An Economic Solution to Spam

1
An Economic Solution to Spam

• Thede Loder, Marshall Van Alstyne, Rick Wash

2
Spam Examples
3
Want these guys in your mailbox?
Spam King Alan Ralsky spewed tens of thousands
of e-mail sales pitches per hour, bringing on the
wrath of Verizon. Tech. Review Aug 03
Richard Colbert, spammer. NYT 09-28-03
4
Self-Cancelling Spam
5
The Spam Problem

• Estimated yearly loss to businesses 10 billion
  (Ferris Research)
• ISPs estimate the cost of spam at 2-3 per user
  per month (IDG)
• 6 spam laws are pending in congress (NYTimes)
• 8 states have already made spam laws including
  recent CA law (NYTimes)
• 50 of all email is now spam (Brightmail)

6
A Working Definition

• Spam is any email received that, if given the
  choice and advanced knowledge of its value, the
  recipient would have chosen not to receive

7
Existing or Proposed Solutions

• Two Categories legislative, technological
• Technological Candidates
• Filtering Rule based (static or dynamic),
  Bayesian Filters, collective/community
  classification
• Challenge Response quasi-Turing tests, return
  address testing, computational challenge (solve
  hard problem)
• Legislative Candidates
• Banning, labeling
• Taxation, stamps

8
Problems with Pure Technology Solutions

• Filtering
• False positives, false negatives
• Costly arms-race
• Consensus definition
• Shuts down exchange
• Challenge response
• Requires human interaction
• Costly arms race
• Blocks automated emails

9
Problems with Legislative Solutions

• Banning and labeling
• Enforceability (jurisdiction?)
• Costly to police and adjudicate
• Labeling lacks incentive compatibility
• Taxation and Stamps
• Blocks wanted along with unwanted email
• Kills cost-effectiveness of email as medium

10
and Spam is Getting Worse
Whats going on?
11
Modeling Email

• Each email has a party-dependent expected value
• value to the sender s
• value to the recipient r
• s and r bounded by Vs and Vr (upper and lower)

Value to Sender (s)
-

0
Value to Recipient (r)
-

__
Vr
0
Vr
__
12
Basic Assumptions

• r and s are expected values
• Sender knows s before sending
• Sender does not know r
• Sender knows or learns their marginal cost, and
  will not send when s lt cost
• Base cost to send cs is the same for all
  distributions
• Recipient knows r upon receipt but only after
  losing an unavoidable receiving/processing cost cr

13
Email Value - Single Distribution
r (value to recipient)
__
cs
Cost for sender to send
Vr
cr
Cost for recipient to receive (sender must send)
cr
s
(value to sender)
region of positive probability in the
distribution for email with values (r, s)
__
Vr
cs
14
Derivation of Payoffs 1

• General form distribution of point
  probabilities

such that

• We assume f is uniform in each dimension and
  independent of s and r.

distribution specific constant (k)
15
Derivation of Payoffs 2

• Baseline Recipient Surplus is

Where
16
Derivation of Payoffs 3

• Similarly, we can determine the Sender Surplus
  (SS) and the Social Welfare (SW)

17
Social Welfare Contribution
r
For a single email
Social Welfare (SW) RS SS
cr
Sent email with (r, s) northeast of the diagonal
line makes a positive contribution to social
welfare if received
s
positive contribution to SW
SW
negative contribution to SW
cs
18
Interpreting Existing Solutions
Flat Tax, Computational Challenge
Filtering (all types)
r
r
Unsent
Unsent
loss
loss
cr
Unwanted
s
loss
loss
Filtered Waste
SW
cs tax
cs/n
cs
cs
19
Existing Solutions 2
Challenge Response - quasi-Turing Test
Criminalizing Spam
r
r
Unsent
loss
Unsent
loss
Unwanted
Unwanted
loss
loss
cs test cost
cs risk of penalty
cs
cs
20
Perfect Filter
r

• Definition A technological filter which
• Operates without cost
• Makes no mistakes (no false positives or false
  negatives)
• Intuits and internalizes all reader preferences
• Works for an arbitrarily large number of
  different distributions
• Eliminates, prior to receipt, any email where r lt
  cr

Received
Unsent
cr
Filtered
s
SW
cs
cs/?
21
In terms of individual and aggregate social
welfare, a system that facilitates valuable
exchange will generally dominate a system that
grants unilateral veto power to either party
22
Not all filtered email is waste
r
positive contribution to welfare
Unsent
negative contribution to welfare
cr
not sent (s lt cs)
s
unwanted by recipient (r lt cr)
Unwanted
Some spam has positive contribution
SW
cs
23
If its not waste, dont destroy it
r
Unsent

• filter nothing
• set bond 0

cr
s
Unwanted
SW
cs
24
The Attention Bond Mechanism

• Follows from mechanism design - an economic,
  rather than pure technological or legislative
  design
• Simple screening mechanism
• Challenge demands an escrowed monetary bond of
  the amount ? (a fee)
• Recipient has sole discretion to keep or return ?
• Effect a recipient-controlled variable tax on
  senders, by type.
• Tax proceeds go to recipient
• Used for non-whitelisted senders

