Unsolicited Commercial Email: An Attention Resource
Perspective
Oleg V. Pavlov"
Worcester Polytechnic Institute
Nigel Melville
Boston College
Robert K. Plice
San Diego State University
Abstract
The growth of unsolicited commercial email (UCE) imposes increasing costs on
organizations and causes considerable aggravation on the part of email recipients. A
thriving anti-spam industry addresses some of the frustration. This paper contributes to
our understanding of the UCE phenomenon by drawing on scholarly work in areas of
marketing and resource ownership and use. Adapting the tragedy of the commons to the
email context, we identify a causal structure that drives the direct e-marketing industry.
Computer simulations indicate that although filtering may be an effective method to curb
UCE arriving at individual inboxes, it is likely to increase the aggregate volume, thereby
boosting overall costs. The analysis advances understanding of the digital commons, the
economics of UCE, and has practical implications for the direct e-marketing industry.
Keywords: SPAM, unsolicited commercial email, UCE, system dynamics, filtering,
computer simulation
Introduction
Computer-mediated communication has become one of the accepted channels in the mix
of outlets that modern companies rely on to advertise their products (Figure 1). Electronic
mail (email) advertising had nearly a billion dollars in revenue in 2001 and is predicted to
reach several billion dollars within a few years (Martin, Durme et al. 2003). Reputable
commercial establishments, such as J.C. Penney, Barnes and Noble, and Borders use
email for communicating with customers (Martin, Durme et al. 2003). The marketing
industry’s search for an optimal portfolio of online and traditional advertising (Kover
1999; Sheehan and Doherty 2001) will eventually evolve into integrated marketing
communication programs (Brackett and Carr 2001).
* Corresponding author. Address: WPI, 100 Institute Road, Worcester, MA 01609. Email:
opavlov@wpi.edu
A cleverly designed direct marketing campaign contributes to overall sales (Chiang,
Chhajed et al. 2003). Email is more attractive than regular mail due to its lower mailing
cost, wider reach, convenience, and faster responses (Mehta and Sivadas 1995; Sheehan
and McMillan 1999; Martin, Durme et al. 2003). The cost of sending email is $5 to $7
per one thousand messages, while it is $500 to $700 for the same volume of regular mail
— two orders of magnitude greater (Martin, Durme et al. 2003). Moreover, digital
marketing campaigns are easier to customize, which can produce better response rates
than for mail campaigns (Ansari and Mela 2003). Timing is also an issue. It takes five to
ten days to receive a response with email, versus ten to fifteen days with postal surveys
(Sheehan and McMillan 1999). Finally, by including hyperlinks, email allows a degree of
interactivity not afforded by conventional direct mail campaigns (Martin, Durme et al.
2003).
Cinema [J 1%
Internet [J] 2.20% .
Figure 1: Shares of
Radio [77] 3.40% total marketing
expenditure in 2003
by channel. The UK
Direct Mail [EE] 14.30% case. Adapted from
DMIS 2004.
Outdoor and Transport [Jj 5.20%
1 25.30%
Press 7] 48.60%
Drawbacks of Direct E-Marketing
Despite the many benefits to senders of direct e-marketing campaigns, there are
pernicious impacts on consumers, email providers, and organizations. Many users are
angry and frustrated, having to sift through mountains of what they perceive to be
unsolicited commercial email in their inboxes. Email administrators struggle to maintain
high service quality in the face of increasing server loads, storage requirements, and
security threats. Of the roughly 31 billion daily emails sent globally, about 12.4 billion
(41 percent) are considered UCE! — MSN alone blocks 2.4 billion per day” The average
email user receives 4.5 adult content emails per day, 16 percent of users change their
email address due to UCE saturation, and 4.5 seconds of corporate time is wasted per
spam message.’ With some users receiving hundreds of UCE messages per day’, it’s no
wonder that frustration is growing. Research conducted by the OECD examined the costs
of UCE, finding that in June 2004, the annual spam cost per employee exceeded $1,900
' http://www.spamfilterreview.com/spam-statistics.html
? http://www.unspam.com/fight_spanv/information/spamstats.html
3 http://www. halverson.org/spam/
and the annual lost productivity per employee equaled 3.1 percent (Anonymous 2004).
