5D Library › Algorithmic Game Theory, Braess's Paradox and the Price of Anarchy in Traffic and Attention Networks · 中文
Chapter 8 Practice and a Pre-Registered Trial
Note
The checklist in this chapter is a thinking tool, not a management standard, a marketing method or investment advice. The example is a fictional “Studio Qinghe”; all names, brands, channels and numbers are invented to show the arithmetic and come from no real organisation, platform or client. Process design, budget allocation and partnership arrangements in any real team need its own data and, where necessary, advice from relevant professionals.
Part 1 A four-step checklist for congestion and diversion
The models of the first six chapters can be organised into four questions. They help a team find the places in its workflow, budget and partnerships where “everyone doing what is best for himself” may make the whole worse.
- A Braess audit of single-point bottlenecks. Is there a “fast track” in the workflow that turns two key people, who used to carry different loads, into the compulsory passage for all business (Chapter 1)? Would closing it actually shorten average handling time?
- The marginal congestion tax in red-ocean niches. Are several projects of the same team competing for the same audience or the same keywords? When the budget is allocated, does the team look at each channel’s average acquisition cost, or at the marginal cost of spending one unit more (Chapters 3 and 4)? The difference is a congestion tax the team levies on itself.
- Throttling with a cap on work in progress. Is there a limit on the number of tasks open at once? The higher the cost of switching between tasks, the more dangerous it is to have no limit (the Guanzi’s “when full, the rear pushes the front” in Chapter 7 is only an analogy).
- A stable-matching check of partnerships. Do current partnerships contain blocking pairs, a creator and a brand who would each rather work with the other (Chapter 6)? If they were re-matched by deferred acceptance, which side would propose?
Calculation
The checklist applied to the fictional “Studio Qinghe”. Step 1: two senior reviewers, W and L, each take days (where is the share of requests they handle), and standard QA and standard compliance take 3 days each; requests used to go “W → QA” or “compliance → L”, and a “fast track” lets W pass work straight to L. Step 2: the team needs 3,000 new customers a week, and in each of three channels the cost per customer rises with the team’s own volume there: hot keyword , mid channel , niche channel ( in thousands). The program compares an allocation that equalises average cost across channels with one that equalises marginal cost. Step 3: tasks arrive at 8 a week, and one person working on one task at a time finishes 10 a week, but each additional open task cuts efficiency by 3 % through switching; the program compares caps on work in progress (tasks beyond the cap wait in a visible backlog). Step 4: the preferences and current pairings of 4 creators and 4 brands; the program checks for blocking pairs and compares deferred acceptance with each side proposing.
Calculation
# The four-step checklist applied to a FICTIONAL team ("Studio Qinghe"); every name and
# number is invented to show the arithmetic, not taken from any real organisation or
# client.
# Step 1. Braess audit of an approval "fast track". Requests flow from intake to
# release. Senior reviewers W and L each take 4x days (x = share of weekly requests they
# handle); standard QA and standard compliance take 3 days each. Route A: W -> QA; route
# B: compliance -> L. A fast-track rule lets work go W -> L directly, skipping both
# standard checks. Equilibrium = minimum of the potential 2w^2 + 2l^2 + 3a + 3b (w, l =
# loads on W and L).
for ft in (False, True):
grid = [(i / 200, j / 200) for i in range(201) for j in range(201 - i)
if ft or i + j == 200]
phi = lambda g: 2 * (1 - g[1]) ** 2 + 2 * (1 - g[0]) ** 2 + 3 * g[0] + 3 * g[1]
a, b = min(grid, key=phi)
z, w, l = 1 - a - b, 1 - b, 1 - a
t = a * (4 * w + 3) + b * (3 + 4 * l) + z * (4 * w + 4 * l)
print(f"step 1: fast track {'on ' if ft else 'off'}: average days {t:.2f};"
f" shares A {a:.2f}, B {b:.2f}, fast track {z:.2f}")
