Decision Making in the Public Sphere

This is a summary comparison of online decision-making systems in the public sphere. It covers systems that occupy roughly the same niche as Votorola. The common purpose is consensus building on political issues, such as legislative proposals and nominee officials. The users are members of the general public, independent of government and political parties.⁸

This is out of date. Here is yet another attempt to create a comparative overview, one that will keep itself up to date.

	1	x	?	A	?	B	B	A	0	0	development stage
7	2	n	n	n	n	n	n	y	y	y	open source
	3	c	c	?	c	c	c	d	d	?	structure of network
	4	?	?	?	n	n	n	y	n	?	scalable network
	5	?	?	?	c	m	m	p	m	?	structure of assent
	6	N	N	?	N	N	N	oN	N	?	issues of assent
6	7	n	n	?	n	n	n	y	n	?	actualization of consensus
	8	dV	dV	?	dV	dV	dV	V	dV	?	media integration
	9	?	?	?	?	n	n	y	n	?	UI localization
5	10	?	?	?	f	f	f	fR	fR	?	framing
1	11	?	y	?	?	y	n	n	?	?	discrimination
	12	?	?	?	y	x	x	n	y	?	forced consensus
3	13	?	?	?	n	n	n	y	n	?	voter authentication
	14	?	?	?	n	n	n	y	n	?	results verification
1	15	?	?	?	n	n	n	y	n	?	open aggregation
2	16	?	?	?	n	n	n	y	n	?	public assent
4	17	?	?	?	n	n	n	n	n	?	private assent option
M5		Op	Tr	Dp	Ci	De	Mi	Vo	Me	Co

Development Stage

Open Politics appears to be inactive. Its development plans for 2007 had included a voting system, to be coupled to its collaborative drafting medium (Wiki).

The Transparent Federal Budget is at an uncertain development stage. CivicEvolution is also at an uncertain stage. According to its home page, the system is “in the midst of an important redesign”.

Open Source

Structure of Network

Most of the systems are centralized. They are designed to provide services to voters from a single location, or from a private network that is centrally administered.

Metascore and Votorola are distributed. They are designed to be locally installed and self-administered by separate communities of users.

Scalable Network

A scalable network will serve any number of small regions, plus their aggregate whole. Thus it might be used by cities and regions for voting on local issues (separately), and national and global issues (together).

Votorola is currently the only system that is designed to scale, in this sense. It consists of a network of core sites that are organized in a tree of super-sites and sub-sites. Together they provide voter services to any number of regions, at various jurisdictional levels. Voter lists are aggregated from sub-sites to super-sites using database synchronization protocols.

Structure of Assent

In most of the systems, consensus is built up by mass assent. The number of voters for a mature consensus may be very large, but the preponderance of them are voting for a relatively few candidate norms (many to few). This is the typical pattern of conventional, mass elections.

CivicEvolution differs in building up its norms by component contributions, one piece at a time. The individual components are separately assented to, but the whole is not. Consequently the result is not a consensus by any direct measure.²

Votorola employs a delegate-cascade voting mechanism, giving it a peer-to-peer structure of assent. Even in a mature consensus, most participants do not vote directly for the consensus norm (or candidate). Instead they vote for their peers — that is, for ordinary voters like themselves.

Issues of Assent

All of the systems support voting for norms, such as laws, plans or policies. Votorola also supports voting for public offices, such as Mayor or Councillor.

Actualization of Consensus

What assurance that a public consensus will be actualized by government? What is the imperative, and what are the affordances for an elected official to follow the consensus? Only Votorola assures action, at present, because only Votorola enables consensus on the election of public officials.⁷

Media Integration

Most of the systems are a combination of special-purpose media: a voting medium to express assent/dissent, and a drafting medium to compose norms. Many of the systems also incorporate their own discussion media. The implementations are highly customized, and all of the media are tightly integrated.

Votorola is a pure voting medium. It is able to interface with a variety of other media (such as Wikis and mailing lists) for the purposes of collaborative drafting, discussion, and so forth. But the other media are hosted at external sites, separate from Votorola.

UI Localization

Localization adapts the user interface (UI) of the system to different languages and cultures. Votorola is currently the only system that supports localization.

