Smartocracy Hicss2007

Smartocracy:
Social Networks for Collective Decision Making
Marko A. Rodriguez∗ , Daniel J. Steinbock† , Jennifer H. Watkins∗ , Carlos Gershenson‡ ,
Johan Bollen∗ , Victor Grey§ , and Brad deGraf¶
∗ Los Alamos National Laboratory, Los Alamos, New Mexico 87545
† Stanford University, Stanford, California 94305
‡ Vrije Universiteit Brussel, Brussel, Belgium 1160
§ 2idi Corporation, Concord, California 94524
¶ Media Venture Collective, Mill Valley, California 94941

Email: marko@lanl.gov

Abstract— Smartocracy is a social software system for collec- sphere. Despite the proliferation of such systems with
tive decision making. The system is composed of a social network traditional methods of aggregation, there has yet to emerge a
that links individuals to those they trust to make good decisions generalized software model for the intelligent aggregation of
and a decision network that links individuals to their voted-on
solutions. Such networks allow a variety of algorithms to convert individual contributions beyond mere vote-counting. If such a
the link choices made by individual participants into specific model did exist, it could systematically improve the state of
decision outcomes. Simply interpreting the linkages differently the art in social software design and promote the innovation
(e.g. ignoring trust links, or using them to weight an individual’s of systems geared more directly toward the aggregation of
vote) provides a variety of outcomes fit for different decision individual knowledge into collective knowledge, i.e. software
making scenarios. This paper will discuss the Smartocracy
network data structures, the suite of collective decision making supported collaborative problem solving and decision making
algorithms currently supported, and the results of two collective [3], [4].
decisions regarding the design of the system.
Smartocracy5 is a web-based social software system for
I. I NTRODUCTION collective problem solving/decision making. Smartocracy uses
a problem-solution model where individuals pose problems
The recent explosion of so-called ‘social software’ on the (i.e. issues, questions) to the Smartocracy community and
Internet has been characterized by democratic approaches propose potential solutions (i.e. options, answers). The
to content generation [1]. The Wiki is an exemplar of this proposed solutions are voted on and the aggregate preferences
approach, where all users have equal power to add or modify of all individual users yield the collective decision. To
the content of any hypertext [2]. Wikis are social software that further facilitate vote-based decision making, a trust-based
support democratic collaborative authorship. Systems exist to social network is used to represent the relations among users
support generative collaboration in many fields including, but and, in some cases, to support the automated delegation
not limited to, journalism1 , scholarly citation2 , photography3 , of decision making power along paths of trust. This paper
and hypertext bookmarking4 . discusses trust-based decision making theory, the Smartocracy
system implementation, and concludes with the results of
While these systems vary in the forms of collaboration the Smartocracy community’s aggregate system development
they support, they have in common an egalitarian social decisions as of March 2006.
structure and they all aggregate user contributions into shared
representations of collective belief. For instance, in addition
to contributing news stories, users of the Digg web service
II. C OLLECTIVE D ECISION M AKING
vote for stories they think highly of and can view the most
popular stories for different subjects. Del.icio.us users save Collective decision making, in the context of Smartocracy,
and categorize bookmarks for websites they like into a involves individuals generating problems, providing potential
common pool and can easily view the most popular sites for solutions, voting for solutions, and the software aggregating
any category. Similar patterns of contribution and aggregation individual votes and ultimately deriving a final collective
occur with varying prominence throughout the social software decision. Figure 1 provides an outline of Smartocracy’s
1 Digg collective decision making process.
available at: http://digg.com
2 CiteULike available at: http://citeulike.org
3 Flickr available at: http://flickr.com
4 Del.icio.us available at: http://del.icio.us 5 Smartocracy available at: http://www.smartocracy.net