Corporations are burdened by the financial and intangible costs of spam, and managers
struggle to find solutions to UCE (Corbitt 2004). They fear that the situation is likely to
become more grave in the future (Fallows 2003).
The Quest For Eyeballs: Attention as a Scarce Resource
Over thirty years ago, Nobel laureate Herbert Simon (1971) observed that attention is a
scarce resource in an information-rich society. In the parlance of modern theory, the
attention of workers can be viewed as a strategic asset (Warren 2002) that determines the
long-term success of an organization. According to Simon (1971, p. 44): “The design
principle that attention is scarce and must be preserved is very different from a principle
of “the more information the better.” When more information arrives than individuals
can process, an information overload (Simon 1971) occurs, and the likelihood of
organizational failure increases.
Examples of information overload are abundant. The first category of information
overload, when there is more information than can be analyzed by given resources, is
demonstrated by recent reports about FBI’s inability to process hours of sensitive
communication that may have high intelligence value (Lichtblau 2004). Perlow (1999)
gives an example of a situation in which information overload occurs due to the loss of
attention resource. Perlow describes a software company in which employees were
continually distracted throughout the day from their primary tasks. The reduced amount
of attention devoted to productive activities led to chronic project time overruns.
The attention squeeze and information overload are exacerbated by the onslaught of
UCE, whether viewed from an individual, organizational, or macro level. At an
individual level, spam is increasing tremendously for some email users*, At an
organizational level, spam as a percentage of regular email is substantial (Melville, Plice
et al. 2005). On an aggregate level, industry reports suggest a steady upward trend for
UCE volume. Brightmail, for example, estimates that spam as a percentage of total email
grew from 49 percent in June 2003 to 65 percent in June 2004, a 33 percent annual
increase. More somber news is that spam is moving beyond email to other platforms,
including instant messaging (spim), blogs, and mobile text messaging. Given these dire
trends and the limited time an individual spends on email, the email recipient is bound to
experience information overload.
We now extend the concept of attention as a limited resource to the realm of the digital
commons. We then construct a causal model of the UCE industry and use the model to
analyze one popular abatement mechanism: filtering. The last section summarizes results
and outlines further extensions.
“For examples see graphs at http://www.raingod.com/angus/Computing/Internet/SpanyStatistics/ .
Tragedy of the digital commons
Rapidly increasing spam volume is a result of decisions made by many self-interested
agents involved in digital marketing. These participants vie for one common resource: the
attention of email recipients. A typical outcome for a situation in which many profit-
seeking agents compete for the same scarce resource is resource overuse. The suboptimal
outcome is often described as the tragedy of the commons (Hardin 1968). A vivid
example is overexploitation of fish stock in the ocean. In an online world in which many
resources are still open to all: “Management of the digital commons is perhaps the most
critical issue of market design that our society faces” (McFadden 2001, pp. 61).
The Physical Commons
When self-interested decisions concerning a common scarce resource degrade the quality
or quantity of the resource (think fish in oceans), there is said to be a tragedy of the
commons (Hardin 1968, Mankiw 2001, Mas-Colell, Whinston et al. 1995). A common
resource is typically identified as one having the following two properties: (i) it is rival,
that is, when it is used, less is available for others; and (ii) it is nonexclusive, that is, no
one can be barred from using it. Individuals seek disproportionate private gains through
the use of the resource but do not bear the full cost. The oceans, forests, grazing lands,
the atmosphere, outer space, and highways are all susceptible to problems of the
commons. History is replete with examples of resource degradation by rational, self-
interested individuals: fish in oceans, oil reserves, etc.
The Digital Commons
The commons problem, however, is not limited to the physical world. Members of early
Usenet discussion groups in the 1980s faced analogous circumstances: the groups were
open to everyone and a small set of users could degrade the environment for all. In this
context “pollution” sprang from various sources — excessive posting or posts that were
off topic, offensive, or contained advertising — and lowered the value for all. The notion
of virtual commons was thus applied to an online common resource whose misuse by the
few degraded the value of the resource for the many (Kollock and Smith 1996).