Framing

Initial framing allows the voter to propose an issue and options, and not just to vote on them.

Reframing allows the voter to openly branch the issue and extend the options, as the voting proceeds. It occurs in the context of a single poll, aimed at a single consensus. It does not raise a separate issue, but it may open a new “channel” that alters the course of the old issue.

Discrimination

Systematic discrimination may arise from the employment of reputation systems. Reputation systems assign ratings to participants on the basis of peer evaluations. The ratings are not generated in the same way as real, social reputation. Social reputation has spatial and dynamic patterns of great complexity, which are generated and renewed in processes of everyday communication. A system generated rating will inevitably function as a substitute for the complexities of this communication. Technically speaking, it will therefore function as a steering medium (like money or power).

A reputation system is therefore liable to strategic use. Participants may employ it as a weapon against their political opponents. If successful, they could skew the results into a forced consensus. Or, the conflict might escalate to the point where the losers had no choice but to depart from the process. The remaining participants would then amount to a self-selected interest group, unrepresentative of the wider public. Any consensus they reached could no longer be characterized as a public consensus.

DemocracyLab relies on a reputation system. The details are undocumented, but it would appear to be liable to systematic discrimination.³

The Transparent Federal Budget (Tr) also relies on a reputation system. The details are undocumented. It may be that Tr is actually intended for private associations and interest groups (as opposed to the public sphere) because it appears to be oriented to lobbying, as opposed to direct legislative voting.⁴

Metagovernment may or may not be using a reputation system; they have yet to decide.

Forced Consensus

Ideally consensus is unforced. This requires that each participant have the freedom to put forward candidates and proposals, to discuss the merits of each, and to express assent or dissent. All of the participants must have equal voices, and no dissenting voices may be hidden or suppressed. When it comes time to vote, all of the proposed options must be on the table, and none may be eliminated from consideration. The measure of consensus is then the quantity of votes for each. If these requirements cannot be met, then the reported consensus is liable to be something of an artifact; arising, not from the choice of the participants, but from the constraints and pressures of the system.

DemocracyLab and MixedInk fail to express any consensus, at all. MixedInk eventually declares a winning proposal on the basis of the voters' ratings, but neglects to reveal the proportion who actually agreed to it, by voting in favour. DemocracyLab reveals the proportions, but fails to tell us how many people have voted. In both cases, the extent of the consensus is unknown. Was it just a few people? Or did everybody agree to it? Or was it something in between?

CivicEvolution filters out dissent by selectively elevating majority positions to the “meta-levels”, and thence to consensus. Dissensus is made impossible, and the resulting consensus is therefore forced.²

Metascore has a synthetic consensus facility in which ‘a synthesizing resolution effectively “steals” points from the resolutions that it synthesizes’.¹ The intent is to encourage consensus. But the component's formal loss of points has no basis in an actual loss of support; its supporters may continue to think that the component is the better proposal. In that case, actual support for the component is suppressed, and a false support is generated for the synthesis. The consequence may be a forced and artificial consensus.

Voter Authentication

Voter authentication ensures that voters are real people (not bots), and are in local residence (not outsiders). It also enforces a single vote per person (no sock puppets). Votorola is currently the only system that supports voter authentication.

The residence criterion is the same, in itself, as Murphy's principle of “standing”.⁶

Results Verification

Verification enables the public to cross-check the results of the system, confirming that the reported counts are accurate. Votorola is currently the only system that supports verification (though not yet implemented).

Open Aggregation

Open aggregation makes the raw voter input available for processing by any system/toolset. So all of the individual votes are available, and anyone may tally a count. So all of the individual voter registrations are available, and anyone may tally a voter list. Votorola is currently the only system that supports open aggregation, as part of its verification interface. (The implementation will not be efficient, however. It might be better if the raw data could be pulled directly from the voters, or indirectly via specialized mediators.)

Public Assent

Public assent means the opposite of a secret ballot. It's like a show of hands. Each vote discloses both the voter's identity, and her choice of candidate.

Private Assent Option

Public assent notwithstanding, this allows an individual voter to opt for a secret ballot. None of the systems supports this option, as yet. (Votorola acknowledges the need for it. See "Nyms", and earlier posts in the same thread tree.)

Notes and References

maintainer Michael Allan