INPUT Generate Potential Individual
problem
solutions via edges that denote a vote. The network data
Solutions Solution Ranking
structure describing this system can be formally denoted
OUTPUT Final Collective as G = {N, W } where N is a collection of human (H),
solution Solution Selection Solution Ranking problem (P ), and solution (S) nodes and W is the set of
(optional) weighted semantic relationships between these entities. Note
λ
that N = H ∪ P ∪ S, H ∩ P ∩ S = ∅, and wni ,nj states that
Fig. 1. Smatocracy’s collective decision making stages there exists a directed, weighted relationship of semantic λ
between ni and nj . The only semantic relationships currently
supported by Smartocracy are λ = {trusts, votedFor}.
Given a particular problem, any individual in Smartocracy
can propose a potential solution. Individuals can also vote The connection between the trust-based social network
for a proposed solution. By voting for particular solutions, and the vote-based decision network is represented in
individuals are explicitly providing their individual ranking Figure 2 where the solid edges represent the weighted trust
of the problem’s solution set. The suite of Smartocracy relations amongst individuals and the dashed edges represent
collective solution ranking algorithms aggregates all the the weighted votes that individuals provide for particular
individual votes into a collective solution ranking. Collective solutions to problems. The following two sections will further
ranking algorithms currently supported by Smartocracy are discuss each network’s constructs.
direct democracy [5], dynamically distributed democracy
[6], and proxy vote [7]. The result of the collective solution
domain 1
ranking algorithm serves as input to a solution selection
problem 1
function. For nominal-, or categorical-, based solution sets, a potential
the highest ranked solution is considered the collective solution
decision. For numeric-, or gradient-, based solution sets, a
weighted average selection function is calculated based on
their respective rank. The result returned by the selection
...
...
function is the collectively derived solution to the problem. It
is important to note that the solution selection function is an
optional step that is used only if a single deﬁnitive solution
is required. In some cases, the solution to the problem may

...
be the collective ranking itself.

trust-based social network
III. T HE N ETWORKS for domain 1
problem n
Social networks are used to represent the relationships
between individuals of a population. Most commercial social Fig. 2. The connection between the social and decision networks
network systems (e.g. MySpace6 and LinkedIn7 ) denote static
relational ties between users, such as ‘is a friend’ or ‘is a
colleague’. For collective decision making, friendship is not A. Trust-Based Social Network
necessarily the desired semantic relationship. In collective
In the Smartocracy trust-based social network, an edge
decision making, stating that an individual is a friend does
going from individual hi ∈ H to hj ∈ H states that hi believes
not identify them as a good decision maker [8]. Therefore, a
that individual hj is ‘good’ at decision making where i = j
collective decision making social network must make explicit
and H ⊆ N . In this model, hj may not trust the decision
the notion of decision making trust. For instance, if individual
making behavior of hi . Therefore, the trust-based edges be-
A provides individual B with a trust edge, then A is stating
tween individuals are directional. Furthermore, individual hi
that he or she trusts B to make a good decision.
may weight the relative trustworthiness of individual hj and,
in such cases, an edge can be represented by the conditional
There exist other artifacts in the system besides human
probability,
individuals. These artifacts include problems (i.e. current
issues being addressed by the collective) and solutions
(i.e. potential options for a particular problem). Therefore, trusts
whi ,hj = P (hj is good | hi ’s knowledge of hj ).
the graph data structure underlying the Smartocracy system
is a graph connecting individuals according to their relative This relationship states that hi trusts, according to some
trust of one another and connecting individuals to particular probability, that hj will make a good decision given hi ’s
previous understanding of hj . What this edge model lacks is a
6 MySpace available at: http://www.myspace.com representation of the domain for which trust is given. Because
7 LinkedIn available at: http://www.linkedin.com humans are multi-dimensional entities composed of various

skills and beliefs, individual hi may trust the decision making In a particle swarm, a particle is considered an ‘atom’ of
behavior of hj in one domain but not another. Therefore, it decision making influence. To calculate a collective solution
may be more appropriate, given a decision making scenario, ranking for a particular problem, particles begin their journey
to state, at human nodes and make their way, in a stochastic manner,
to the solution nodes of that problem via the trust-based social
trusts network and vote-based decision network. Because particles
whi ,hj =
are discrete, indivisible entities, the diffusion of particles
P (hj is good in domain dl through a network requires a sufficient initial distribution to
| hi ’s knowledge of hj in domain dl ). expose the underlying network topology. The more particles
initially supplied to the network, the more accurate the
There are multiple ways to represent domains in a social collective ranking [9]. At the end of the particle propagation
network context [7]. Currently, due to the focused use of algorithm, when all particles have either been destroyed or
Smartocracy, there exists only a single social network and have reached a solution, the distribution of particles over
therefore, no explicit representation of domains. As will the solution set of the problem determines the collective’s
be demonstrated in the results section, domain specific solution ranking. This section will describe the specifics
representation is a collectively desired feature for future of each of the three collective solution ranking algorithms
implementation. currently supported by Smartocracy.

