Network Reality of Virtual Worlds
The Three Rules
of Social Capital
In This World
And Those
These slides are from teaching handouts, research papers, and recent books recent. All rights are
reserved (© Ronald S. Burt 2016). The course syllabus, course slides, book overviews, and related
research papers can be downloaded from http://chicagobooth.edu/fac/ronald.burt.
Now the
Social Network
Asia
US
Lines indicate frequent and
substantive work discussion;
heavy lines especially
close relationships.
Bill
EU and
Emerging
Markets
Bob
B
Front
Office
B
B
B
B
Strategic Leadership
Foundations (page 3)
B
CEO
C-Suite
Heir Apparent
Back
Office
B
R&D
Other Senior Person
from Burt (2017, Structural Holes in Virtual Worlds).
Figure
4. Scenes
in Second
Scenes
in Second
Life Life
(February
(February2008,
2008,just
justafter
afterthe
thedata
datadownload)
download)
Strategic Leadership
Foundations (page 12)
Attend a Meeting or Class
Wander the Earth
Shop
Meet, Engage, and Exchange with Folks
from Burt (2017, Structural Holes in Virtual Worlds).
EverQuest
II Characters
Figure
5.
EverQuest
II Characters
(during the 2006 observation period, descriptions
from EQ2.WIKIA.COM)
(during the 2006 observation period, descriptions from EQ2.WIKIA.COM)
Froglock
Dwarf
High
Elf
Halfling
Wood
Elf
GOOD: Stout and strong, Dwarves are known for bravery and a sense of honor, though not particularly for their intellect. Mentally
sharp and morally virtuous, Froglocks strive to eliminate villainy and corruption. Courageous Halflings are good-natured and
friendly, known for their humor. High Elves embody nobility and wisdom, but their stoic nature can be mistaken for arrogance.
Wood Elves are pleasant and friendly, but fierce protectors of the woodlands, battling any who would taint the purity of nature.
Dark
Elf
Iksar
Ogre
Ratonga
Troll
EVIL: Sinister, cunning, and dangerous, Dark Elves coolly prey on the weak and the ignorant. Calculating and cold, the Iksar are a
harsh but disciplined people who delight in cruelty and conquest. Ogres are aggressive brutes whose physical might is matched
only by their hunger for power. Ratonga are keenly perceptive and highly intelligent, but tend to be selfish and manipulative. Trolls
care only about satisfying their hunger for food and lust for battle, making them fearsome and deadly opponents.
Strategic Leadership
Foundations (page 13)
Barbarian
Erudite
Gnome
Half
Elf
Human
Kerra
NEUTRAL: Hearty and strong, Barbarians are loyal companions and unforgiving enemies. Erudites eschew their human heritage,
seeking arcane knowledge and mystical power. Gnomes make up for their small stature with tenacity and ingenuity. Descended from
humans and elves, Half Elves are known for fierce determination and independence. Humans are diverse and adaptable, at once
wise, foolish, and brutal. Worshiping spirits of the land, the Kerra’s docile manner can mask the fearsome predators that they are.
from Burt (2017, Structural Holes in Virtual Worlds).
Rule 1 of Social Capital
Asia
Now the
Social Network
US
Lines indicate frequent and
substantive work discussion;
heavy lines especially close
relationships.
RULE 1: Trust, reputation, and other shared beliefs are facilitated and
maintained when embedded in a closed network (Sherif, 1935; Festinger
Bill
et al., 1950; Asch, 1951; Granovetter, 1985, 1992; Burt, 1987; Coleman,
1988; Ellickson, 1991; Bernstein, 1992, 2001; Krackhardt, 1992; Barker,
Bob
1993; Burt & Knez, 1993; Burt, 2005;
Putnam, 1993; Uzzi, 1997).
B
Front
Office
B
B
B
B
Strategic Leadership
Foundations (page 6)
EU and
Emerging Markets
CEO
C-Suite
Heir Apparent
Other Senior Person
B
Back
Office
B
R&D
3
Continuing Colleague
(first cited two years
ago, z = 0.81)
2
1
All Colleagues
(z = 14.88)
B. Investment Bankers
10+
10+
Strategic Leadership
Foundations (page 10)
2
3
4
Bold line through black dots describes above
average reputations (8.1 routine t-test). Dashed
line through hollow dots describes reputations
average and below (6.1 routine t-test).
