Recombinant Text

Michael Allan. 2007. Recombinant text., project textbender, release 0.2.2, file d/overview.xht,*

Recombinant text is a collaborative medium of design and composition based on a mechanism of evolutionary genetics. In theory, it works much like a system of natural populations, or evolutionary algorithms; but instead of running automatically, the mechanism is human controlled, and directed toward the design of cultural artifacts.

Implementations of recombinant text have been developed for the purpose of collaborative writing. In typical use, a text is replicated into a population of variants, one per author; authors swap fragments and arrangements of the text, from peer to peer; and genes encode a traceable record of authorship throughout.

Communication Pattern  Typical Characteristics  Technical Definition
Use Cases  Related Fields  Applications  Notes and References

Communication Pattern

Compared to other collaborative media, recombinant text has a distinctive communication pattern:

                               O  <---- O
     \       /
      \     /                    \
       \   /                      \
------>  O  <------       O  -------\------>  O
       /   \                \         \     /
      /     \                \         \   /
     /       \
                               O         O

    centralized                distributed
(Wiki and other media)      (recombinant text)

Unlike the centralized pattern of most other collaborative media (such as Wikis), recombinant text has a distributed, peer-to-peer pattern. Instead of a single copy, the text is distributed in multiple copies, one per author. Instead of pushing contributions to the center, authors pull contributions from peer to peer. As a result, the text diverges into multiple variants that co-exist side by side. Collectively, the text has spatial diversity. In statistical terms, it has a population.25

Typical Characteristics

The text exists as a population of variants, one per author.

This is a natural consequence of the distributed, peer-to-peer communication pattern (above). The resulting population forms as a collection of documents posted at various locations on the Web. All documents are related, either by common origin or by cross-transfer. Each is the working copy of a single author, who has independent creative and editorial freedom.

Authors swap fragments and arrangements of text, peer to peer.

This is how authors collaborate, by pulling contributions from each other. Different methods of transfer are possible. The most basic is like copy and paste — an author selects part of another author's work, and transfers it to his own. But recombinant transfer has advantages over ordinary copy and paste, because it is aided by genetic alignment, a feature that reveals patterns of similarity among texts, and helps to guide each transfer. As well, the source authors are automatically acknowledged in the target text, with each transfer.

Genes encode a traceable record of authorship, throughout.

Despite all the copying back and forth, the system remembers the authorship of specific contributions wherever they ultimately survive in the evolving text. An individual document is a patchwork of discrete genes, each of which encodes a piece of the document. In addition, each gene records its own origins and ancestry. The combined ancestries of all genes of a document are, therefore, a record of that document's authorship. When a piece of text is transferred to another document, a portion of authorship is transferred too. When that document is later subject to a genealogical trace, it will reveal the contributions made to it by all surviving transfers, and the historical authorship of each.

Technical Definition

Recombinant text is a collaborative medium of design and composition based on an underlying mechanism of evolutionary genetics. It has four structural and functional components: a genetic code, populations, mutation, and recombination.

A genetic code is a set of rules for encoding and decoding recombinant texts. Current implementations use a form of XHTML, extended with genetic structures. The most important structures are genes. A gene is a datum encoding a part of the text. Genes serve as:

A population is a community of individual members that exchange genetic information. The individuals are separate copies of the text. Each is a ‘working document’, associated with a single author or editor.

Mutation is essentially text editing. Each author has independent editorial freedom over his own working document. Consequently the population acquires variant documents, and the gene pool acquires variant genes. These variations accommodate diversity. They are, in turn, grist for the mill of recombination.

Recombination allows an author to improve a working document by transferring select bits of text from the documents of other authors. On the surface, recombination can be as simple as copy and paste; underneath, it involves the transfer of discrete genes.

Use Cases

Several approaches to recombinant text have been explored to date, each corresponding to a different use case, or mode of use.


This is illustrated by the following scenario: Suppose you are working on a text, revising it, drafting and redrafting, when you come to a stop at a particularly difficult passage. Something is wrong with it, but exactly how to correct it is unclear. Imagine at this point that you could look out through your working copy and into the population, where you would see an array of variations — alternatives of the same passage drafted by other writers. If one appeared to be an improvement, you could select it, adopt it as a replacement, and continue working... (Past development has focused on this approach.)