# Step 2. The team's own campaigns crowd each other. Cost per acquisition in channel k
# rises with the team's weekly acquisitions q_k (thousands): CPA_k = a_k + b_k q_k.
# Target: 3,000.
CH = {"hot keyword": (6.0, 10.0), "mid channel": (9.0, 3.0), "niche": (12.0, 1.0)}
def allocate(marginal, Q=3.0):
lo, hi = 0.0, 100.0 # common level of average or marginal CPA
for _ in range(100):
m = (lo + hi) / 2
q = {k: max(0.0, (m - a) / (b * (2 if marginal else 1)))
for k, (a, b) in CH.items()}
if sum(q.values()) > Q: hi = m
else: lo = m
return q
for marginal in (False, True):
q = allocate(marginal)
spend = sum(q[k] * (a + b * q[k]) for k, (a, b) in CH.items())
rule = "equal marginal CPA" if marginal else "equal average CPA "
print(f"step 2: {rule}: " + ", ".join(f"{k} {v:.2f}" for k, v in q.items())
+ f"; spend {spend:.2f}")
q = allocate(False)
print(" congestion tax q*b, average rule: "
+ ", ".join(f"{k} {q[k] * b:.2f}" for k, (a, b) in CH.items()))
# Step 3. A cap on work in progress (WIP). Tasks arrive at 8 a week; one person
# finishing one task at a time does 10 a week, but each extra open task costs 3 % of
# capacity in switching (rate 10 / (1 + 0.03 (open - 1))). Tasks beyond the cap K wait
# in a visible backlog.
lam = 8.0
mu = lambda n: 10 / (1 + 0.03 * (n - 1))
print("step 3: WIP cap K open tasks in backlog days request->done"
" days start->done")
for K in (1, 2, 3, 5, 8, None):
p, stable = [1.0], True
for n in range(1, 5000):
p.append(p[-1] * lam / mu(n if K is None else min(n, K)))
if p[-1] > 1e12: stable = False; break
if not stable:
print(f" {'none':>8} no steady state: open tasks pile up without limit")
continue
s = sum(p); p = [x / s for x in p]
L = sum(n * x for n, x in enumerate(p))
O = sum(min(n, K) * x for n, x in enumerate(p))
print(f" {K:>8} {O:5.2f} {L - O:5.2f} {7 * L / lam:5.1f}"
f" {7 * O / lam:5.1f}")
# Step 4. Stable-matching check of the current creator-brand pairings.
CRE = ["Ana", "Bo", "Chen", "Dee"]; BR = ["tea", "gear", "skin", "books"]
PC = {"Ana": ["tea", "gear", "skin", "books"], "Bo": ["gear", "skin", "tea", "books"],
"Chen": ["skin", "tea", "gear", "books"], "Dee": ["books", "tea", "gear", "skin"]}
PB = {"tea": ["Bo", "Chen", "Ana", "Dee"], "gear": ["Chen", "Ana", "Bo", "Dee"],
"skin": ["Ana", "Bo", "Chen", "Dee"], "books": ["Dee", "Ana", "Bo", "Chen"]}
now = {"Ana": "books", "Bo": "tea", "Chen": "gear", "Dee": "skin"}
def blocking(m):
inv = {b: c for c, b in m.items()}
return [(c, b) for c in CRE for b in BR if b != m[c]
and PC[c].index(b) < PC[c].index(m[c])
and PB[b].index(c) < PB[b].index(inv[b])]
def da(prop, pp, rp): # deferred acceptance, "prop" proposes
held, nxt, free = {}, {x: 0 for x in prop}, list(prop)
while free:
x = free.pop(0); y = pp[x][nxt[x]]; nxt[x] += 1
if y not in held: held[y] = x
elif rp[y].index(x) < rp[y].index(held[y]): free.append(held[y]); held[y] = x
else: free.append(x)
return held
pairs = lambda m: ", ".join(f"{c}-{m[c]}" for c in CRE)
bp = lambda m: ", ".join(f"{c}-{b}" for c, b in blocking(m)) or "none"
print(f"step 4: current pairs {pairs(now)}\n blocking pairs: {bp(now)}")
for lab, m in (("creators propose", {c: b for b, c in da(CRE, PC, PB).items()}),
("brands propose ", da(BR, PB, PC))):
m = {c: m[c] for c in CRE}; inv = {b: c for c, b in m.items()}
print(f" {lab}: {pairs(m)}; blocking pairs: {bp(m)}"
"\n ranks of partners: "
f"creators {[PC[c].index(m[c]) + 1 for c in CRE]}, "
f"brands {[PB[b].index(inv[b]) + 1 for b in BR]}")