A. Analysts and
Investment Bankers
Number of Third Parties
Linking Employee
with Colleague this Year
4
banker
Mean Number of Third Parties
Connecting Banker with
Colleagues This Year
1
Mean Correlation for
Banker’s Reputation
from this Year to Next
(13-person subsample)
Probability that Relationship is Cited
Next Year as Good or Outstanding
“Reputation cannot arise in an open structure.”
(AJS, Coleman, 1988:S107)
4
3
2
1
2
3
4
1
banker
Closure-Trust Associations, Management
Dots are average Y scores within intervals of X. Graph A describes 46,231 observed colleague relations with analysts and bankers
over a four-year period (adapted from Burt, 2010:174-175). Vertical axis is the proportion of relations cited next year as good or
outstanding. Horizontal axis is number of mutual contacts this year. Logit z-score test statistics are estimated with controls for
differences in network size and adjusted for autocorrelation between relationships (Stata "cluster" option). Graph B describes for the
bankers subsample correlations between positive (above average) and negative (below average) reputations this year and next year
(adapted from Burt, 2010:166; routine t-tests reported across 1,179 banker-year observations).
from Burt (2017, Structural Holes in Virtual Worlds).
Second Life prediction from closure
in social network (n = 2,218,770)
Y = 5.17 X - .276 X2 + other
(104)
(-56)
EverQuest II prediction from closure
in social network (n = 216,677)
Y = 5.23 X - .287 X2 + other
(30)
(-19)
EverQuest II prediction from closure
in economic network (n = 199,118)
Y = 1.31 X - .058 X2 + other
(20)
(-11)
Strategic Leadership
Foundations (page 11)
More
And the Same Holds for Online Social Relationships
Dots are average Y scores within intervals of X. Second Life trust is friendship rights granted to contact as predicted in Table
3.1 by Model 2. EverQuest II trust is housing rights granted to contact as predicted in Table 3.2 by Model 4 for social relations
and Model 5 for economic relations. Standard errors in parentheses are adjusted for autocorrelation between relations from
same character using STATA “cluster” option.
from Burt (2017, Structural Holes in Virtual Worlds).
Rule 2 of Social Capital
Asia
Now the
Social Network
US
Lines indicate frequent and
substantive work discussion;
heavy lines especially close
relationships.
Bill
Bob
B
Front
Office
Strategic Leadership
Foundations (page 19)
B
RULE 2: Achievement is facilitated Bin an open
network (information breadth, timing, and
B
arbitrage advantages created by bridging
structural
1973; Freeman,
EUholes;
and Granovetter,
B
Emerging
Markets
1977; Burt,
1980, 1992, 2005; Lin et al., 1981;
Lin,CEO
2002; Cook etBack
al., 1983; Gould & Fernandez,
1989,
1994; Aral &Office
Van Alstyne, 2011).
C-Suite
Heir Apparent
Other Senior Person
B
B
R&D
Social Capital of Brokerage
2.5
Manifest as better ideas, more-positive evaluations, higher compensation,
earlier promotion, and faster teams.
E
(compensation, evaluation, promotion)
Z-Score Relative Performance
Brokerage is a large percentage of explained performance differences.
E
2.0
E
1.5
E
1.0
E
E
J
E
E
E J
E
EE E
JE E
E E EE
E
J
E E
E E
E
E
E
E
E
J
E
E
EE
E E
E
E E
E
EJ
E EE
0.5
0.0
-0.5
E
E
E EE
E
E E
E
E
E
E
E
J E J J
E
E
E
EE
EE
-1.5
E
E
E E
E
E
E
E
J
J E
E
17%
EE
J
E
E
J
E
E
J
J E
E
E
EE
J
J
E
E
9%
E
E
E
E
E
E
5
15
25
35
E
45
55
E
E
81%
65
75
Z = 2.78 - .82 ln(C)
r = -.53
85
95
Network Constraint (C)
large, open
10%
E
J
E E
-2.0
Strategic Leadership
Foundations (page 22)
E
E
-1.0
-2.5
E
55%
28%
median network
constraint (35 points)
many ——— Structural Holes ——— few
Circles are average z-score performance (Z) for a five-point interval of network constraint
Robert
(C) within each of eight study populations. Dashed line goes through mean values of Z for
intervals of C. Bold line is performance predicted by the natural log of C.
33%
2%
small, closed
James
64%
Brokerage
Contributes
"Slightly More
than Half"
of Predicted
Variance in
Performance
Differences
between
Managers:
Network
constraint (white),
job rank (red),
and other factors
(striped). First
pie is investment
banker
compensation
and analyst
election to the
All-America
Research Team.
Second pie is
supply-chain and
HR manager
compensation
in corporate
bureaucracies.