This is where you are reading another author's document, and discover a desireable piece in it. If you like, you can copy it, and paste it into your own working document; either as a new insert, or as replacement for prior text. And then continue reading... (The current implementation of textbender is focused on this approach.)


This is where you are reading another author's document, and discover it has a desireable quality, but it does not correspond to a simple piece of text that you could copy and paste; rather it is a complex, with parts scattered in separate areas. For example (if the text is a novel) it might be a new twist in the plot, or a new character introduced to the story. With appropriate tools, you could copy it as a whole, and ‘paste’ it into your own working document, more-or-less automatically. (This approach has yet to be implemented.)

              |  \
              |   \
              |    \
              |     \  complex-wide
simplex-wide  |      \
              |       \
              |        \
              |         \
              |          \


As suggested in the diagram above, the complex-wide case draws from the combination of the two basic cases, simplex-wide and paired-regions. In fact, an author equipped with nothing but the basic tools could discover a complex variation in another document, and then gather its scattered parts to obtain a new copy of it, thus effecting a complex transfer. Such a manual process would be feasible especially for a localized and tight-knit variation. But the larger the text, the more scattered variations are likely to be, and the more likely that specialized tools will be needed to transfer them. In addition to formalizing and automating the process, such tools could improve it, in particular by employing statistical methods across the population, both to search out new patterns, and to sharpen the resolution of each transfer. (The techniques for this approach are a topic of planned research.) Much of the implementation, too, could be built up by the extension and integration of the basic infrastructure and tooling that already exists.

Biological and Bio-Metaphorical  Wiki  Distributed Revision Control
Memetics  Rhapsodic Verse in the Oral Tradition

Biological and Bio-Metaphorical

The theoretical underpinning of recombinant text is evolutionary genetics — the synthesis of Darwinian evolution and modern genetics. Aside from its significance as a biological theory, evolutionary genetics has practical applications in fields such agriculture, engineering and art. In these fields, it underpins systems that combine optimization and search in detail, with creative composition and diverse expression in the large. Recombinant text is one such system, and Table 1 places it in context.

Table 1. Evolutionary genetic systems and their refinements.
  system or refinement genetic
selectand innovator selector
  natural populations nucleotide individual nature nature
    artificial selection nucleotide individual nature man
      genetic engineering1 nucleotide individual man man
      recombinant text data gene man man
      human-based evolutionary algorithms2 data individual man man
    interactive evolutionary algorithms3 data individual computer man
  evolutionary algorithms data individual computer computer

In this context, recombinant text is most clearly distinguished along lines of human agency. Each of these systems runs a mechanism of evolutionary genetics. But instead of running automatically, the mechanism of recombinant text is human controlled. It compares most closely, in this regard, with genetic engineering and human-based evolutionary algorithms. In all three systems, the innovators and selectors are people (see table): they mutate and recombine the genetic material to produce new phenotypes; and they select for those phenotypes in the population at large. But recombinant text differs in that it extends human control of the mechanism further, to the direct selection of genes and recombinant sequences, throughout the population. In fact, it extends so far that the detailed construction of every genotype of the population is under direct and exclusive human control.4

Another Collaborative Medium: a Wiki

As a collaborative medium, recombinant text is comparable to a Wiki. It has a Web-based implementation and a similar, reactive pattern of collaboration. But recombinant text differs in its distributed, peer-to-peer architecture (populations), formalized data units (genes), and semantic labels (genetic loci). These place it in the arena of Web 3.0.

Distributed Revision Control

As a collaborative medium, recombinant text is also comparable to distributed revision control. Both are applied to distributed text composition; both provide mechanisms to pull changes from document to document; and both maintain historical data on changes and ancestry. They differ in several ways, but most fundamentally in purpose. Revision control is designed to keep a team of collaborators in unison, so that everyone has the same changes, or very nearly. Change propagation (patch flow) depends on a common context (literally) between source and target documents. As documents become more diverse, it becomes more difficult to 'cherry-pick' changes, while maintaining diversity. Recombinant text, on the other hand, has a looser design, in which communications are context free, depending instead on a common meta-context (genome) between source and target. Recombinant text is therefore a better medium where freedom and diversity are important, as in creative composition, its primary application. It would be less suitable if consistency mattered more, as in software engineering, the primary application of revision control.