To formalize the application of commons logic to the Internet, two conditions are
necessary (Regan 2002). First, the Web must be a “place,” just as the earth is a place. The
Internet is commonly and consistently recognized as a place for conducting a wide array
of economic and social activity. Everyday metaphors provide evidence in this regard,
with terms such as “going online,” “size of the internet,” “internet storm,” “virtual
community,” and “virus” illustrating the mapping of the physical to the virtual. The place
metaphor is also a fundamental concept used in Internet law: “the cyberspace as place
metaphor operates as one of the most compelling theories of how we have regulated
cyberspace to date, and how we are likely to regulate it in the future” (Hunter 2003, pp.
446).
The second necessary condition of online commons is that it must contain resources
(analogous to fish stock in oceans) characterized by sharing, the lack of clearly defined
private ownership, overuse, and negative externalities. A common resource is shared by
many and private ownership is unclear or non-existent, just as no one owns the depths of
the oceans and the fish stock in it. The Internet has many resources that involve sharing,
including public discussion groups, peer-to-peer file sharing networks, and email. Also,
these areas of cyberspace are characterized by the fact that no one can be barred from
using them. There is the presence of overuse by rational individuals leading to pollution
that affects all. In this context, email is a common good (Regan 2002). Spam is clearly
the result of rational businesspersons whose private revenues exceed costs to society.
Spam exhibits signs of a negative externality, which results in production that is higher
than society desires. This is evidenced by the bombardment of email addresses with spam
and the resulting financial and non-pecuniary costs borne by each of the millions of users,
their respective email administrators, and employers. Table 1 draws parallels between
UCE and fish population, which is a canonical common resource suffering from the
tragedy of the commons. As the online commons is not a biological system, we take care
in drawing the analogy homomorphically, i.e., by “paying attention to the peculiarities of
the digital environment as well” (Greco and Floridi 2004).
Table 1: Comparison of a Physical and Online Commons
FISHERIES UCE
COMMON RESOURCE Fish stock in oceans Attention of email users
SELF-INTERESTED BEHAVIOR Fish as much as possible — Send as much UCE as possible
TECHNIQUE Fishing expeditions Marketing campaigns
TRAGEDY Over fishing Information overload
Dynamic analysis of filtering
Conventional commons problems such as over-fishing have been modeled as dynamic
systems (e.g. FishBanks interactive computer simulation’). The system is comprised of at
least two agents whose quests for private gain reinforce each other until curtailed by
limits in the environment. In the case of fishing, each agent will maximize revenue or
profit until the system is overrun and fish stocks become depleted. Our approach is to
adapt this model to the case of the online commons, specifically, UCE. To the best of our
* http://www.unh.edu/ipssr/Lab/FishBank.html
knowledge, this approach to studying spam is unique and it allows leveraging what we
already know about physical commons to the problems of online commons.
We model the UCE value chain as having four participants: 1) inbox owners, 2)
harvesters, 3) operators, and 4) sponsors. The inbox population is the set of feasible
recipients of unwanted commercial e-mail. Harvesters are in the business of discovering
inboxes and compiling them into lists of e-mail addresses, which they sell to UCE
operators. UCE operators administer spam campaigns, which promote products from
sponsors. Finally, sponsors support campaigns based on their success rate. We now
explain the dynamic processes relating these four value chain participants illustrated in
Figure 2.
Address Harvesting
To receive UCE, an inbox address must be discovered by a harvester. There are hundreds
of ways to collect inbox addresses (Brain 2004). One of them is via directory harvest
attacks (DHA), in which automated programs query email servers for the existence of
millions of commonly designated usernames®. A study by the Center for Democracy and
Technology (Center for Democracy & Technology 2003) reports that harvesters are also
very effective at gathering email addresses posted on the web. There are also clearly
illegal harvesting techniques, such as when an AOL employee was recently arrested for
stealing the email addresses of 92 million AOL users (Swartz 2004a). The employee then
sold the list to an operator of an online gambling business in Las Vegas for $100,000.
