[Air-L] Vote mirroring as a counter-monopoly measure

Michael Allan mike at zelea.com
Sat Jul 30 04:41:27 PDT 2011


Please reply if you see a flaw in this argument.  I'm hoping it's
strong enough to underwrite our development plans, but I'm only a
practitioner.  I lack the resources to properly research the topic.


VOTE MIRRORING AS A COUNTER-MONOPOLY MEASURE [0]

Vote mirroring is arguably a sufficient measure in itself to preclude
the formation of a monopoly in the provision of online voting services
for participatory democracy.  A vote mirror can unilaterally
interconnect the voters of multiple service sites into a single
"inter-network" of participation and thus reduce the network effects
that might contribute to the dominance of a single provider.  Social
media in general is prone to monopolies, but their formation is not
inevitable [1].  By posing an early challenge, vote mirroring can
level the playing field and preclude a monopoly in voting media.

Vote mirroring works by copying votes from multiple service sites
(vote-servers) and reproducing them as images at another site (the
mirror) [2].  The technique is illustrated in figure 1 below.  The top
half of the figure shows three vote-servers together with a sample of
the votes cast on each.  Vote-servers A and C are providing a
communicative type of service in which votes are delegated
transitively, while server B is providing a conventional mass type of
service [3].


                       |                    |
             A         |         B          |           C
                       |                    |
                       |                    |
              (0)      |                    |        (0)   (0)
         (0)   |       |                    |  (0)    | 1  /
           \ 1 | 1     |                    |    \    |   / 1
            \  |       |                    |     \ 1 |  /
   (0)  (0)  \ |       |                    |      \  | /  (0)  (0)
     \ 1 |    \|       |        (0)         |       \ |/    | 1 /
      \  | 1  (2)      |  (0)    |          |        (3)    |  / 1
       \ |     |       |    \ 1  | 1 (0)    |         |     | /
        \|     | 3     |     \   |   /      |         | 4   |/
        (2)    |       |      \  |  / 1     |         |    (2)
          \ 3  |       |       \ | /        |   (0)   |    /     (0)
           \   |  (0)  |        (6)----(0)  |     \ 1 |   /       |
            \  |  /    |      1 / \  1      |      \  |  /    (0) | 1
             \ | / 1   |       /   \        |       \ | / 3     \ |
              \|/      |     (0)  1 \       |        \|/       1 \|
              (7)      |            (0)     |        (8)         (2)
                       |                    |
   --------------------+--------------------+------------------------

               (0)               D                (0)   (0)
          (0)   |                           (0)    | 1  /
            \ 1 | 1                           \    |   / 1
             \  |                              \ 1 |  /
    (0)  (0)  \ |                               \  | /  (0)  (0)
      \ 1 |    \|              (0)               \ |/    | 1 /
       \  | 1  (2)       (0)    |                 (3)    |  / 1
        \ |     |          \ 1  | 1 (0)            |     | /
         \|     | 3         \   |   /              | 4   |/
         (2)    |            \  |  / 1             |    (2)
           \ 3  |             \ | /          (0)   |    /       (0)
            \   |  (0)         (6)----(0)      \ 1 |   /         |
             \  |  /         1 / \  1           \  |  / 3    (0) | 1
              \ | / 1         /   \              \ | /         \ |
               \|/          (0)  1 \              \|/         1 \|
               (7)                 (0)            (8)           (2)


   FIG 1.  Vote mirroring.  Three separate vote servers (A, B, C) and
           a mirror that reproduces the votes of all three on a single
           server (D).  Each number in brackets shows the quantity of
           votes received by a candidate.  Note that a candidate may
           be a text, as well as a person. [4]


The mirror (bottom) reproduces the votes of all three servers.  This
involves translating the votes from their various native formats on
the original servers into the single format used on the mirror.
Voting methods differ greatly and the translation may therefore
entail a degree of information loss, making for an imperfect image.
Such imperfections cannot invalidate the overall technique, however,
because a best effort at an image is always a better reflection of
reality than no image at all.  For example, knowing that these
particular votes (figure above) all concern a single issue, the mirror
has represented the four end-candidates as rivals, with 7, 6, 8 and 2
votes respectively, because this gives a truer picture of the options
available to the voters. [5]


USES OF A MIRROR
----------------

This section concerns the direct uses of a vote mirror by a
participant.  The table below summarizes these uses for different
categories of participant together with the associated costs and
benefits.  Of the four uses shown, three are of benefit to the users
of any voting service, while the fourth is of benefit to users of mass
type services in particular.