Recombinant text is also comparable to memetics. Both are concerned with the evolution of cultural artifacts, and both employ structures or concepts analogous to genes; but the analogy is loose in the case of memetics. Memes are quite different in their behaviour from genes: they have no loci; their copying fidelity is low; and they re-associate by blending instead of by discrete recombination.5 These differences detract from the comparison.

Rhapsodic Verse in the Oral Tradition

There is a notion too, that recombinant text may be a re-creation, in a sense, of rhapsodic verse in the oral tradition. Substitute the genes of a recombinant text for the formulae, epithets and stock phrases that were the raw materials of improvised recital — what Ong calls its 'prefabricated parts';6 and substitute the Web for the wandering singers who formed the collective memory and communication medium of the oral tradition; and it appears that recombinant texts are composed and re-composed from posting to posting, much as songs were once 'stitched together' from recital to recital. In this sense, a recombinant text re-creates an environment similar to the one in which the ancient epics evolved; a fluid dynamic and dialogue among artists, long since frozen by the invention of writing.


Literary composition is the only application that is currently supported in practice. This section also speculates on other, potential applications.

Design  Composition  Utopian Literature  Education
Law Making  Brainstorming  Decision Support


Recombinant text might be applied to the design of any artifact, provided it is composeable of interchangeable and recombinable parts, and provided the design itself can be expressed as interchangeable and recombinable parts. No general survey of such artifacts has been attempted, to date.

We divide recombinant design into sub-applications, in various domains. Some of these are detailed in the following sections:


Recombinant text may be applied to the composition of literary and musical ‘artifacts’. The beta target for textbender, for example, is verse and creative prose. For this application, the medium promises some unique benefits for artists.

Various other sub-applications of recombinant composition are discussed in the following sections:

Utopian Literature

Recombinant text is suited to utopian literature, particulary to the expression of what historian Howard P. Segal calls ‘serious utopian visions’. According to Segal, a serious utopian vision is not just the prediction or dream of an isolated individual, but serves as a reflection of larger society. And it functions best not as a literal blueprint for the future, but as a critical representation of the present, and as a possible motivation or guide for change.7

Recombinant text is suited to the expression of serious utopian visions, because it would amplify the qualities that distinguish them as serious: reflection, motivation and guiding. Expressed in the medium, a utopian vision would:

A utopian vision would reflect society more accurately if it were expressed in recombinant text. The argument here is that recombinant text is a social medium, one that assumes the structure and dynamics of the society in which it is embedded. Specifically, the structure of the medium is a population, one that mirrors the community of its authors. The composition of the text is driven by tensions among these authors that mirror similar tensions in the society. As the text evolves in response to these tensions, it is pulled into a shape that reflects the society in all of its similarity and diversity (and even in its contradictions).

A recombinant utopian text could change continuously as society changes. The medium is evolutionary, so the text-as-mirror would be carried into the future. Furthermore, the medium is optimizing, so the reflective power of the mirror would improve with time. By contrast, a utopian vision expressed in a traditional medium is static, and becomes difficult to interpret as it recedes into the past.

Recombinant utopianism could motivate society by continuous ‘contact’. Again, the medium is evolutionary. The text would never become outdated; it would forever be composed and read within the present, where it would have full effect. And that effect would gradually increase to a maximum over time, again because the medium is optimizing. By contrast, a traditional utopian text exercises its full effect on society just once, as though by a single ‘shove’ from the past.

Finally, a recombinant utopian text could guide society by continual ‘steering’. Its authors would effectively accompany the text into the future, where they could make continual adjustments in response to unforseen events. They could steer the text, and thus attempt to steer society. By contrast, a traditional utopian text is locked into a single direction, a single forecast, and it becomes less relevant as society diverges from it.


The most obvious application is text books and other course materials. Authors, instructors and students could collaborate in creating, refining and customizing them.

Recombinant text could also serve as a teaching aid for creative composition. Students could intercommunicate and learn from each other. Instead of focusing exclusively on their own work, they could see the work of other students, and engage in creative dialogues.