B. Vote-Based Decision Network
A. Direct Democracy
In Smartocracy, any individual that votes for a potential
Direct democracy embodies the idea of ‘one-person/one-
solution is creating a directed edge from themselves in H
vote’ [5]. In direct democracy, the trust-based social network
to a solution in S. The semantic of this relationship is
is not used to calculate the collective decision. If the
λ = votedFor and can be represented by the conditional
individual does not vote, then the individual does not
probability,
participate. If an individual does not participate, then he
or she does not influence the collective solution ranking.
votedFor To implement this algorithm within the particle swarm
whi ,pj (sm ) =
P (sm is a good solution for problem pj framework, each individual is supplied with 100 particles. A
votedFor
particle can only traverse a votedFor edge, whi ,pj (sm ) . The
| hi ’s knowledge of pj ).
probability of traversing one votedFor edge over another
The above vote edge states that human hi believes that, is dependent upon the edge weight assigned by the voter.
according to some probability, sm is a good solution for Strongly weighted edges have a higher probability of being
problem pj . traversed by the particle. If the individual has not voted,
then the particle destroys itself. After one step, all particles
The weighted ranking that an individual provides for are either destroyed or are at a particular solution node to
the solution set to a particular problem is their subjective problem pj . The distribution of particles over the solution
evaluation of the relative optimality of the solutions for the nodes represents the direct democracy collective solution
problem. These are called the individual solution rankings. ranking.
To move from an individual solution ranking to a collective
solution ranking, an aggregation algorithm is required. The Figure 3 provides an example of the possible paths of a
next section will discuss the algorithmic framework used to direct democracy swarm. Humans h1 , h2 , and h3 are each
calculate collective decisions within the trust-based social provided 100 particles. After the first step, because particle
network and the vote-based decision network. diffusion is a stochastic process biased by edge weights,
s1 will have accumulated approximately 50 particles, s2 60
particles, and s3 90 particles. Note that the particles given to
IV. A LGORITHMS FOR C OLLECTIVE S OLUTION R ANKING h2 are destroyed because h2 has not voted on a solution. The
Given a trust-based social network and a vote-based normalized distribution over the solution set is s1 = 0.25,
decision network, Smartocracy supports a family of algorithms s2 = 0.30, and s3 = 0.45. This normalized distribution is the
for aggregating individual votes into a collective solution direct democracy collective solution ranking. If the solution
ranking. The currently implemented algorithms are direct set is a nominal solution set, then s3 is the collectively
democracy, dynamically distributed democracy, and proxy derived solution according to a highest rank solution function.
vote. All of these algorithms are implemented under the
parameterized particle swarm framework described in [7], [9]
and are similar to the idea of constrained spreading activation B. Dynamically Distributed Democracy
[10]. Dynamically distributed democracy, or DDD, was
developed to handle fluctuating levels of participation. In

human Furthermore, in the case that every individual votes for
1 2 3
collective a particular solution, the DDD algorithm will provide a
collective solution ranking that is equivalent to the direct
0.50 democracy algorithm’s ranking. DDD is useful in problem
0.50 0.10 0.90
domains where it is desirable for every individual to have
an equal say (i.e. influence) in the collective’s decision, but
solutions it is not feasible for every individual to actively participate
1 2 3
to problem via voting. In such cases of reduced participation, ad hoc
representative structures emerge to simulate full participation.
Fig. 3. Direct democracy particle paths