Third pie is
early promotion
to senior job
rank in a large
electronics firm.
Graph is from Figure 1.8 in Brokerage and Closure. Data are pooled
across eight management populations. Pie charts are from Figure
2.4 in Neighbor Networks. On causal order, see Appendix VIII.
Second Life prediction from
constrained social network
Second Life prediction from
nonredundant social contacts
EverQuest II prediction from
nonredundant social (upper)
versus economic (lower) contacts
EverQuest II prediction
from constrained social (upper)
versus economic (lower) network
Strategic Leadership
Foundations (page 23)
25+
Effective Size
(Number of NonRedundant Contacts)
Network Constraint (x 100)
Dots are average Y scores within integer (left) or five-point (right) intervals on horizontal axis. EverQuest II achievement variable is the predicted
character level in Model 8, Tables 3.4 and 3.5. Second Life achievement is the canonical correlation dependent variable in Model 15, Tables 3.5 and 3.6
(associations with individual achievement dimensions in Second Life are given in appendix graphs on page 65).
from Burt (2017, Structural Holes in Virtual Worlds).
Predicted Avatar Z-score Achievement
Predicted Avatar Z-score Achievement
Returns to Brokerage Are Evident Online in the
Network-Achievement Connection within Virtual Worlds
Rule 3 of Social Capital
Asia
Now the
Social Network
US
Lines indicate frequent and
substantive work discussion;
heavy lines especially close
relationships.
Bill
Bob
B
Strategic Leadership
Brokerage Contingencies (page 8)
B
EU and
Emerging Markets
CEO
C-Suite
Heir Apparent
Other Senior Person
B
Back
Office
RULE B3: Returns to an open
network require acceptance in the
Front
Btarget audience, or are greatly
reduced by lack of acceptance Office
B
(typically
measured by job status,
network status, reputation;
Merton, 1968; Granovetter, 1985,
1992; Tullock, 1985; Greif, 1989;
Bernstein, 1992, 2001; Uzzi,
1996, 1999; Burt, 1997, 1998;
Macy & Centola, 2007; Rider,
B
2009; Tortoriello
& Krackhardt, R & D
2010; Burt & Merluzzi, 2014; de
Vaan, Vedres, & Stark, 2015).
Si = Sj zji Sj
Network Status
Network status is on the vertical axis of the top graph. Status
is defined in the same way that price is defined in the general
equilibrium model: Si = Sj zji Sj, where Si is status of person i, and
zji is connection from j to i. Like price, status is only meaningful
in reference to the status of some numeraire benchmark person.
Here, status is normalized at the mean, so a score of 1.0 indicates a
person of average status in the network.
(eigenvector score / mean score)
Audience Effect, II: Broker's
Network Status Reassures or
Concerns the Target Audience
Regions indexed
by shading, functions
by shape
Sales
Network Constraint
Z-Score Compensation
Regional Ops
Product Support
Administration
Strategic Leadership
Brokerage Contingencies (page 12)
R2 = .74
r = .96 for
high status
r = .03 for
low status
Network Constraint
Sociogram is Figure 3.2 in Neighbor Networks and the graphs are from Figures 1 and 2 in Burt & Merluzzi discussion of the link between
brokerage and network status as a reputation measures (2013, "Embedded brokerage," Research in the Sociology of Organizations)
Other Status (bottom 50%)
n = 14
b = .28 for x, -.01 for x2
t = 3.90 for x, -2.44 for x2
Other Status
(bottom 50%)
n = 19
r = -0.21
t = -0.88
Predicted Avatar Z-Score Achievement
Predicted Avatar Z-Score Achievement
High Status (top 50%)
n = 25
b = .57 for x, -.01 for x2
t = 30.12 for x, -16.86 for x2
High Status (top 50%)
n = 21
r = -.92
t = -10.35
24+
NonRedundant Contacts
Strategic Leadership
Brokerage Contingencies (page 13)
few ——— Structural Holes ——— many
Network Constraint (x 100)
many ——— Structural Holes ——— few
Brokerage-Achievement Association,
Contingent on Network Status in Second Life
Achievement is the canonical correlation dependent variable in Model 15, Table S5. “High” status is above median. Scores on the
y-axis are average achievement scores for avatars within integer intervals of nonredundant contacts (left) or five-point intervals of
network constraint (right). Statistics are based on averages plotted in the graph. See Table S8 for regression results with controls
showing higher returns to brokerage for individual avatars with high status.
from Burt (2017, Structural Holes in Virtual Worlds).