That person in turn repackaged and resold the addresses to spammers for over a million
dollars. Considering the many ways in which harvesters add email addresses to their
lists, it is reasonable to assume that it is only a matter of time before an email account is
discovered (see Figure 2). We model this by including an average inbox discovery delay.
Delays, including the discovery delay, are shown in Figure 2 as two short lines crossing
an arrow.
Attention and Information Overload
The attention resource can be measured in terms of time (Simon 1971). According to a
recent survey conducted by the American Management Association, an employee
typically spends about a quarter of her day on email (Swartz 2004b). Employees whose
inboxes have been discovered possess a limited attention resource. The total demand for
attention from regular and UCE email is proportional to their respective volumes
delivered to inboxes. Assuming that regular email has a higher priority than spam, the
time left for UCE is the difference between the attention resource and the attention
devoted to regular email (see Figure 2). If the arriving volume of electronic messages is
greater than what an individual is comfortable handling, then, using Herbert Simon’s
terminology, information overload occurs.
° See http://www. postini.com/services/what_are_dhas.html
Response Rate
Advertisers have known about the negative relationship between advertising volume that
an individual is exposed to and the response rate to advertisements (Rudolph 1947;
Starch 1966; Houston and Scott 1984). Houston and Scott (1984), for example,
statistically showed a negative convex relationship between advertising readership and
the number of pages in a journal. Recent research shows that the negative relationship
holds equally well for electronic marketing. Martin et al. (2003) found that in the case of
permission-based advertising for a company operating from Finland, the likelihood of
visiting a link advertised within an email decreased as volume of email from the company
increased. A recent survey by the Pew Internet & American Life Project (Rainie and
Fallows 2004) found a decline in the readership of UCE while the UCE volume
increased, Anecdotal evidence from UCE operators (Hansell 2003) also confirms the
existence of a negative relationship between the amount of spam that a finite group
receives and the response rate. This is indicated by the loop in Figure 2 comprising UCE
volume - attention required by UCE - information overload - response rate.
Profitability
For a given overall response rate, the total number of responses a company receives
increases with its share in the email volume (in Figure 2, this is captured by positive links
between UCE Volume from Operator and Responses to UCE from Operator). More
responses imply more revenue (see Figure 2). More revenue means more profit. Greater
profit implies that with some delay (shown as two short lines crossing an arrow in Figure
2) more budget is allocated for UCE by a sponsor and thus expenditure on UCE volume
increases. The UCE volume that a sponsor can buy for a given expenditure is inversely
proportional to the UCE price that an operator charges for sending electronic messages.
It is clear from the graph in Figure 2 that there is a tendency to step up UCE volume
while profits from UCE campaigns increase. This is captured by two positive Sponsor
Profit Loops. Starting new campaigns is easy and quick thanks to specialized software
packages (Lemke 2003). An example of such a tool is iBuilder from VerticalResponse’.
Hence, campaigns have low marginal cost, and therefore cost recovery is unimportant
(Kraut, Sunder et al. 2002). A campaign requires a very low response rate to break even:
0.001 percent is often sufficient (Fallows 2003: 26). The causality acting through the
response rate forms the Attention Limit Loop (Figure 2), which checks the exponential
growth of spam.
7 http://www.verticalresponse.com/
4
Sooner At
Sponor As vert
Employes ntows: gH mpoyes ntors
discovered pee
; ~\ , gf" Unuscovered
tention resource
:
Say i . ya for UCE
UCE price SO? information
“ra
= UCE volume <<
: ener
A
“ i) ite ie
sote Wo ruse
Sponsor's AUTENnEW +
Sponsor B's
fee) gi
sonore
Figure 2: The causal structure of the UCE system
Results: Base Case
We conducted computer simulations for a fictitious organization with 10,000 inboxes.