C. Proxy Vote
Proxy vote is an extension of DDD where the initial
such cases, ad hoc representative structures are created to distribution of particles is biased by the trust-based in-degree
ensure that every individual can influence the collective of an individual. That is, the more a particular individual is
solution ranking even if only through a proxy representative. trusted, the more particles that individual initially receives.
This algorithm has been shown to be an accurate way to This algorithm is used for expert-based problem domains
model the collective’s perspective as voter participation wanes where an equal say for every individual is not desired. For
[6]. proxy vote, it is assumed that the in-degree of the individual
is a representation of his or her level of expertise.
In DDD, a particle, if it is unable to take a vote edge to
a particular solution, uses the trust-based social network to In Figure 5, human h1 is supplied with 200 particles, h2
move to a proxy representative. If that representative has 200 particles, and h3 100 particles. The number of particles
voted, then the particle moves to one of the representative’s at h1 and h2 is double that of h3 because both h1 and h2
chosen solutions. If the representative has not voted, the have one incoming trust-based edge and each incoming edge
particle traverses a trust edge to move to yet another provides an extra 100 particles to an individual. As in the
representative. This iterative process continues until a solution DDD example, h2 disseminates his or her received particles
to the problem is found. to h1 because h2 trusts only h1 and h2 has not voted on a
solution to the problem. Therefore, h1 has 400 particles to
In Figure 4, human h2 has not voted on a particular distribute to his or her chosen solutions. After the second
solution. In DDD, h2 ’s decision making influence (in the step, solution s1 will have 200 particles, s2 210 particles,
form of particles) is delegated to human h1 because h2 trusts and s3 will have 90 particles. The normalized distribution
trusts
h1 , ∃wh2 ,h1 . Therefore, h1 has 200 particles to provide to his over the solution set is s1 = 0.40, s2 = 0.42, and s3 = 0.18.
or her chosen solutions. After two steps, solution s1 will have Given a highest rank selection function, s2 would be the
100 particles, s2 will have 110 particles, and s3 will have collective decision.
90 particles. The normalized distribution over the solution
set is s1 = 0.33, s2 = 0.36, and s3 = 0.30. If the selection
function selects the highest ranked solution, then solution s2
would be the collectively derived solution. human
1 2 3
collective

0.50
0.50 0.10 0.90
human
1 2 3
collective
solutions
0.50 1 2 3
to problem
0.50 0.10 0.90

Fig. 5. Proxy vote particle paths
solutions
1 2 3
to problem

V. S MARTOCRACY B ETA R ESULTS
Fig. 4. Dynamically distributed democracy particle paths
Smartocracy was released for beta testing in February
of 2006. New users join the system via an invitation from
Notice that because h3 provided a vote, the particles a currently existing user. An invitation of someone new
initially provided to h3 do not traverse the social network. automatically creates a directed edge from the inviter to the
trusts
That is, wh3 ,h2 is not used in the DDD computation. invitee. Throughout the life of the system, individuals can

reassign their outgoing trust-based edges. The interface to the Smartocracy social network’s edge distribution follows a
view one’s incoming and edit one’s outgoing trust edges is power-law. For the function P (in-degree) = a × in-degree−γ ,
shown in Figure 6. γ ≈ 2.4. This function computes the probability that any
given node will have a particular in-degree. As can be seen
in Figure 8, there exist many individuals with few incoming
trust edges and few individuals with many incoming edges.
As depicted in Figure 8, more than 75% of all individuals in
the network have 1 incoming edge while only 0.4% of the
individuals have 8 incoming edges.

0.500
0.100
probability
0.020
0.005
1 2 3 4 5 6 7 8
trust−based in−degree

Fig. 8. Smartocracy social network in-degree edge distribution

The remainder of this section will provide some preliminary
results of two initial problems posed to the Smartocracy
Fig. 6. An individual’s in and out trust-based edges collective. The two problems, in the form of questions, are:

1) What should the name of the system be?
Figure 7 provides a visualization of the 276 participants in 2) What features should we add to the system?
the Smartocracy trust-based social network as of March 7,
2006. These problems are self-reﬂective, as they concern the
system itself. The goal of the beta test period is to develop
181 176
a system that is palatable to the collective. For this reason,
256 173

237
106
177 129
127
it is thought best to allow the users to provide development
180