# step 1: fast track off: average days 5.00; shares A 0.50, B 0.50, fast track 0.00
# step 1: fast track on : average days 6.00; shares A 0.25, B 0.25, fast track 0.50
# step 2: equal average CPA : hot keyword 0.70, mid channel 1.33, niche 0.98; spend 38.93
# step 2: equal marginal CPA: hot keyword 0.45, mid channel 1.01, niche 1.53; spend 37.73
# congestion tax q*b, average rule: hot keyword 6.98, mid channel 3.98, niche 0.98
# step 3: WIP cap K open tasks in backlog days request->done days start->done
# 1 0.80 3.20 3.5 0.7
# 2 1.50 3.16 4.1 1.3
# 3 2.14 3.34 4.8 1.9
# 5 3.47 4.70 7.1 3.0
# 8 6.47 21.43 24.4 5.7
# none no steady state: open tasks pile up without limit
# step 4: current pairs Ana-books, Bo-tea, Chen-gear, Dee-skin
# blocking pairs: Ana-skin, Bo-skin, Chen-skin, Dee-books
# creators propose: Ana-tea, Bo-gear, Chen-skin, Dee-books; blocking pairs: none
# ranks of partners: creators [1, 1, 1, 1], brands [3, 3, 3, 1]
# brands propose : Ana-skin, Bo-tea, Chen-gear, Dee-books; blocking pairs: none
# ranks of partners: creators [3, 3, 3, 1], brands [1, 1, 1, 1]Step 1: without the fast track, requests split evenly and take 5.00 days on average; with it, half the requests go W → L, W and L each handle 75 % of requests, all three routes take 6.00 days, and every request is 20 % slower. This is Braess’s network from Chapter 1 again: the fast track is “faster” for every request, until everyone uses it. The conclusion is to check whether the fast track should be open only to the few truly urgent requests, not to abolish review.
Step 2: with average costs equalised, the hot keyword brings 700 customers and total spending is 38.93 (in thousands of currency units, here and below); with marginal costs equalised, the hot keyword falls to 450 customers and the niche channel rises to 1,530, for total spending of 37.73, about 3 % less. Under the average-cost rule the “congestion tax” on the hot keyword is 6.98, seven times that on the niche channel (0.98): spending one more unit on the hot keyword not only costs the average price but also raises the cost of all the team’s other spending in that channel. The saving is modest, consistent with Chapter 2’s finding that losses are limited with linear costs, but its direction is clear.
Step 3: with a cap of 1, a request takes 3.5 days on average from being raised to being finished, and only 0.7 days once work starts; with a cap of 3, 4.8 days; with a cap of 8, 24.4 days, only 5.7 of them in progress: switching leaves completion barely faster than arrivals, so the backlog grows long; with no cap, the rate of completion falls as open tasks pile up until it drops below the arrival rate, the system has no steady state, and the backlog grows without limit. In this model a smaller cap is always better because the model has no reason for parallel work, such as waiting for others’ feedback; in real work the right cap has to be set from one’s own data. The size of the switching cost (3 %) is illustrative. The relation between the average number in a queue and the average waiting time follows Little’s law (Little 1961).
Step 4: the current pairings contain four blocking pairs; for example, Ana and skin each prefer the other. When the creators propose, all four creators get their first choice and three brands get their third; when the brands propose, the reverse. Both results contain no blocking pairs and are stable; which one to use is a choice of whom the rule favours and should be stated in advance. All names and preferences are invented.