Law Making

Recombinant text could serve as a medium for the proposal, drafting and promulgation of statutory law. Legislators could employ it both in house and publicly. In house, it could provide a collaborative medium for drafting legislation. Publicly, it could provide a participative medium for opening the process to public comment and involvement. Ultimately, it could serve to transfer full legislative power to communities, allowing citizens to take responsiblity for all stages of civil law making.

Another medium that has been proposed for this purpose is a Wiki.21 Wikis have achieved prominence as a collaborative medium, particularly for the composition of compendia, such as encyclopedias. It is therefore natural to consider adapting them to other purposes, including legislative purposes. But could a Wiki be adapted to the realities of law making? As Peter Leyden, director of the New Politics Institute, explains:

When you have people with opposing points of view, Wikis can get to be a battleground, and it's hard to broker a deal in the same way that you can in the traditional process. That process of give and take is often how the real stuff of politics gets worked out.14

The communication conflicts that can turn a Wiki into a ‘battleground’ — and could wreck a legislative process — cannot occur if the process is mediated, instead, by recombinant text. Recombinant text is a collaborative medium, having the same structure as a community. It allows for the expression of opposing views, just like a community. Embedded in a particular community, it can serve the two-fold purpose of expressing that community's concerns, and collaboratively drafting its legislative bill.

In order to elevate the bill into law, the community would also require a voting system. The voting system we propose below is based on a delegate cascade.23 Its purpose would be to express trust within a legislative context, and to provide the basic currency for give and take, and brokerage. Together, this hybrid of two systems (recombinant text and voting system) is sufficient to institute an open legislature: a ‘community legislature’, housing a process of ‘community law-making’.

A community legislature would be representative of society, much as a traditional legislature is. Its definition (as we will see) employs a similar concept of constituencies. Its constituencies are finer scaled, however, and more dynamic. New ones can form on demand, and scale themselves to fit any group in the community. And voters can quickly shift their votes from representative to representative, so their support is never taken for granted. Because of these improvements, a community legislature would represent society more directly and accurately. This, in turn, would strengthen the authority of civil law, by removing it from state control, and basing it more directly on the higher authority of the public.

The public consists of private individuals brought together by the voluntary exchange of ideas to form a whole greater than the sum of its parts. Not only is it independent of the state, it claims a superior status for both quantitative and qualitative reasons: because its numerical strength allows it to claim to be representative of civil society, and because its insistence on free expression and public debate allows it to claim enhanced authority.15

At the end of this section, we will describe the pre-conditions and steps necessary for the institution of community law-making. In what follows, we outline how it could work, once fully instituted.16

  1. Any citizen could draft a proposal (bill) for a new law; or the amendment or abrogation of an existing law.
  2. Other citizens (drafters) could copy the bill, modify it, and thus create their own variants (drafts) of it.
  3. Each citizen would have a single vote per bill, which he might use to ‘back’ any drafter of the bill. A drafter could thus acquire a ‘constituency’ of backers.
  4. A drafter's vote would carry with it the votes of her constituency.
  5. A drafter who accumulated sufficient backing might submit her draft for ‘deliberation’ as a ‘final bill’. During deliberation, the text of the final bill would be frozen.
  6. Every bill would have a ‘null drafter’. A vote for the null drafter would count as a vote against the bill.
  7. At the end of deliberation, and provided it retained sufficient backing, the final bill would become law.
  8. The text of the final bill would be traced, and all of its contributing drafters would receive credit as legislators.

Any citizen could draft a bill for a new law; or the amendment or abrogation of an existing law. The bill would be formatted as an ordinary recombinant text, and posted on the Web.

Other drafters could copy the bill, modify it, and thus create their own variant drafts of it. The drafters would collaborate in this, using ordinary recombinant tools and techniques.

Each citizen would have a single vote per bill, which he might use to back any drafter of the bill. He would thus mark his preference for a draft by formally backing its drafter. He could back at most one drafter at a time. He would remain free, however, to withdraw his backing at any time, or to transfer it to another drafter.

A drafter could thus acquire a constituency of backers. A drafter with a constituency would effectively become a representative law maker. Her task would be to represent the views of her constituents (and thus to hold on to their backing) as she drafts the bill to completion.