   Table 1.  Uses, costs and benefits for different categories of
             participant.

   Use of original vote server     Use of mirror
   ==============================  ==================================
   SERVICE PARTICIPANT USE         COST         BENEFIT
   TYPE    ROLE                    (if any)
   ======= =========== ==========  ============ =====================
   Any     Candidate   Soliciting  [5]          Overview of voters,
                       votes                    Overview of rivals
                                  
           ----------- ----------  ------------ ---------------------
           Voter       Comparing                Overview of
                       candidates               candidates
                                  
                       ----------  ------------ ---------------------
                       Voting      User         Convenience
                                   registration
   ------- ----------- ----------  ------------ ---------------------
   Mass    Trailing   Delegation   User         Visibility to voters,
           candidate               registration Viability as
                                                candidate
   ==================================================================


Soliciting votes
----------------

A mirror may be used by a candidate as a tool for soliciting votes.
It provides two benefits for this purpose: an overview of all the
voters in a single interface, plus an overview of the rival candidates
(compare top and bottom in figure 1).  Seeing all the voters is
especially beneficial, because online voting tends to be a continuous
process in which voters are free to shift their votes. [5]

Comparing candidates
--------------------
A mirror may be used by voters as a tool for comparing candidates.
The benefit it provides for this purpose is an overview of all the
candidates, together with the "live" structure of their voter support.
Only a mirror can provide such an overview.

Voting
------

A mirror will typically provide a voting service of its own, making it
a combined mirror and vote-server.  This is a convenience for any
voter who uses the mirroring part to compare candidates and then
wishes to shift his vote.  Rather than having to register as a user on
the candidate's home server, he simply registers on whatever
mirror/vote-server he prefers, which is most likely the one he is
currently using.  He can then shift his vote to any candidate.

Delegation
----------
A mirror can provide a delegation service to a candidate on a mass
voting site, where it would otherwise be unavailable.  This may be
particularly useful to a candidate who is trailing in votes, as it
increases both her visibility and viability.  A new voter will often
explore upstream and inspect the supporting structure of votes before
deciding where to place his own vote.  A candidate who presents
herself in that context is more likely to receive a vote than one who
appears as number 29 (or 229) in an ordered list.


                        |                    |
            A           |         B          |           C
                        |                    |
                        |                    |
            (0)         |                    |        (0)   (0)
       (0)   |          |                    |  (0)    | 1  /
         \ 1 | 1        |                    |    \    |   / 1
          \  |          |                    |     \ 1 |  /
 (0)  (0)  \ |          |                    |      \  | /  (0)  (0)
   \ 1 |    \|          |        (0)         |       \ |/    | 1 /
    \  | 1  (2)         |  (0)    |          |        (3)    |  / 1
     \ |     |          |    \ 1  | 1 (0)    |         |     | /
      \|     | 3        |     \   |   /      |         | 4   |/
      (2)    |          |      \  |  / 1     |         |    (2)
        \ 3  |          |       \ | /        |   (0)   |    /     (0)
         \   |  (0)     |        (6)----(0)  |     \ 1 |   /       |
          \  |  /       |      1 / \  1      |      \  |  / 3  (0) | 1
           \ | / 1      |       /   \        |       \ | /       \ |
            \|/         |     (0)  1 \       |        \|/       1 \|
            (8)----(0)  |            (0)     |        (8)         (2)
                 1      |                    |
                        |                    |
 -----------------------+--------------------+------------------------