126
179

solutions and thus regulate the system’s evolution via the
272
109 252
251 125 172
130
131 175
174
178

128
collective. It is important to note that the results of the
249
254

145 124 226 40
46
various collective solution ranking algorithms differ. We
223 116
255

229
108
107
253
148 195
50 224
42 225
220
118

156 are not attempting to state whether one algorithm is more
60 115
62 41 159
241

111
242 63
144
250 265
43
49

140
51

44
3
211
98
221 117
121
122
optimal than another, only to demonstrate that different
55 196
266 210 199

240
230

110
207
64
61

193
257
146 36
77

66
32

132
81

82
222
5

70
114
80
119

259
123

154
metrics for collective decision making exist and that it is
33 147 143 150 155
153 268 1
245

214
218

57
263
35
184 2
72
264

0
160

96
76
267

67
269

270
142 151
120
important to study in which contexts which algorithms are
235 194 162 161
182 88 113 141

227
52
271
188
165
192 68
65
74
200

244
94
75
69

71
84

198
258
157 158
most appropriate.
212 164 89 197
163 228
273 202
83
208 186
247 87 201
170 92
189 91 86
246 185 135
171 93
213 187
191 137
219 90 261

236
204
169 203 168 85 139

134
A. System Name
101
232 260
190 138
216 104 262
97 205
275
The system name problem was posed by the developers in
112
103
102
order to determine the most appropriate name for the service.
248
217
206
276
274
136
133
Any individual could provide a potential name solution as
99 100

well as vote on which name they preferred. Table I provides
Fig. 7. Visualization of the Smartocracy network as of March 7, 2006 the 13 potential names that were provided by the group.

A visualization of the vote interface of the system is
The current in-degree edge distribution is presented on provided in Figure 9. Between the time the initial decision
a log/log plot in Figure 8. Like most real world networks, making data was analyzed and the screenshot was taken, a

potential solution
1 DecisionNet

0.6
2 DDD
3 Dis Dis Sys

0.5
4 Cell of the Elite

0.4
5 Smart Mob Rule
6 Smartocracy

0.3
7 Holocracy
8 netocracy

0.2
9 Wisism

0.1
10 Stupocracy
11 Cheerocracy

0.0
12 antpile 1 2 3 4 5 6 7 8 9 10 11 12 13
13 Decision Network
TABLE I Fig. 10. Direct democracy system name ranking
S YSTEM NAME SOLUTIONS

new system name, Dynamocracy, was proposed. Note that
the new solution expresses the dynamic nature captured in
the difference between Table I and Figure 9.

0.6
0.5
0.4
0.3
0.2
0.1
0.0

1 2 3 4 5 6 7 8 9 10 11 12 13

Fig. 11. Dynamically distributed democracy system name ranking

Finally, Figure 12 provides the ranking of the 13 potential
solutions according to the proxy vote algorithm. Smartocracy
is once again the highest ranked solution.
0.6
0.5
0.4

Fig. 9. Creating and voting on solutions screenshot
0.3
0.2

Figure 10 provides the ranking of the 13 potential solutions
0.1

according to the direct democracy algorithm. Of the 276
0.0

participants in the system, 95 voted. Therefore, only 34% 1 2 3 4 5 6 7 8 9 10 11 12 13

of the group contributed to the direct democracy collective
solution ranking. With a highest rank selection function, the Fig. 12. Proxy vote system name ranking
name Smartocracy, is the final collective decision.

Figure 11 provides the ranking of the 13 potential Table II provides the Spearman ρ correlations between
solutions according to the dynamically distributed democracy the various algorithms’ collective solution rank distributions
algorithm. In DDD, the 181 non-participants delegated their where dd is direct democracy, ddd is dynamically distributed
decision making influence (particles) to their trusted proxy democracy, and pv is proxy vote. The strong correlations,
representatives. Given a highest rank selection function, with p-value < 2.2−16 , between the algorithms demonstrates
Smartocracy is again the collective decision. the definitiveness of the collective’s desired name choice.