The organization could be, for example, a company with employees or an email provider
hosting email accounts. We allow only a small portion of the accounts to be initially
known to spammers. In the model, we assume that UCE budget is proportional to the
profitability of UCE campaigns; that is, the more profitable the UCE, the greater budget
will be allocated for e-marketing. Table 2 summarizes the assumptions of the model.
Parameter values were suggested by published surveys and anecdotal evidence from
experts. Appendix offers a detailed mathematical description of the model.
Table 2: Model assumptions
Item Implementation Source
Organization size 10000 inboxes ‘Assumption
Inia poration of discovered 10 Inboxes ne
Time spent on email by anemployee 2 Hours/day Swartz 2004b
Average regular email volume 20 Messages/day Assumption
Base price of sending 1,000 UCE
messages
Response rate
Email marketing budget
$5 per 1000 messages
A declining function of UCE volume
Allocated proportionally to the past
profit from the UCE campaign
Martin, Durme et al. 2003
See Response Rate section above
Assumption
Figure 3 shows the base run for the simulated two years of life of our fictitious
organization. As expected, given the fixed number of inboxes, the total number of
discovered inboxes grows monotonically and asymptotically toward the total inbox
population (Figure 3a). Within two years, more than 80% of the inboxes have been
discovered. Positive profits accrued through spam campaigns (Figure 3b) encourage
sponsors to allocate even more resources for electronic marketing (Figure 3c and Figure
2). Greater UCE budgets allow each sponsor to spend more on email (sponsor’s
expenditure in Figure 2), which contributes to the growth of the UCE volume from
operator A and B (Figure 2). Hence, global UCE volume grows as well (Figure 3d and
Figure 2). The UCE volume that arrives to an individual inbox also grows exponentially
(Figure 3e), which is consistent with real life examples (see Footnote 4).
730
Inboxes Profit
15,000 we
Total \_I
11,250 mane 300,000
7,300 Discovered 200,000
inhowee Sh
3750 iii
o 0 a
( » @ © @ = 2 oS Se Sa a ee a a ee
Time Day) ‘ete
(a) (b)
UCE volume
UCE budget
/|
: 4
° = — - o 3 146219 365.438 ~—CSTSNdOST
Time (Day)
(c)
(d)
Average UCE volume to a discoverd inbox.
Figure 3: UCE system simulated for 2
years
0 ———
0 73 146 219 292 365 438 SIT S84 657730
Time (Day)
(e)
Results: The Impact of Spam Filtering
When foreseeing an approaching information overload, Herbert Simon suggested in the
early 1970s that filtering might be a possible solution (Simon 1971). Filtering of
unsolicited email has proven to be capable of reducing demand for attention: users report
lesser burden of spam at work than on their personal email accounts due to active email
screening efforts at work (Fallows 2003). The popularity of this solution feeds the growth
of a new and active anti-spam software industry. The method, however, has its flaws.
Many inbox users fear that aggressive filtering may lead to some legitimate email being
discarded. A survey by the Pew Internet Project (Fallows 2003: 29) found that about one
third of the respondents feared their incoming email might be blocked, and 13 percent
were convinced that it happened to them. About a quarter of respondents feared that their
outgoing emails might be filtered out by the intended recipient.
In this section we study the effect of filtering on UCE volume using the computer model
introduced earlier. We assume that the organization starts filtering email in the third year.
To address fears that legitimate email may be discarded, the organization discards only
UCE-suspects that it is most confident about. An interested reader can find a detailed
account of the filtering procedures followed by a typical medium-sized organization in
Melville et al. (2005).
10
Inboxes UCE volume
13000 ssM
Total inboxes
11250 4125M it fkering
With fering
7,500 4 275M
Discovered
7s 75
3750. inboxes 5M
F No filtering
of 0
0365730 ~~«1095~—~—«dM6)~—=«CRDS~—«DND «SSS 0365730 ~~«1098—«td6)~—~«CRDS—<2190 2585
Time (Day) Time (Day)
(a) (b)
Average UCE volume to a discoverd inbox Response rate
i
0 365 —«730~=~=«*109S=«146.=—=«TSZS—«2190 «58S 0 365—~=«730=S«109SsGO.C
2190 2585
Time (Day) Time(Day)
(c) (d)
Sponsor A's profit due to UCE
«60000
4500000 x 5S aS
Figure 4: Organization simulated for
With fering. . . .