Part 2 A pre-registered randomised controlled trial
Question
In a multi-round route-choice game that contains a shortcut liable to trigger Braess’s paradox, does showing participants at the end of each round “how much delay you added for others this round”, in addition to each route’s travel time, and deducting congestion points equal to that amount, reduce the share of groups trapped in the inefficient Braess equilibrium?
Evidence
Earlier experiments. In 2009 Rapoport, Kugler, Dugar and Gisches had fairly large groups choose routes repeatedly in two networks, with and without added links; their results rejected the view that Braess’s paradox is marginal and fades with experience: financially motivated participants converged with experience on the equilibrium routes of the augmented network, even though their earnings fell sharply as a result (Rapoport et al. 2009). In 2009 Morgan, Orzen and Sefton found that in networks changed in Braess’s way, participants’ route adjustments lowered their payoffs, and that aggregate flows were close to equilibrium levels but with systematic deviations explained by a simple reinforcement-learning model (Morgan et al. 2009). In 2007 Selten and colleagues had groups of 18 choose repeatedly between two roads; mean numbers were close to equilibrium, but fluctuations persisted, and the group that knew only its own travel times switched roads significantly more often (Selten et al. 2007). In 2012 Gisches and Rapoport compared “knowing only one’s own payoff” with “knowing everyone’s routes and payoffs” and found that aggregate route choices converged to equilibrium under both information conditions (Gisches and Rapoport 2012).
These results show that simply providing more information about traffic is not enough to lead participants out of the Braess equilibrium. This trial tests a different intervention: letting each person see and bear her own externality. The program of Chapter 3 showed that once points are deducted for the externality, “everyone on the bridge” is no longer an equilibrium.
Design
| Element | Specification |
|---|---|
| Design | Two parallel arms, randomised by six-person group, stratified by site, 1:1 allocation |
| Participants | Adults aged 18 or over recruited publicly through universities and research institutes, paid according to task earnings as is usual in experimental economics; not recruited from the author’s community, students or customers, including the clients of the author’s businesses |
| Exclusions | People who have taken part in route-choice or congestion-game experiments; participants in this study’s pilot |
| Task | 40 rounds. In each round each person chooses the upper route U, the lower route L or the bridge route Z (the network of Chapter 3: 10 per user on the congestible roads, 70 on the fixed roads, 0 on the bridge); earnings per round = 200 − travel time (arm A also subtracts congestion points). The rules are explained in writing with examples before the start, identically in both arms |
| Common interface | After each round: each route’s travel time in that round, the participant’s own travel time and earnings, and cumulative earnings |
| Arm A (externality shown + points) | Also shows “the delay you added for others this round”, namely the number of other users on each congestible road you used × 10, and deducts the same number of congestion points from earnings; with a two-minute neutral explanation (“each extra person on a congestible road slows each other person on it by 10”), without suggesting which route to take |
| Arm B (control) | Common interface only |
| Blinding | Participants know their own interface but not the study hypothesis or the other arm’s interface; the primary outcome is computed automatically; analysts do not know arm labels until the analysis plan is locked |
| Primary outcome | Trapped in the Braess equilibrium: in at least 8 of rounds 31–40, five or six of the group’s six members chose the bridge route. One binary outcome per group |
| Secondary outcomes | Average travel time (excluding points) in rounds 31–40; the share choosing the bridge in each round; number of route switches; earnings; post-task understanding of the “externality” (coded by raters blind to arm) |
| Analysis | Group-level logistic regression: trapped ~ arm + site stratum, reported as a risk difference with its 95 % confidence interval; two-sided |
There is only one primary outcome. Average travel time is reported as the most important secondary outcome, because arm A’s points change earnings, and earnings cannot be compared directly.