              (0)               D                (0)   (0)
         (0)   |                           (0)    | 1  /
           \ 1 | 1                           \    |   / 1
            \  |                              \ 1 |  /
   (0)  (0)  \ |                               \  | /  (0)  (0)
     \ 1 |    \|                                \ |/    | 1 /
      \  | 1  (2)                                (3)    |  / 1
       \ |     |                                  |     | /
        \|     | 3                                | 4   |/
        (2)    |              (0)                 |    (2)
          \ 3  |        (0)    |            (0)   |    /       (0)
           \   |  (0)     \ 1  | 1 (0)        \ 1 |   /         |
            \  |  /        \   |   /           \  |  / 3    (0) | 1
             \ | / 1        \  |  / 1           \ | /         \ |
              \|/        7   \ | /               \|/         1 \|
              (14)------------(6)----(0)         (8)           (2)
                            1 / \  1
                             /   \
                           (0)  1 \
                                  (0)

   FIG 2.  Cross-server delegation in a mirror. A candidate on a mass
           server (B) has cast a vote using a communicative server    
   (A).  The mirror (D) recognizes this as delegation and carries the
   6 votes of the first candidate to the second candidate for a total
   of 14.


Note that up to three servers are involved.  A delegate is both a
candidate and a voter.  Here the role of candidate is provided by the
mass server (B) on which the delegate continues to receive her
original votes, while that of voter is provided by the second server
(A) on which she casts her own vote.  The third server (D) is the
mirror that combines both roles and reveals her as a delegate.  If the
mirror were to provide its own voting service, as well, then it would
be more convenient to cast the delegating vote there directly. [6]


COUNTER-MONOPOLY EFFECTS OF MIRRORING
-------------------------------------

If the benefits outlined above are roughly accurate, then the
counter-monopoly effects are clear.  Votes will shift in favour of
preferred candidates regardless of their home servers.  Users will
tend to migrate to those servers that provide a convenient combination
of both mirroring and voting services rather than relocating with
every vote shift.  But any vote-server is capable of hosting a
mirroring service and sharing equally in the full extent of voter
participation [8].  A new server without any actual users, for
example, would still have all the voters.  Such an advantage cannot
easily be dispensed with.  Network effects are likely therefore to
entrench vote mirroring as a practice; and with that, no imbalance in
the distribution of users could ever tilt over into a monopoly [9].


   Table 2.  Non-defences against mirroring.

   -------------------------------------------------------------------
   Defence                  Why it is likely to fail
   ----------------------   ------------------------------------------
   Copyright                Non-creative factual information is exempt

   Closing data APIs        Data scraped from user interfaces

   Opaque client platform   There is none, all can legally be hacked
   -------------------------------------------------------------------


Copyright
---------
This defence is ineffective because non-creative, factual information
is generally exempt from copyright protection.  A vote conveys the
simple fact of assent or agreement and people are free to inform
others of that fact.

Closing data APIs
-----------------
Closing the data APIs is an ineffective defence because the data
necessary for vote mirroring can instead be scraped from a user
interface.  This defence would have to be combined with an opaque
client platform (below) in order to be effective.

Opaque client platform
----------------------
This defence is ineffective because no such platform exists.  The only
opaque platforms with sufficent coverage to support a monopoly bid are
Flash and Java, both of which have open source implementations on the
client side [10].  An open source client can always be hacked and
internally scraped.


   Table 3.  Defences against mirroring.

   ------------------------------------------------------------------
   Defence                 Cost
   ----------------------  ------------------------------------------
   Secret ballot           Lower credibility of results,
                           De-socializing the medium
   ------------------------------------------------------------------


Secret ballot
-------------

In this defence, the identity of the voter is kept hidden.  This is an
effective defence because votes cannot accurately be imaged without
knowing the voters' identities [7].  One of the costs associated with
this defence is a lower credibility of results.  There are methods of
verifying private votes, but none is so simple and credible as the
method of full disclosure available to public vote-servers.

Another cost is the de-socializing of the medium.  This cost is more
serious, because the social attractions of a medium are largely
eliminated if the participants can no longer identify each other.  It
is unclear therefore how this defence could be reconciled with a
competitive strategy, especially one in which the goal is a monopoly.