300000 7 years with and without filtering
— policy in place
No file
0
0 «365 730~—=«1095-—=—«1460—«N2S 2190-2855
Time (Day)
(e)
Figure 4 shows a simulation that extends the run in Figure 3 for five more years (2555
days total). The figure compares the case when filtering is used to the case when no
filtering is implemented. Under each scenario, during the seven-year period, harvesters
discover all inboxes (Figure 4 a). Figures 7b — 7e depict the case when the organization
performs no active filtering with dotted lines. As in the base case simulation discussed
earlier, driven by the two Sponsor Profit Loops (see Figure 2), each operator continues to
increase UCE production. This leads to the growth in global volume of unsolicited
messages (Figure 4 b), which in turn contributes to the increase in spam arriving to
individual accounts (Figure 4 c). Eventually, attention required by UCE outgrows
attention available for UCE and information overload becomes more strongly felt (see
Figure 2). Email recipients, who are overwhelmed by increasing volumes of spam (Figure
4c), tend to delete most of it, thus driving the overall response rate down (see Figure 2
and ‘no filtering’ case in Figure 4 d). The declining response rate leads to lower revenue
and lower profit (see Figure 2 to trace the logic and Figure 4 e for the resulting profit
11
trajectory). With some delay, the declining financial performance of the electronic
marketing campaigns affects the sponsors’ UCE budgets (Figure 2). As a result, the VCE
volume tapers off (later portions of Figure 4 b).
Let us now examine the effect of filtering, which is clearly visible starting in year three.
Filtering reduces the burden from UCE on an individual by lowering UCE volume
arriving to an individual inbox (solid line in Figure 4 c). Feeling less overwhelmed by
incoming spam, readers tend to read a greater fraction of incoming messages, thus
increasing the overall response rate (solid line in Figure 4 d). Better response rates drive
profitability upwards (solid line in Figure 4 e). Better profitability encourages greater
UCE budgets (Figure 2), thus allowing sponsors to spend more on email marketing.
Greater expenditure by each sponsor boosts overall UCE volume (filtered case in Figure 4
b). Hence, an interesting result is that, while lowering the burden of spam on individuals
(Figure 4 c), filtering is likely to increase overall volume of UCE (Figure 4 b).
Discussion
In this paper we addressed the growing problem of unsolicited commercial email.
Adopting the viewpoint that in an information-rich society attention is a limited resource
—a notion proffered by Herbert Simon (1971) — allowed us to describe the problem of
spam in terms of a common resource. The common resource framework is well
understood in economic literature and has been helpful in explaining many phenomena
that lead to the overexploitation of limited resources. The situation of overexploitation of
a resource by self-interested agents is generally referred to as the tragedy of the
commons. The framework has been applied to other Internet-related problems. When
applied to the virtual world, the phenomenon has been dubbed the tragedy of the digital
commons.
Filtering has been proposed as a possible solution to information overload long before
email became popular (Simon 1971). Filtering, however, may impose costs that exceed
the benefit (Cranor and Lamacchia 1998). The benefit is the reduction of spam volume
arriving to an individual inbox. However, as our discussion in the previous sections
showed, the use of filtering is likely to stimulate production of spam. Greater spam
volume consumes more of organizations’ bandwidth and processing resources (Melville,
Plice et al. 2005). Secondly, organizations and spam senders iteratively improve their
filtering and electronic marketing tools, respectively, with no clear end to, or winner of,
such an arms race. The continuous anti-spam effort is costly. Thirdly, inbox owners bear
the cost too because false positives during spam filtering lead to the deletion of desired
email. From a theoretical standpoint, this reduces consumer surplus (Loder, Van Alstyne
et al. 2004). Table 3 summarizes benefits and drawbacks of the filtering solution.