Sample size
Calculation
# Chapter 8 trial. Part 1: the 6-player Braess game of Chapter 3 (links s->v and w->t
# cost 10 per user, v->t and s->w cost 70, bridge v->w 0) played for 40 rounds by simple
# learning BOTS: a MODEL of the hypothesis, not data about people. Control bots see
# every route's time and minimise time; treatment bots also pay congestion points equal
# to the delay they add to others, (n_e - 1) x 10 on each congested link. Primary
# outcome per group: TRAPPED = in at least 8 of rounds 31-40, five or six players took
# the bridge route. Part 2: sample size.
import math, random
from statistics import NormalDist
rng = random.Random(15)
ROUTES = {"U": ("sv", "vt"), "L": ("sw", "wt"), "Z": ("sv", "vw", "wt")}
COST = {"sv": lambda n: 10 * n, "vt": lambda n: 70, "sw": lambda n: 70,
"wt": lambda n: 10 * n, "vw": lambda n: 0}
def route_cost(others, r, points):
"""Cost of route r for a player when the other five chose 'others'."""
n = {e: sum(e in ROUTES[o] for o in others) + (e in ROUTES[r]) for e in COST}
t = sum(COST[e](n[e]) for e in ROUTES[r])
if points: t += sum((n[e] - 1) * 10 for e in ROUTES[r] if e in ("sv", "wt"))
return t
def group(points, heed=1.0, rounds=40, beta=0.2, phi=0.7):
att = [{r: -100.0 + rng.gauss(0, 10) for r in ROUTES} for _ in range(6)]
uses = [points and rng.random() < heed for _ in range(6)] # bots counting points
trap, late = 0, 0
for t in range(rounds):
ch = []
for a in att:
w = {r: math.exp(beta * (a[r] - max(a.values()))) for r in ROUTES}
u, acc = rng.random() * sum(w.values()), 0.0
for r in ROUTES:
acc += w[r]
if u <= acc: ch.append(r); break
for i, a in enumerate(att): # learn from all routes' costs this round
others = ch[:i] + ch[i + 1:]
for r in ROUTES:
a[r] = phi * a[r] - (1 - phi) * route_cost(others, r, uses[i])
if t >= rounds - 10:
trap += ch.count("Z") >= 5; late += ch.count("Z")
return trap >= 8, late / 60
for pts, heed, lab in ((False, 1, "route times only"),
(True, 1, "plus externality points"),
(True, 0.5, "points, half the bots ignore them")):
res = [group(pts, heed) for _ in range(1000)]
print(f"bots, {lab:33s}: trapped {sum(r[0] for r in res) / 1000:.0%};"
f" bridge share in rounds 31-40 {sum(r[1] for r in res) / 1000:.0%}")
def n_per_arm(p1, p2, alpha=0.05, power=0.90):
z = NormalDist().inv_cdf; pb = (p1 + p2) / 2
num = (z(1 - alpha / 2) * math.sqrt(2 * pb * (1 - pb))
+ z(power) * math.sqrt(p1 * (1 - p1) + p2 * (1 - p2)))
return math.ceil(num ** 2 / (p1 - p2) ** 2)
print("assumed share trapped: control -> treatment"
" analysable GROUPS per arm (90 % power)")
for p1, p2 in ((0.60, 0.30), (0.60, 0.35), (0.70, 0.40), (0.50, 0.25)):
print(f" {p1:.2f} -> {p2:.2f} {n_per_arm(p1, p2)}")
n = n_per_arm(0.60, 0.30); rec = math.ceil(n / 0.9)
print(f"chosen 0.60 -> 0.30: {n} groups per arm; allowing 10 % incomplete groups,"
" recruit")
print(f" {rec} groups per arm = {6 * rec} participants per arm, {12 * rec} in all")
hits = 0
