies the mathematics behind referral growth, paywalls and tiered pricing: branching processes, hazard rates, signal detection and price discrimination. Its numbers describe models, not any real website; all data are generated by the programs printed in the book. Countries and platforms have specific rules on “share to unlock”, multi-level commissions, advertising for fortune-telling, and personalised pricing; Chapters 3 and 8 list several that need checking, but the book is not legal advice.
The growth engine of a paid reading website usually has three parts: a sharing mechanism by which users bring in others, a free hook that makes users want to know more, and a ladder of paid tiers from cheap to dear. Behind each part stands a slogan: “with K above 1 growth explodes exponentially and acquisition costs fall to zero”; “state the pain point bluntly, add an identity label, and users will share”; “a five-tier paywall takes several times what one price does”. This book writes each slogan back out as the model it rests on, runs the model, and sees where the slogan holds and where it does not.
Like the other books of 5D Library, it tries to refute before it sets bounds, and it reports negative results as they are — several of which run directly against the slogans.
What the book contains
Chapter 1 writes the two channels, “invite friends to unlock free” and “pay and earn a commission”, as a Galton-Watson branching process and finds the critical boundary of K. Chapter 2 adds a finite audience and a clustered social network, and computes the acquisition cost the slogan leaves out. Chapter 3 treats the incentive compatibility of commissions and share-to-unlock, and how fake accounts distort growth metrics. Chapter 4 writes the “information gap” as a purchase hazard and uses signal detection theory to ask where the feeling of “it’s so accurate” comes from. Chapter 5 compares, under Pareto willingness to pay, one price, menus and the ceiling of perfect price discrimination. Chapter 6 treats bundles, decoys and paying per question. Chapter 7 reads the chapters “Rubbing” and “Flying and Clamping” of the Guiguzi and the “State Reserves” chapter of the Guanzi alongside the models. Chapter 8 gives a four-step check, a compliance checklist and a pre-registered paywall test.
The book borders on several others in 5D Library and recalls them briefly only where they overlap: Book 6 treated the economics of attention and information cascades; Book 47 choice architecture and price anchoring; Book 48 acquisition funnels and attribution; Book 50 mechanism design and incentive compatibility; Book 58 the unit economics of subscriptions.
Claims corrected after checking
- The share rate caps K under share-to-unlock. A sharer brings at most new users, so however high the acceptance rate, K cannot exceed plus the pay channel’s contribution. With 30 % or 50 % of users willing to share, no threshold and no acceptance rate takes K above 1; with 80 % sharing and a threshold of 2, the acceptance rate must reach 0.124 (p01).
- Supercritical does not mean explosive. At K = 1.15, 0.729 of single seeds still die out within a few generations and the median cascade is 3 users; explosion belongs only to the few seeds that survive the first steps (p02).
- Acquisition costs do not fall to zero. In a finite audience K = 1.15 reaches only 0.255 of it, and the free readings given to everyone who never pays are the main cost: 3.22 per paying user, 0.325 of the 9.9 price (p03).
- Real social networks give a lower K than the tree estimate. When friends know each other, the K measured in the first generations is 0.934 against a tree estimate of 1.35, and the cascade fills one pocket and stops (p04).
- Fake accounts make growth metrics lie. When fake accounts are cheap, the platform measures K at 1.992 while the real K is only 0.065, and every payer refers themselves, turning a 50 % commission into a 50 % discount (p06).
- The feeling of accuracy needs no information. A generic reading identical for everyone is judged 0.838 accurate, close to an informed reading (0.939), yet scores only the chance level of 0.333 in a swap test; adding an “identity label” raises sharing from 0.583 to 0.745 (p08).
- For one good, no menu beats one price. At a Pareto tail index of 1.5 the ceiling of perfect discrimination is 3.00 times the revenue of one price, but the best two-option menu earns 2.961, below the 3.000 of one price (p09).
- A multi-tier paywall earns more because it sells more goods. Pricing four different products separately raises the share of the value ceiling captured from 0.150 for one product to 0.339; a bundle adds a little more only when customers differ in which product they want (p10, p11).
- The effect of a decoy tier is a parameter to be measured. Without an attraction effect, a decoy raises revenue only because some people buy the dominated option (p12).
- Testing a paywall needs far more data than intuition suggests. Because one referral tree can hold hundreds of users, seeing a difference of 0.10 in net revenue per seed needs 166,788 referral trees per arm (p16).
How to read the boxes
The boxes separate published evidence, runnable Python calculations (16 programs, standard library only, fixed random seeds, each actually run at build time with its real output printed beneath it), classical sources, notes and the author’s hypotheses; every hypothesis states what result would show it to be wrong.
Disclosure
The author uses the works of 5D Library, including this book, as the methodological basis of his community courses, the 5D Ascension Cards (Dimensional Shift Card Decks) and the Yafu AI knowledge base, and develops and runs paid online reading software and services built on traditional divinatory arts, including share-to-unlock, referral commission and multi-tier payment features; the models of this book are the mathematical basis of those features, so he has a direct commercial interest in its subject. The book promotes no software, course, card deck or service, uses no user data from the author’s businesses, and accepts no sponsorship. It is not business, legal or investment advice, and its pricing and growth models are not a recommendation of any particular charging scheme. The test in Chapter 8 must be run and analysed with pre-registered code by people independent of the pricing decision, and its results will be published whether positive or negative.
Fan, Y. (2026). Supercritical Viral Cascades, Information-Gap Hazard Rates and Multi-Tier Paywall Optimization: Galton-Watson Referral Thresholds, Share-to-Unlock Incentives and Second-Degree Price Discrimination [超临界病毒级联、信息缺口风险率与多层付费墙优化] (Version 1.0.0). 5D Library. https://doi.org/10.5281/zenodo.24090174 · Licensed under Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International (CC BY-NC-ND 4.0).