12
Table 3: Benefit and drawbacks of the filtering solution
Benefit
Lower UCE volume to an inbox
Drawbacks
Greater global volume of UCE
Emergence of the costly “anti-spam arms race”
False-positives lower consumer surplus
Our analysis has several practical implications. Because of its effectiveness at the
individual level, filtering will inevitably increase in popularity. Considering that many
individuals still do not use filtering software and only 28 percent of small and medium
size enterprises have active anti-spam measures in place (PR Newswire 2004), the growth
in the filtering effort is likely to be very significant. As our simulations suggest,
proliferation of filtering will further boost the global UCE volume. It will also encourage
an arms race between IT managers and the filtering software companies on the one side
and spam operators on the opposing side. Owners of filtering software companies are
well positioned to benefit financially from the situation. Consumers and IT managers, on
the other hand, will pay for the new upgrades of the filtering software and for the addition
of processing and bandwidth capacity needed to handle the increased UCE volume.
Moreover, greater reliance on filtering will make the task of legitimate electronic
advertisers harder, as without the reliance on obfuscating tricks used by spammers, direct
marketers are easy targets for filtering. Hence, to remain effective, the direct marketing
industry will need to do the following: (i) foster relationships with customers; (ii) be
sensitive to the optimal emailing frequency; (iii) limit email to customers who consented
to receiving such communication; (iv) target and customize email, so as to make email
communication useful to the recipient; (v) use email as only one dimension in a
comprehensive marketing campaign; and (vi) actively distinguish itself from UCE.
13
Appendix
Below are the model equations.
Address Harvesting
Average time to discovery
Undiscovered inboxes
Discovered inboxes
Total inboxes
Fraction of inboxes discovered
Attention and Information Overload
Time an employee can spend on email
Attention resource of an organization
Discovered attention resource
Time it takes to read an email
Regular email volume to discovered inboxes
Attention devoted to regular mail
Attention available for UCE
UCE volume
Attention required by UCE
Information overload
Response Rate
Maximum response rate
Response rate
F(0)
Profitability
Sponsor A’s profit
Budget duration
Average daily profit
Sponsor A’s expenditure
Price of UCE
UCE volume from operator A
Responses to UCE from operator A
Average revenue per sale
Sponsor A’s revenue
Marketing fraction
Adjustment of spam budget
Sponsor A’s UCE budget
14
t
‘a
N
(d/dt)I, =1,/t,
T=L4+1,
i=1,/1
=
a
Bm
= 8
&s
2
W
8
References
Anonymous. "Squashing Spam." OECD Observer, 242 (2004): 37.
Ansari, A. and C. F. Mela. "E-Customization." Journal of Marketing Research 40, 2
(2003): 131-146.
Brackett, L. K. and B. N. Carr. "Cyberspace Advertising Vs. Other Media: Consumer Vs.
Mature Student Audiences." Journal of Advertising Research 41, 5 (2001): 23-32.
Brain, M. "How Spam Works." How Stuff Works (on 7/7/04 available at
http://computer.howstuffworks.com/spam.htm), (2004).
Center for Democracy & Technology. "Why Am I Getting All This Spam?" (2003).
Chiang, W. Y. K., D. Chhajed and J. D. Hess. "Direct-Marketing, Indirect Profits: A
Strategic Analysis of Dual-Channel Supply-Chain Design." Management Science 49, 1
(2003): 1-20.
Corbitt, T. "Are You Suffering from Spam?" Management Services 48, 2 (2004): 22-23.
Cranor, L. F. and B. A. Lamacchia. "Spam!" Communications of The ACM 41, 8 (1998):
74-83.
DMIS. Letterbox Factfile. 2004.
Fallows, D. "Spam: How It Is Hurting Email and Degrading Life on the Internet." PEW
INTERNET PROJECT Report, October (2003).
Greco, G. M. and L. Floridi. "The Tragedy of the Digital Commons." Ethics and
Information Technology 6, (2004): 73-81.
Hansell, S. "An Unrepentant Spammer Vows to Carry on, within the Law." The New
York Times, December 30 (2003).