A drafter's vote would carry with it the votes of her constituency. Initially, a drafter might back herself. But later, she might decide to transfer her backing to another drafter, especially one who had a larger constituency, and thus a better chance of making law. In transferring her vote, she would effectively transfer the backing of her own constituents. In this too, she would hope to represent their views (and thus to hold on to their backing). The variant drafts would thus become organized in a hierarchical representation of the diverse views of the community (bottom), and their ability to broker a consensus (top).

Any citizen might aquire a consitituency in this manner; the citizen need not be an actual drafter. Thus all citizens could participate in the delegate cascade. Citizens who were unsure which drafter to back, for example, could instead back a non-drafter — someone they know and trust — who would serve as their delegate, to represent their views.

A drafter who accumulated sufficient backing might submit her draft for deliberation as a final bill. This could occur only if the community acheived a broad consensus on a single top drafter. Her draft could then be submitted as soon as she judged it ready.

During deliberation, the text of the final bill would be frozen. Deliberation would commence from the time of submission, and would extend for a period of time, perhaps a month or longer. During this period, work could continue on other drafts of the bill. Citizens would remain free to transfer their votes to and from the final bill (i.e. the top drafter), and among the various other drafters, as usual.

Every bill would have a null drafter. A vote for the null drafter would count as a vote against the bill. The null drafter would exist prior to deliberation, as well. If it accumulated sufficient backing, it would indicated a consensus against the bill. In many cases, such a bill would never come to deliberation at all.

At the end of deliberation, and provided it retained sufficient backing, the final bill would become law. This would be automatic. No additional vote would be necessary, because all the votes would already have been cast, in the form of backing. The backing would simply be tallied one last time. Then the final bill would either be rejected, or elevated to law.

The text of the final bill would be traced, and all of its contributing drafters would receive credit as legislators. Regardless of the outcome of deliberation, the text would be subject to a genealogical trace, in order to reveal who had drafted it, and in what proportions. The state would then acknowledge them for their service as legislators, and reward them accordingly.

Some of the pre-conditions and steps necessary for the institution of community law-making are:

The medium and tools of recombinant text would have to be developed to sufficient maturity. For the most part, the tools would be the same as for other, non-legislative applications of recombinant text. They might therefore be proven in one of those applications, prior to any legislative use.

A voting system would have to be developed, and integrated with the medium. Project Votorola is developing a system that would be suitable for this purpose.

Legislative responsibility would have to be transferred in gradual stages. At each stage, the technology could be tested and refined, and additional responsibility could be transferred to the community. The stages might be:

  1. Rulemaking for non-legislative applications.
  2. In house, serving for collaborative drafting among traditional legislators.
  3. Consultative, as a medium to open the process to public comment.
  4. Participative, as a medium to open the process to public involvement. This might culminate in opening the legislature to citizen's bills, and to public voting.
  5. Transfer of full legislative power to the community. The legislature would formally transfer its constitutional authority to a new civil institution, an open legislature based on community law-making.

An open legislature is bound to be a popular idea. No democracy could easily deny it to the public once it was proven feasible.24 And, because it can be built on free and open technology, its proof may be short in coming. As Katherine Noyes summarizes:

Just as in the world of free software ... people need to be shown that the idea could work, they need to have a rough draft to dig in with, and they need to be willing to collaborate to make it happen.13


Recombinant text could mediate a formal brainstorming session.8 This approach might be more effective than the traditional, face-to-face method. Although the traditional method is widely used, studies reveal it to be ineffective compared with non-group, non-collaborative methods;10 and they attribute its failure to mechanisms associated with group dynamics. But most of these mechanisms would be inoperable in a recombinant process, or would affect it less.

In recombinant brainstorming, participants would each receive a copy of the text. In it, they would read the session rules, the topic, and lead questions. The session would then proceed in three phases:

  1. In phase one, each participant would work alone on his own copy of the text. He would append a list of his initial ideas, and draft a short description of each.12
  2. Next, the participants would work together using recombinant techniques to exchange ideas, generate new ones, and collaboratively compose their descriptions.
  3. Finally, the results would be gathered.