CONCLUSION
----------

There is no viable defence against vote mirroring.  The immediate
effects of mirroring are sufficient in themselves to prevent the
formation of a monopoly in voting services.



  [0] A working draft of this essay is posted at
      http://zelea.com/w/User:Mike-ZeleaCom/Vote_mirroring_as_a_counter-monopoly_measure

  [1] Public telephone networks are an example of a social media
      monopoly.  Online examples are Twitter, Facebook and Skype.
      Email is a counter example.

  [2] Vote mirroring is the invention of Thomas von der Elbe.  See the
      original description:
      http://zelea.com/w/User:ThomasvonderElbe_GmxDe/Vote_mirroring

  [3] On the distiction between communicative and mass voting, see:
      http://zelea.com/project/votorola/d/theory.xht#medium

  [4] How a candidate may be both a person and a text is described at:
      http://zelea.com/project/votorola/d/theory.xht#candidate
      http://zelea.com/project/votorola/d/theory.xht#action-norm

  [5] Vote mirroring depends on knowing whether two sets of votes on
      otherwise distinct servers are concerned with a common issue.
      When they are, the candidates receiving the votes may faithfully
      be represented as rivals in a common pool of voters.  The method
      of determining this on a case by case basis is still under
      active development (mid-2011), but probably the users of the
      mirror are the best qualified for the decision work, as they
      have the most at stake.  In that case, it would have to be
      accounted as a cost of using the mirror.

  [6] If the vote were cast instead on mirror D, then the total for
      the end candidate on server A would have remained 7.  It would
      be incorrect in any case, as the results calculated on all the
      non-mirroring servers (A, B, C) are incomplete.  None of them
      takes into account the full set of votes on the issue.

  [7] Vote mirroring depends on shared standards of voter
      identification and authentication.  These may be grouped
      together under the category of "voter registration".  Voter
      registration differs from user registration in tying the user's
      identity to a physical place of residence, among other
      real-world properties.  The purpose of voter registration is not
      only to establish the user's eligilibility for voting on local
      issues, but also to prevent the user from multiplying his votes
      under false identities (sock puppets).  The dependency on voter
      registration is not particular to vote mirroring, but is shared
      by voting in general; vote mirroring need only be sensitive to
      the various methods of registration employed on the
      vote-servers.

      The dependency is moot, however, because no public vote-server
      currently implements and enforces voter registration.  Users on
      any server may employ sock puppets if they wish.  Vote mirroring
      might make the practice somewhat easier to get away with, but
      the incentive and opportunity to exploit this at an effective
      scale would depend on vote mirroring itself becoming entrenched
      as a practice.  In that light, the problem goes beyond the scope
      of this essay.  But the solution is simply to enforce voter
      registration when it becomes necessary.  The costs of that
      enforcement will be easier to bear after the problem has exposed
      itself.

  [8] Glossed over in the main text is the requirement of shared
      standards for the replication of the mirrors themselves.
      (Thanks to Alex Rollin for pointing out this omission.)  Such
      standards include issue identification and voter registration
      [5][7].  It may be enough to say that the requirement is not
      absolute.  Having two sets of standards that are mutually
      incompatible would not result in the fragmentation of the
      mirroring inter-network into two halves, but only in the
      creation of two inter-networks that were layered, as it were,
      one atop the other.  Each would be complete in itself and would
      reveal the full complement of participants.  The individual user
      would then have a choice of which to frequent.  In other words,
      a mirroring inter-network may itself be mirrored, and it follows
      that none could ever be locked in by network effects and imposed
      as a kind of monopoly solution.

  [9] Crucially vote mirroring unbinds the role of participant from
      that of user.  Only the former is subject to network effects,
      and unbound it can no longer reinforce an imbalance in the
      latter.  A user is free to choose her own vote-server based on
      quality of service, but that choice will not affect the fact of
      her broader participation and thereby the choices of others.

 [10] Java itself is open source.  For Flash clients, there is
      Lightspark. http://sourceforge.net/apps/trac/lightspark


-- 
Michael Allan

Toronto, +1 416-699-9528
http://zelea.com/



More information about the Air-L mailing list