For this reason, Smartocracy is the current name of the system. feature. Smartocracy was originally developed with a single
trust-based social network. Therefore, decision making
influence is delegated in the system irrespective of the
dd ddd pv
dd 1.0 0.93 0.91
domain of the problem. It became apparent to many users
ddd 0.93 1.0 0.97 that they may trust someone in one domain, but not another.
pv 0.91 0.97 1.0 The DDD/highest rank decision shows that the collective is
TABLE II more interested in having domain specific proxies, or domain
S YSTEM NAME S PEARMAN ρ RANK CORRELATIONS specific representatives. That is, decision making influence is
delegated to representatives depending on the domain of the
problem and the domain of the trust-based representation.

B. Future Features
In line with today’s widely-used strategies for social
software development, Smartocracy was originally

0.30
implemented in a bare-bones fashion with the desire
that the group steer its future development. With this idea

0.20
in mind, the problem “What features should we add to the
system?” was posed to the group. Table III provides the list
of all proposed solutions generated by members of the group.

0.10
0.00
potential solution 1 2 3 4 5 6
1 social network visualization
2 know if my proxies have taken action
3 domain specific proxies Fig. 14. Dynamically distributed democracy collective solution ranking for
4 Golightly Community features
5 dialog with other participants
6 geo-based network viz
TABLE III
Finally, the proxy vote solution ranking is perfectly
F UTURE FEATURES SOLUTIONS
correlated with the DDD solution ranking and therefore,
proxy vote shows that the collective is interested in having
domain specific proxies. The proxy vote collective solution
Of the 276 participants in the system, 98 voted for a future ranking is provided in Figure 15.
feature. Figure 13 depicts the direct democracy collective
solution ranking over the 6 proposed solutions. According
to direct democracy with a highest rank selection function,
the collective desires to have a feature that allows individuals
to know if their representatives, or proxies, have voted for a
0.30

solution to a problem.
0.20
0.10
0.30

0.00

1 2 3 4 5 6
0.20

Fig. 15. Proxy vote collective solution ranking for features
0.10
0.00

1 2 3 4 5 6 The Spearman ρ correlations between the various
algorithm’s collective solution rankings are provided in Table
Fig. 13. Direct democracy collective solution ranking for features IV where p-value < 2.2−16 . Given the perfect correlation
between DDD and proxy vote, the developers of Smartocracy
are currently working towards implementing domains into the
In Figure 14, the DDD collective solution ranking provides system as articulated in [7].
a different perspective of the collective’s desired future

dd ddd pv
dd 1.0 0.88 0.88
school admissions, popularity contests, artistic awards, ethical
ddd 0.88 1.0 1.0 decisions, judicial sentences. These are only some of the
pv 0.88 1.0 1.0 potential niches where the research presented in this paper
TABLE IV could be exploited.
F UTURE FEATURES ALGORITHM S PEARMAN ρ CORRELATIONS