The benefits would be:

Recombinant brainstorming would be open to remote and asynchronous participation. It would occur online, instead of face-to-face. Ideas would be exchanged through the text, not through the open air. Timing would be asynchronous, so the session would be relatively long, and more time would be available during the process to reflect on the ideas.

Unlike traditional brainstorming which requires a trained facilitator to be fully effective,10 recombinant brainstorming would require little or no facilitation. Each participant would have an exclusive copy of the text in which to express his ideas, guaranteeing him equal ‘air-time’. Even without a facilitator to moderate the session, no idea would fail to find expression; as long as it had the support of at least one participant, it would remain visible to the others, and would show up in the final results.

Recombinant brainstorming would suffer fewer process losses owing to social-psychological and economic mechanisms. Such mechanisms include cognitive uniformity, social influence, evaluation apprehension, observational learning, distraction and free-riding.10 These loss mechanisms are described in further detail in Table 2. Generally, they are expected to be weaker in remote, asynchronous sessions, and therefore to have less effect in recombinant brainstorming.

Recombinant brainstorming would also be immune to production blocking. Production blocking occurs in face-to-face sessions. It occurs because only one person at a time can speak. Ideas formed in the meantime by other participants must wait to be expressed. And often, being forced to wait, they never do get expressed.10 This cannot occur in a recombinant session, because the medium is asynchronous; expression occurs in parallel, and is never delayed or blocked.

After the session has ended, one or more ideas might be selected and adopted as a basis for further action. The selection process would be similar to literary end-publication. A judge (in the role of editor) would select the winning idea, and employ recombinant techniques to compose a description of it (mostly by copying it from the population). In this final act of composition, the judge would behave like any other participant, but one who has deferred joining until the very end, in order to avoid biasing the session.

Finally, idea-contributions would be automatically traceable and creditable to specific participants. If the winning idea-description were subject to a routine genealogical trace, it would reveal each participant's contributions to it. These could be tallied and translated into idea-shares or credits, and then apportioned back out to the participants.

Table 2. Process gains and losses in face-to-face (FF) and recombinant (Re) brainstorming.11
Mechanism FF Re Hypothesis
Procedural gains
Cognitive stimulation + + New ideas inspired by ideas of others, affects both approaches.
Procedural losses
Duplication   - Idea duplication, affects only an asynchronous process (Re).
Cognitive interference -   Over-engagement in others' ideas, weaker in remote, phased method (Re).12
Production blocking -   Limit of one speaker at a time affects ideation, but inapplicable to asynchronous process (Re).
Social psychological gains
Task orientation   + Focus on task of ideation, instead of on other participants, stronger in remote, asynchronous process (Re).
Observational learning +   Imitation of performance (may also be a loss), weaker in remote, asynchronous process (Re).
Social recognition +   Desire for it motivating performance, weaker in remote, asynchronous process (Re).
Social psychological losses
Cognitive uniformity -   Going too much with the flow, weaker in remote, asynchronous process (Re).
Personalisation -   Over-associating ideas with personal matters between participants, weaker in remote, asynchronous process (Re).
Social influence -   Individual participants dominating others, weaker in remote, asynchronous process (Re).
Evaluation apprehension -   Fear of negative evaluation, weaker in remote, asynchronous process (Re).
Observational learning -   Imitation of performance (may also be a gain), weaker in remote, asynchronous process (Re).
Distraction -   Inapplicable to remote, asynchronous process (Re).
Economic gains
Drive arousal + + Group activity motivates and enhances performance, affects both approaches.
Formal reward   + Apportioning of idea-shares or credits, infeasible in traditional method (FF).
Economic losses
Free-riding, loafing -   Hiding in the crowd, impossible in distributed, text mediated process (Re).
Production matching - - Matching productivity to others in group, affects both approaches.

Table 2 compares hypothetical process gains (+) and losses (-) between the two approaches. If the hypotheses in the table are correct, then the balance probably weighs against the face-to-face method, and in favour of the recombinant method. The relative effectiveness and weight of the various mechanisms is not defintitely known, but evidence points to either production blocking losses, or some combination of the various social-psychological losses.10 None of these are expected to affect the recombinant method. Furthermore, it is reasonable to expect that the economic gains of formal reward would be among the most significant, perhaps even dominant in some situations; and these gains would be specific to the recombinant method. However, the recombinant method may also have specific losses (and other specific gains) that are yet to be discovered. These would have to be weighed in the balance. But as things stand, the recombinant method looks promising.