Hardin, G. "The Tragedy of the Commons." Science 162, 3859 (1968): 1243-1248.
Houston, F. S. and D. Scott. "The Determinants of Advertising Page Exposure." Journal
of Advertising 13, 2 (1984): 27-34.
Hunter, D. "Cyberspace as Place and the Tragedy of the Digital Anticommons."
California Law Review 91, 2 (2003): 439-519.
15
Kollock, P. and M. Smith. Managing the Virtual Commons: Cooperation and Conflict in
Computer Communities. Computer-Mediated Communication: Linguistic, Social, and
Cross-Cultural Perspectives. S. Herring. Amsterdam, John Benjamins: 109-128, 1996.
Kover, A. J. "Another Unsolved Problem." Journal of Advertising Research 39, 3 (1999):
5.
Kraut, R. E., S. Sunder, J. Morris, R. Telang, D. Filer and M. Cronin. "Markets for
Attention: Will Postage for E-Mail Help?" Yale ICF Working Paper No. 02-28, August
(2002).
Lemke, T. "How Spammers Exploit the News: Current Events Trigger Onslaught of E-
Mail Ads." The Washington Times, October 14 (2003): AOI.
Lichtblau, E. "F.B.I. Said to Lag on Translating Terror Tapes." The New York Times,
September 28 (2004).
Loder, T., M. Van Alstyne and R. Wash. "Information Asymmetry and Thwarting
Spam." MIT Working Paper, (2004): 1-11.
Mankiw, N. G. Principles of Microeconomics. Mason, Ohio, Thomson, 2001.
Martin, B. A. S., J. V. Durme, M. Raulas and M. Merisavo. "Email Advertising:
Exploratory Insights from Finland." Journal of Advertising Research 43, 3 (2003): 293-.
Mas-Colell, A., M. D. Whinston and J. R. Green. Microeconomic Theory. Oxford,
Oxford University Press, 1995.
McFadden, D. (2001). The Tragedy of the Commons. Forbes ASAP. 168: 61-62.
Mehta, R. and E. Sivadas. "Comparing Response Rates and Response Content in Mail
Versus Electronic Mail Surveys." journal of the market research society 37, 4 (1995):
429-439.
Melville, N., R. Plice and O. V. Pavlov. "Unsolicited Commercial Email: Shedding Light
on a Grey Industry." mimeo, (2005).
Perlow, L. A. "The Time Famine: Toward a Sociology of Work Time." Administrative
Science Quarterly 44, 1 (1999): 57-81.
PR Newswire. "Sophos Survey Reveals Workers Want Employers to Take Responsibility
for Blocking Offensive Spam." September 14 (2004).
Rainie, L. and D. Fallows. "Pew Internet Project Data Memo: The Impact of Can-Spam
Legislation." March (2004).
16
Regan, P. "Privacy as Common Good in the Digital World." Jnformation,
Communication and Society 5, 3 (2002): 382-405.
Rudolph, H. J. Attention and Interest Factors in Advertising. New York, Funk &
Wagnalls Company, 1947.
Sheehan, K. B. and C. Doherty. "Re-Weaving the Web: Integrating Print and Online
Communications." journal of interactive marketing 15, 2 (2001): 47-59.
Sheehan, K. B. and S. J. McMillan. "Response Variation in E-Mail Surveys: An
Exploration." Journal of Advertising Research, July-August (1999).
Simon, H. Designing Organizations for an Information-Rich World. Computers,
Communications, and the Public Interest. M. Greenberger. Baltimore, MD, The Johns
Hopkins Press: 38-52, 1971.
Starch, D. Measuring Advertising Readership and Results. New York, McGraw-Hill,
1966.
Swartz, N. "AOL Employee Charged for Stealing Screen Names." /nformation
Management Journal 38, 5 (2004a): 16.
Swartz, N. "The Worldwide War on Spam Continues." /nformation Management Journal
38, 2 (2004b): 16.
Warren, K. Competitive Strategy Dynamics. New York, NY, John Wiley & Sons, Inc.,
2002.
17