for _ in range(4000):
a = sum(rng.random() < 0.60 for _ in range(n))
b = sum(rng.random() < 0.30 for _ in range(n))
pa, pb = a / n, b / n; pp = (a + b) / (2 * n)
hits += abs(pa - pb) / math.sqrt(pp * (1 - pp) * 2 / n) > 1.96
print(f"simulation check of power with {n} groups per arm: {hits / 4000:.2f}")
# bots, route times only : trapped 77%; bridge share in rounds 31-40 87%
# bots, plus externality points : trapped 0%; bridge share in rounds 31-40 21%
# bots, points, half the bots ignore them: trapped 4%; bridge share in rounds 31-40 50%
# assumed share trapped: control -> treatment analysable GROUPS per arm (90 % power)
# 0.60 -> 0.30 56
# 0.60 -> 0.35 82
# 0.70 -> 0.40 56
# 0.50 -> 0.25 77
# chosen 0.60 -> 0.30: 56 groups per arm; allowing 10 % incomplete groups, recruit
# 63 groups per arm = 378 participants per arm, 756 in all
# simulation check of power with 56 groups per arm: 0.89The first part of the program is not data but a model of the hypothesis: each “bot” chooses routes weighted by their past costs (cheaper routes are more likely to be chosen, but with some randomness) and updates its estimates after every round from the costs of all routes. Among bots that see only route times, 77 % of groups are trapped in the Braess equilibrium and 87 % of choices in the last 10 rounds are for the bridge, in line in direction with Rapoport and colleagues’ finding that most participants ended up on the bridge; among bots charged points, no group is trapped and the bridge share falls to 21 %; even when half the bots ignore the points entirely, only 4 % of groups are trapped. This gap is an upper bound: the bots’ response to points is built in, and real people may not understand the points, not care about them or distrust their meaning.
The sample size is therefore based on much more conservative assumptions: 60 % of control groups trapped and 30 % of arm A groups, two-sided , 90 % power. Each arm needs 56 analysable groups; allowing about 10 % of groups to be incomplete because someone drops out, each arm recruits 63 groups, that is, 378 people per arm and 756 in all; the simulated power is 0.89. If the true difference is only 25 percentage points (60 % against 35 %), each arm needs 82 groups. These parameters are assumptions and should be checked against pilot data before the trial is run.
Note
Why the group is the unit. The six people in a group influence each other’s choices, and the outcome is itself the state of the group, so both randomisation and analysis are by group and the sample size is counted in groups; with one outcome value per group, no intraclass correlation or design effect is needed. With only six people per group, this is the finite-player game of Chapter 2; the experiments of Rapoport and colleagues (2009) used groups of about 18 (Rapoport et al. 2009), so this design is smaller in scale and its conclusions cannot be extended directly to larger groups.