[tentative sketch]

Decision Support for Planning and Policy Making

[needs revision in light of the more recent, open legislative and electoral systems]

A recombinant text could host a decision support network for planning and policy making. The process might be roughly similar to a brainstorming session as above, but larger in scale, more complex, and longer sustained. The participants might be drawn from an urban community, a government department, or a business firm. The subject for decision might be a proposed transportation development, a foreign affairs initiative, or a business strategy. And the outcome would be a final plan, policy or course of action.9 Roughly speaking, this approach might combine the:

This has yet to be elaborated, grounded in theory, and explained in the context of each domain (social, organizational, business). In a business context, one way to explain it might be to compare the proposed recombinant approach to the existing Wiki approach [only partly understood by the author, and maybe inaccurate]:

A Wiki may be employed within a firm for internal communication purposes. In some cases [guessing here], the Wiki might serve as a node within a decision support network. In this capacity, the Wiki would serve two basic functions: information sharing, and information aggregation. People would use it to share information during the decision support phase; and, in the end, the aggregate result would be read from the Wiki, and a decision made.

One problem with the Wiki approach is that, in combining the two functions of information aggregation and information sharing, the first tends to interfere with the second. Before information can be shared among participants, it must first pass through a single Wiki node, and must therefore be aggregated. And often this aggregation bottleneck entails overwriting of other information, information that another participant is trying to share. Consequently, essential information is routinely lost by overwriting. This in turn leads to further bottleneck effects such as: editing wars; the suppression of diverse viewpoints; and the imposition of an artificial and premature consensus.

If such a Wiki node were replaced by a recombinant text, it would alter the decision support system, both structurally and functionally. Structurally, a single node (Wiki text) would become a network of multiple nodes (population of text copies). These would be distributed out, one copy per participant. The recombinant text, as a communication medium, would thus assume the detailed shape of the organization in which it was embedded.

Functionally, as a consequence, information sharing and aggregation would become disassociated. Intermediate aggregation would still occur, but only in a participant's own copy of the text, and only under that participant's guidance. Final aggregation into a single result would be deferred until the end of the process, just prior to a decision. Until then, information would continue to flow peer-to-peer, at the finest scales, unimpeded by the bottleneck effects of premature aggregation. Rather than being forced to follow each other through that bottleneck single file, as it were, participants could explore the problem space in parallel, and in different directions. The end result would be a collective perspective from different vantage points, and a broader understanding of the subject to support a final decision.

Notes and References


Although genetic engineering is not an evolutionary system in itself (only an isolated technique), it is applied to systems that are, at least in potential, evolutionary; namely populations in the laboratory. So it may be considered a technique in refinement of an evolutionary system. Its evolutionary application is more explicit in population genetic engineering. See Austin Burt, 2003: Site-specific selfish genes as tools for the control and genetic engineering of natural populations. (Proc. Biol. Sci. 270: 921-928.)

See also Austin Burt and Robert Trivers, 2005: Genes in Conflict: The Biology of Selfish Genetic Elements. (Harvard University Press. pp. 223-)


Human-based evolutionary algorithms are also called ‘human-based evolutionary computation’. See


Interactive evolutionary algorithms are also called ‘interactive evolutionary computation’, and ‘collaborative evolutionary algorithms’. They are used commonly in evolutionary art. See

See also Peter J. Bentley and David W. Corne, 2002: An introduction to creative evolutionary systems. (In Peter J. Bentley and David W. Corne eds., Creative Evolutionary Systems. Academic Press, San Diego.)


This comparison of evolutionary genetic systems is a summary. See the more detailed comparision in approach-simplex-wide.xht.


Richard Dawkins. 1982. The Extended Phenotype: The Gene as the Unit of Selection. W.H. Freeman, Oxford.


Walter J. Ong. 1982. Orality and Literacy: the Technologizing of the Word. Methuen, London.

On the other side of this correlation between genes and rhapsodic formulae is the suggestion that a kind of genetic, recombinant mechanism may have driven the evolution of pre-literate songs. No critical analysis of this has been attempted, but the leading quotations here will provide sources for anyone who wishes to pursue the suggestion further.