VIII. ACKNOWLEDGMENTS
The Smartocracy project was organized by Brad deGraf of
VI. F UTURE D IRECTIONS the Media Venture Collective8 whose vision is to revolutionize
In spirit with the previously presented collective decision the media to better serve public interest through grassroots
making processes, future development of the system will collective action. This project is funded by grants from the
be guided by the requirements generated by the collective. Threshold Foundation and the Wallace Global Fund.
As shown in the future features section, a major push in
the development of this system will be the inclusion of R EFERENCES
domain specific representation. In [7], a model for adding
[1] M. Tepper, “The rise of social software,” netWorker, vol. 7, no. 3, pp.
domain specific representation to social networks is provided. 18–23, 2003.
Depending on the requirements of the Smartocracy group, one [2] B. Leuf and W. Cunningham, The Wiki way: quick collaboration on the
of the proposed models will be implemented. Furthermore, Web. Addison-Wesley Longman Publishing Co., 2001.
[3] F. Heylighen, “Collective intelligence and its implementation on the
the current platform can be extended to support collaborative web: Algorithms to develop a collective mental map,” Computational
discourse as presented in [4], [11]. Collaborative discourse & Mathematical Organization Theory, vol. 5, no. 3, pp. 253–280, 1999.
promotes interaction amongst the participants by means of [4] M. Turoff, S. Hiltz, H. Cho, Z. Li, and Y. Wang, “Social decision support
systems (SDSS),” in 35th Annual Hawaii International Conference on
argumentation. System Sciences (HICSS‘02), vol. 1, Hawaii, USA, 2002, p. 11.
[5] A. Lijphart, Democracies: Patterns of Majoritarian and Consensus
Finally, in terms of future research in the area of collective Government in Twenty-One Countries. Yale University Press, 1984.
[6] M. A. Rodriguez and D. Steinbock, “A social network for societal-scale
decision making, we intend to explore various collective decision-making systems,” in NAACSOS ’04: Proceedingss of the North
ranking algorithms for different classes of problems to American Association for Computational Social and Organizational
understand which algorithms are best suited for particular Science Conference, Pittsburgh, PA, USA, 2004. [Online]. Available:
http://arxiv.org/abs/cs.CY/0412047
decision making situations. Given a problem set with known [7] M. A. Rodriguez and D. J. Steinbock, “The anatomy of a large scale
optimal solutions, which algorithms best aggregate the collective decision making system,” Los Alamos National Laboratory
perspectives of the individuals to yield the optimal solution? Technical Report [LA-UR-06-2139], Tech. Rep., March 2006.
[8] C.-N. Ziegler and J. Golbeck, “Investigating correlations of trust and
Such a study would expand the use of Smartocracy into interest similarity - do birds of a feather really flock together?” Decision
the domain of collective intelligence research and would Support Systems, vol. [in press], 2006.
bring Smartocracy to a level that is a generalization of [9] M. A. Rodriguez and J. Bollen, “Simulating network influence
algorithms using particle-swarms: Pagerank and pagerank-priors,” [sub-
domain specific human collective intelligence systems like mitted], 2005. [Online]. Available: http://arxiv.org/abs/cs.DS/0602002
[4], [12]–[14]. [10] F. Crestani and P. L. Lee, “Searching the web by constrained spreading
activation,” Information Processing and Management, vol. 36, no. 4, pp.
585–605, 2000.
[11] M. Turoff, S. R. Hiltz, M. Bieber, J. Fejermestad, and A. Rana,
VII. C ONCLUSION “Collaborative discourse structures in computer mediated group commu-
Social systems, in order to make the ‘right’ decisions, nications,” in 32nd Annual Hawaii International Conference on System
Sciences (HICSS‘02), vol. 1, Hawaii, USA, 1999.
require sensors capable of delivering relevant information. [12] N. Johnson, S. Rasmussen, C. Joslyn, L. Rocha, S. Smith, and M. Kan-
Polls have been used to collect information relevant for tor, “Symbiotic intelligence: self-organizing knowledge on distributed
societal-scale decision making. The drawbacks of polls are networks driven by human interaction,” in 6th International Conference
on Artificial Life, 1998.
well known: either they require public participation from a [13] C. A. Kaplan, “Collective intelligence: A new approach to stock price
large sector of the population, or they might be biased and forecasting,” in Proceedings of the 2001 IEEE Systems, Man, and
therefore, not representative of the collective opinion. By Cybernetics Conference, New York, USA, 2001.
[14] D. Steinbock, C. Kaplan, M. A. Rodriguez, J. Diaz, N. Der, and
exploiting social network structures, systems such as the one S. Garcia, “Collective intelligence quanitifed for computer-mediated
presented here can overcome some of these drawbacks. With group problem solving,” University of California at Santa Cruz, Tech.
many collective solution ranking algorithms, other than just Rep., 2004. [Online]. Available: http://arxiv.org/abs/cs.CY/0412064
direct democracy, it is possible to learn how these metrics
effect the collective solution outcome. In turn, over time
it may be possible to understand in which context which
algorithm is best.

Smartocracy could be used to make social decisions within
communities of any type, from frivolous to scientific, from
political to educational: scientific peer review, project funding, 8 Media Venture Collective available at http://www.mediaventure.org

Smartocracy Hicss2007

More Related Content

What's hot (7)

Viewers also liked (15)

Similar to Smartocracy Hicss2007 (20)

Recently uploaded (20)

Smartocracy Hicss2007