5D Library › Supercritical Viral Cascades, Information-Gap Hazard Rates and Multi-Tier Paywall Optimization · 中文
Chapter 7. “Rub It with Its Own Kind”, “Lift Then Clamp” and “Gather When Light”
Why read the Guiguzi and the Guanzi here
The Guiguzi is a pre-Qin text of the strategists on persuasion and gauging other minds; the “light and heavy” chapters of the Guanzi are China’s earliest systematic writing on prices and fiscal policy. The first is about how to make people respond, the second about how to direct goods by price — matching exactly the two halves of this book: referral growth and pricing. This chapter reads a few passages alongside the models and marks clearly where reading becomes interpretation. The texts are read to understand their structure, not to recommend techniques for manipulating others.
The texts
Source text
Guiguzi, “Rubbing” (Mo)
摩者,揣之術也;內符者,揣之主也。用之有道,其道必隱。微摩之以其所欲,測而探之,內符必應;其所應也,必有為之。 “Rubbing is the art of gauging; the inner tally is the master of gauging. There is a way to use it, and the way must be hidden. Rub lightly with what the other desires, probe and test, and the inner tally will respond; and what responds will surely act.”
故曰:摩之以其類,焉有不相應者;乃摩之以其欲,焉有不聽者。 “Hence it is said: rub it with its own kind, and what will fail to respond? Rub it with its desire, and what will fail to heed?”
Guiguzi, “Flying and Clamping” (Feiqian)
引鉤箝之辭,飛而箝之。鉤箝之語,其說辭也,乍同乍異。 “Draw out with hooking and clamping words: lift them up, then clamp them. Hooking-and-clamping speech, as persuasion, now agrees and now differs.”
Guanzi, “State Reserves” (Guoxu) (phrase)
斂之以輕,散之以重。 “Gather it in when it is light; release it when it is heavy.”
Note
Checking the text. The quotations from the two Guiguzi chapters have been checked against the full text in Project Gutenberg. “Gather it in when it is light; release it when it is heavy” comes from the “State Reserves” chapter of the Guanzi (its source was corrected when Book 56 was checked); only the phrase is quoted here, and its context has been checked against a standard punctuated edition.
“Rub it with its own kind” and like responding to like
“Rub it with its own kind, and what will fail to respond?” In the branching process of Chapter 1 this corresponds to homophily: invitations go more often to people like oneself, and people like oneself accept more readily. The fourteenth program imagines two communities, A with a share rate of 60 % and B with 20 %; each sharer sends 8 invitations, a share of them within their own community, accepted with probability 0.30 within a community and 0.10 across. The overall growth factor is the largest eigenvalue of the next-generation matrix.
Calculation
# Chapter 7. "Rub it with its own kind": homophily in a referral cascade.
# Two communities. In A 60 % of users share, in B 20 %. A sharer sends 8
# invitations, a share h of them to their own community; an invitation is
# accepted with probability 0.30 inside one's own community and 0.10
# across. The next-generation matrix has entries
# K[a][b] = share rate of a x 8 x (h if a = b else 1 - h)
# x (0.30 if a = b else 0.10)
# (new users in b per user in a). Its largest eigenvalue R is the growth
# factor per generation; its eigenvector gives where the new users land.
# R is compared with the K of each community taken alone.
import math
def matrix(h):
s = {"A": 0.6, "B": 0.2}
k = {}
for a in "AB":
for b in "AB":
same = a == b
k[a, b] = s[a] * 8 * (h if same else 1 - h) * (0.30 if same
else 0.10)
return k
def leading(k):
a, b = k["A", "A"], k["B", "A"] # column: new users in A
c, d = k["A", "B"], k["B", "B"]
tr, det = a + d, a * d - b * c
r = tr / 2 + math.sqrt(tr * tr / 4 - det)
va, vb = b, r - a # eigenvector of [[a, b], [c, d]]
if abs(va) + abs(vb) < 1e-12:
va, vb = 1.0, 0.0
return r, abs(va) / (abs(va) + abs(vb))
if __name__ == "__main__":
print(" own-share h K within A K within B R overall"
" new users in A")
for h in (0.5, 0.7, 0.9):
k = matrix(h)
r, share_a = leading(k)
print(" {:3.1f} {:5.3f} {:5.3f} {:5.3f}"
" {:5.3f}".format(h, k["A", "A"], k["B", "B"], r,
share_a))
print("\nReading: the more invitations stay inside one's own kind, the")
print("higher the overall growth factor - like responds to like - but")
print("the growth lives almost entirely in the community that shares.")
print("A cascade that is supercritical overall can be subcritical in")
print("every other community it touches.")
# own-share h K within A K within B R overall new users in A
# 0.5 0.720 0.240 0.757 0.683
# 0.7 1.008 0.336 1.018 0.826
# 0.9 1.296 0.432 1.297 0.947
#
# Reading: the more invitations stay inside one's own kind, the
# higher the overall growth factor - like responds to like - but
# the growth lives almost entirely in the community that shares.
# A cascade that is supercritical overall can be subcritical in
# every other community it touches.The larger , the higher the overall growth factor: from 0.757 to 1.297 — like does respond to like and does speed up referral. But the growth happens almost entirely in the community willing to share: at , 0.947 of new users come from A. Inside B, K stays below 1 throughout (0.432 at most). A cascade that is supercritical overall can be subcritical in every other community it touches: “like responding to like” makes referral faster, and also more confined.
“Lift then clamp” and the free hook
“Draw out with hooking and clamping words: lift them up, then clamp them” — first raise the other with praise (lift), then hold them fast (clamp).
Hypothesis · not established
The author’s interpretation (not established; a hypothesis about the text): the free hook and paywall of Chapter 4 have exactly the form of “lift then clamp”: first give content that makes people feel understood, then charge where the gap is widest. The calculations of Chapter 4 add two limits to this reading: the feeling of being lifted (perceived accuracy) needs no real information, and once the clamp is delayed, curiosity fades fast. Using this structure where people can be made to pay without real information is something this book explicitly does not recommend. What would show the reading to be inapt: if the commentators agree that “flying and clamping” applies only to persuading rulers and has nothing to do with exchange, the reading is only a modern analogy.
“Gather when light, release when heavy” and price
“Gather it in when it is light; release it when it is heavy” — take in goods when prices are low, put them out when prices are high. In the Guanzi this is a means for the state to steady prices and raise revenue.
Hypothesis · not established
The author’s interpretation (not established): the phrase is about adjusting prices over time to use the rise and fall of demand, which is a different thing from Chapter 5’s price discrimination, which adjusts prices across people. Reading the first as a form of intertemporal pricing — cutting prices to absorb slack demand and raising them when demand is strong — is this book’s interpretation, not the meaning of the text; the subject of the text is the state and its aim is to steady prices, not to maximise a website’s revenue.
Summary
“Rub it with its own kind” corresponds to homophily in referral: with invitations sent to one’s own kind, the overall growth factor rises from 0.757 to 1.297, but the growth is concentrated in the community willing to share and the others remain subcritical. “Lift then clamp” can be read as the structure of a free hook plus a paywall, to which the calculations of Chapter 4 add limits. “Gather when light, release when heavy” is about adjusting prices over time, not across people. All three readings are marked as interpretations.
References
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).