Howard P. Segal. 2005. Technological Utopianism in American Culture. Twentieth Anniversary Edition. Syracuse University Press.


The initial idea for the brainstorming application came from a discussion between Arnold Wytenburg and Michael Allan (Open Coffee meetup, Toronto, July 2007).


The initial idea for the decision support application came from a discussion between Martin Cleaver and Michael Allan (Open Coffee meetup, Toronto, July 2007).


Jonathan Davies. 2004. Wiki Brainstorming and Problems with Wiki Based Collaboration. University of York Computer Science Project.


Most of the process mechanisms, and their classification, are from Davies.10


Jonathan Davies suggested the utility of an explicit initial phase. Its purpose would be to ‘allow for more distinct and original brainstorming ideas to emerge.’ (Personal communication, 2007.)

Although an initial phase might be implemented in a face-to-face session (e.g. by writing ideas on paper) that practice is uncommon; wheras it fits naturally with a recombinant session. Even without explicit phasing, for example, a participant in a recombinant session would always be free to ‘tune out’ in order to concentrate exclusively on his own ideas. He might suddenly find he is ‘on a roll’, for example, and wish to avoid being distracted. No special facility is needed to support this, it is simply built into the method.


Katherine Noyes. Open Legislation, Part 1: What If Everybody Got to Write Laws? LinuxInsider, July 13 2007,


Peter Leyden, as quoted by Katherine Noyes in Open Legislation, Part 2: It's the People's Choice. LinuxInsider, July 16 2007,


T. C. W. Blanning. 2002. The Culture of Power and the Power of Culture: Old Regime Europe 1660—1789. Oxford University Press. p. 111.


The feasibility of community law-making, as originally proposed here, was discussed in several forums.17 18 19


Open source legislation, community law-making. TorCamp Discussion Group, August 2007.


Community law-making, a system based on recombinant text. APSA_ITP, mailing list of the American Political Science Association, Information Technology and Politics Section, August 2007.


—— E-Democracy and E-Government Researchers Network, DoWire Groups, August 2007.


Marcus Pivato. 2007. Pyramidal democracy.


Michael Evans. 2007. Toward (sensible) collaborative democratic policymaking? Submission to 24th Human-Computer Interaction Laboratory Symposium, Maryland.


Marko A. Rodriguez, Daniel J. Steinbock, Jennifer H. Watkins, Carlos Gershenson, Johan Bollen, Victor Grey, Brad deGraf. 2007. Smartocracy: social networks for collective decision making. 40th Annual Hawaii International Conference on System Sciences (HICSS'07).


Marcus Pivato describes a delegate cascade in which the top delgates would sit in a traditional legislative assembly. He also analyzes the potential benefits of the resulting ‘pyramidal meritocracy’.20

Rodriguez et al. describe a system of 'dynamically distributed democracy', based on delegate cascade, for the general purpose of collective decision making.22


It is likely that any re-founding of the legislative branch, whether upon community law-making, or upon another form of open legislation, would occur in the context of a re-founded executive (and possibly judiciary), and an altered political atmosphere generally. The reason is that the delegate cascade which forms the backbone of community law-making could also form the backbone of an open electoral system; and such a system could be implemented with relative ease. For example, see project Votorola.


A critical comparison of centralized and distributed collaboration (push vs. pull) is the topic of discussion in: CivicEvolution — pushing information up to the meta teams. Online Consultations, Dialogues, and E-Participation, DoWire Groups, February 2008. and


Copyright 2007-2008, Michael Allan. Permission is hereby granted, free of charge, to any person obtaining a copy of this software and associated documentation files (the "Textbender Software"), to deal in the Textbender Software without restriction, including without limitation the rights to use, copy, modify, merge, publish, distribute, sublicence, and/or sell copies of the Textbender Software, and to permit persons to whom the Textbender Software is furnished to do so, subject to the following conditions: The preceding copyright notice and this permission notice shall be included in all copies or substantial portions of the Textbender Software. THE TEXTBENDER SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE TEXTBENDER SOFTWARE OR THE USE OR OTHER DEALINGS IN THE TEXTBENDER SOFTWARE.