Decision rule
Calculation
# Decision rule for the Chapter 8 trial, fixed before any data are collected. diff =
# share of TRAPPED groups in the control arm (route times only) minus the treatment arm
# (own extra delay to others + congestion points); POSITIVE diff = fewer trapped groups
# with the treatment. 95 % CI from the pre-specified model (logistic regression on the
# group-level outcome, arm + site stratum, reported as a risk difference). Smallest
# difference of interest: 0.15 (15 percentage points). Fidelity: the treatment screen
# must have been shown as specified in >= 90 % of treatment-arm rounds.
SESOI = 0.15
def verdict(diff, lo, hi, fidelity, time_change):
if fidelity < 0.90:
return (f"diff {diff:+.2f}: NOT INTERPRETABLE - screen shown as"
f" specified"
f" in only {fidelity:.0%} of rounds")
if lo >= SESOI: v = "treatment cuts trapping by >= 15 points"
elif lo > 0: v = "treatment cuts trapping, >= 15 points not shown"
elif hi < 0: v = "treatment INCREASES trapping"
elif lo > -SESOI and hi < SESOI: v = "no difference worth acting on"
else: v = "unclear: the interval is too wide to decide"
return f"diff {diff:+.2f} ({lo:+.2f} to {hi:+.2f}): {v}; time {time_change:+.0%}"
for case in ((0.33, 0.17, 0.49, 0.97, -0.09), (0.18, 0.02, 0.34, 0.95, -0.04),
(0.02, -0.12, 0.14, 0.96, 0.00), (-0.20, -0.36, -0.04, 0.95, 0.05),
(0.35, 0.19, 0.51, 0.82, -0.10), (0.10, -0.08, 0.28, 0.94, -0.02)):
print(verdict(*case))
# diff +0.33 (+0.17 to +0.49): treatment cuts trapping by >= 15 points; time -9%
# diff +0.18 (+0.02 to +0.34): treatment cuts trapping, >= 15 points not shown; time -4%
# diff +0.02 (-0.12 to +0.14): no difference worth acting on; time +0%
# diff -0.20 (-0.36 to -0.04): treatment INCREASES trapping; time +5%
# diff +0.35: NOT INTERPRETABLE - screen shown as specified in only 82% of rounds
# diff +0.10 (-0.08 to +0.28): unclear: the interval is too wide to decide; time -2%The example results are invented. If arm A’s interface was not shown as specified in more than 10 % of rounds, the report says “not interpretable” whatever the result. Only if the lower limit of the 95 % confidence interval for the difference in trapped shares is at least 15 percentage points does the report say “the intervention reduced trapping by at least the smallest difference of interest”. Average travel time and earnings are reported only as estimates with intervals and are not used to override the primary conclusion.
What this trial cannot show
First, arm A receives two things at once: information about the externality and a price based on it. A positive result cannot tell which of them worked; Chapter 3’s visible-externality hypothesis needs a third arm, “shown but not charged”, to test it. Second, this is a laboratory game of 40 rounds with six people per group, in a known, deliberately constructed Braess network; the structure of real roads, workflows and attention markets is unknown, and they involve many more people. The conclusions cannot be extrapolated directly to any real transport policy, organisational practice or platform rule. Third, the points change arm A’s payoff structure, and participants may change their behaviour to avoid deductions; this is the intervention’s intent, but it also means that part of arm A’s “good result” comes from an external incentive, and whether it would persist once the points were removed is a question this trial cannot answer. Fourth, most participants will be university students. Fifth, only one set of parameters is used. Sixth, the primary outcome is a binary judgement of a group’s state, with thresholds (at least 8 of 10 rounds, at least 5 of 6 people) fixed in advance; with other thresholds the conclusion might differ, so the secondary outcomes will report the full curve of the bridge share by round.
Ethics and conflicts of interest
This is a low-risk study: the task is a route-choice game on a computer, involves no health intervention and collects no health information, and the only foreseeable discomfort is frustration at low earnings. Even so, the trial must be reviewed and approved by an independent ethics committee, every participant must give informed consent, and the protocol and analysis plan must be registered before the first group is enrolled (Nosek et al. 2018). The instructions state truthfully how earnings and points work; withholding the hypothesis is not deception, and everyone is debriefed afterwards. Payment rules are published in advance; anyone may withdraw at any time without giving a reason and is paid for the part completed, and the group is then recorded as incomplete. Data are stored anonymously. The results will be published whether positive or negative.
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 therefore has a commercial interest in the subject of this book. The author also founded and runs Asia Influencer X and Yafu Global, whose business includes multi-platform traffic distribution and digital-marketing networks, and therefore has a direct commercial interest in the subject of this book. The book promotes no course, card deck or service, is not investment or business advice, and uses no client data from these businesses; the Chapter 8 trial is designed to be run and analysed by researchers independent of the author’s business and does not recruit from the author’s community, students or customers, including the clients of these businesses; all the code and the study protocol are published openly for anyone to check.
References
Fan, Y. (2026). Algorithmic Game Theory, Braess's Paradox and the Price of Anarchy in Traffic and Attention Networks: From Wardrop Equilibria and Selfish Routing to Platform Congestion, Read Beside Classical Chinese Hydraulic Engineering [算法博弈论、布雷斯悖论与流量分配的无秩序代价] (Version 1.0.0). 5D Library. https://doi.org/10.5281/zenodo.23782094 · Licensed under Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International (CC BY-NC-ND 4.0).