25
Architecture for ABM
26
Recipient Payoff
r
For any particular distribution, the recipient
can remove the sender incentive to send emails
for which s lt csp?, while at the same time
gaining p?.
Unsent
Wanted
cr
Unwanted
s
Here, p 0
Positive payoff to recipient
Negative payoff to recipient
SW
cs
27
Recipient Payoff
r
For any particular distribution, the recipient
can remove the sender incentive to send emails
for which s lt csp?, while at the same time
gaining p?.
Unsent
Wanted
cr
cr -p?
s
Unwanted
Positive payoff to recipient
Negative payoff to recipient
SW
csp?
cs
28
Recipient Payoff
r
For any particular distribution, the recipient
can remove the sender incentive to send emails
for which s lt csp?, while at the same time
gaining p?.
Unsent
Wanted
cr
cr -p?
s
Unwanted
Positive payoff to recipient
Negative payoff to recipient
SW
csp?
cs
29
Recipient Payoff
r
For any particular distribution, the recipient
can remove the sender incentive to send emails
for which s lt csp?, while at the same time
gaining p?.
Unsent
Wanted
cr
s
cr -p?
Unwanted
Positive payoff to recipient
Negative payoff to recipient
SW
csp?
cs
30
Recipient Payoff
r
For any particular distribution, the recipient
can remove the sender incentive to send emails
for which s lt csp?, while at the same time
gaining p?.
Unsent
Wanted
cr
s
cr -p?
Unwanted
Positive payoff to recipient
Negative payoff to recipient
SW
csp?
cs
31
Recipient Payoff
r
For any particular distribution, the recipient
can remove the sender incentive to send emails
for which s lt csp?, while at the same time
gaining p?.
Unsent
Wanted
cr
s
cr -p?
Unwanted
Positive payoff to recipient
Negative payoff to recipient
SW
csp?
cs
32
Recipient Payoff
r
For any particular distribution, the recipient
can remove the sender incentive to send emails
for which s lt csp?, while at the same time
gaining p?.
Unsent
Wanted
cr
s
cr -p?
Positive payoff to recipient
Unwanted
Negative payoff to recipient
SW
csp?
cs
33
Recipient Payoff
r
For any particular distribution, the recipient
can remove the sender incentive to send emails
for which s lt csp?, while at the same time
gaining p?.
Unsent
Wanted
cr
s
Positive payoff to recipient
cr -p?
Unwanted
Negative payoff to recipient
SW
csp?
cs
34
Recipient Payoff
r
For any particular distribution, the recipient
can remove the sender incentive to send emails
for which s lt csp?, while at the same time
gaining p?.
Unsent
Wanted
cr
s
Positive payoff to recipient
cr -p?
Unwanted
Negative payoff to recipient
SW
csp?
cs
35
Extended Model and Comparison

• How do RS, SS, and SW compare?
• Multiple Distributions
• Perfect Filter average cost per receipt goes to
  cs/?
• ABM - Cost to sender increases by pd ?. Cost now
  cs pd ?

36
The no intervention baseline
Reader Surplus defined
Baseline surplus
37
The Readers choice of screen
Reader Surplus defined
38
The Perfect Filter
Reader Surplus defined
Reader Surplus
Bonding wins if
39
Social Planners choice of screen
Welfare defined
Optimal Screen
Always
40
Comparison of ABM to PF2-Distribution Case

• V distribution expected to be mostly Valuable
• W distribution expected to be mostly Waste
• Sender does not know r
• Sender knows to which distribution (V or W) their
  email belongs

41
Distribution V
r
V
Valuable
cr
s
SW
cs
42
Distribution W
r
W
cr
s
Waste
SW
cs
43
V Sent With Attention Bond
r
V
Unsent V
cr
s
For each distribution, the recipient can choose a
policy with a seize probability px
SW
cs
cs pv ?
44
W Sent with Attention Bond
r
W
cr
Unsent W
s
For a mostly unwanted email distribution, the
recipient is best off if they seize with a high
probability (greater p? product)
SW
cs
cs pw?
45
Recipient Surplus - ABM
r
V
W
cr
W
s
cs pw ?
SW
cs pv ?
cs
46
Recipient Surplus Perfect Filter
r
V
cr
W
s
cs /?w
SW
cs /?v
cs
47
Welfare Basis - ABM
r
V
W
cr
s
cs pw ?
SW
cs pv ?
cs
48
Welfare Basis - Perfect Filter
r
V
cr
W
Filtered Waste
Filtered Waste
s
cs /?w
SW
cs /?v
cs
49
Policy Independence

• Choose any one p or choose ? (subject to boundary
  conditions)
• No social inefficiency for adding as many
  distributions as you want (true for readers and
  senders). Each distribution can be separately
  optimized
• Policy can be adjusted to individual senders, not
  just spammers and good-guys
• Contrast to filter, which suffer from
  dramatically increased type 1 and type 2 errors
  with more distributions
• There is no incentive to misappropriate the bond
  ex-post

50
Caveats and Adoption Issues

• bottom chop
• Email with high value to the recipient but a low
  value to the sender will not be sent (sender not
  willing to risk bond)
• Infrastructure and Transaction costs
• requires escrow service(s), server
  infrastructure changes, low cost transaction
  system for small transactions, inter-escrow
  payment network (later)
• Network effects
• Mulitiple escrow companies, if not connected,
  slow adoption

51
Additional Social Benefits

• Reduced friction recipients have incentive to
  publish their contact information, not obscure it
• No costly arms race
• Direct Marketing still possible
• Permits communication otherwise eliminated (e.g.
  political speech)
• Costs remain cheaper than post office-style
  direct marketing
• Tailors to an individuals unique preferences
• Signaling information about a recipient via claim
  history

52
Conclusion

• Enabling transactions helps readers more than
  unilateral veto
• Screening mechanism forces senders to reveal
  their type
• Many desirable secondary effects

53
Future Work

• Signaling
• Non-uniform distributions
• Allow cs and cr to vary with distribution?
• What happens when sender knows r? (with some
  certainty or exactly)
• Infrastructure issues
• Show determination of minimal ? for multiple
  distributions