Now obsolete, this is posted as an example of the simplex-wide approach to recombinant text.
The centerpiece tool is the editor. We use an XML editor for this demonstration, a Web launch version of Jaxe. To launch it, click on:
link and other content deleted
The editor should appear, complete with a pre-loaded text, as shown below.
|
The text contains XML elements. Many of the elements have locus attributes. For example, scroll down to the first line of the text, and click on the Attributes tab. You will see its locus attribute.
 |
If an element has a locus attribute, then it is a gene. In a typical genetic text, most elements have a locus attribute, and are therefore genes. This is important because genes have variations.
Altering the text of a gene creates a new variation of the gene. For example, go ahead and alter the first line:
div >My mind inclines [me] to speak of forms changed into new < div
To:
div >
 |
That is all you need at this point, just one new variation.
Texts are published to a population server. Once published, their variations become part of the gene pool, visible to other writers.
The server's Web publication interface has been removed. A new, mail-based interface will soon be added. The remainder of the demo will still work, much the same...
In the previous sections, you edited a gene and published [or would have published] the altered text. As a result, you contributed a new variation to the gene pool. In this section, you will do the reverse: take an existing variation from the pool, and merge it into your own text.
With the cursor inside of a gene, use the menu command Recombinant | Target-Gene.
 |
The gene is highlighted, and a menu bar appears on your desktop.
 |
Use its New menu to launch the Allele Table tool.
 |
The allele table (above) shows variations from the gene pool. It serves as a window into other writers' texts. Highlighted (top of table) is the reference variation. This is the variation from your own text. This variation also happens to exist in the wild population (twice).
A single variation may occur multiple times in the population, among the published texts of different writers, editors, etc., as shown in the Occurences column (far left). Your own variation is more popular than it was before, by exactly one copy, because you just published an extra copy of it (along with your text) a moment ago.
If you select a variation, the Copy button is enabled (bottom left of table):
 |
Press the Copy button, and the variation is merged into your text.
 |
Next, we demonstrate how to merge a whole canto, instead of a single line.
With the cursor just inside a canto tag (one of the parent div's) use Recombinant | Target-Gene to target it.
 |
The canto is a structural parent gene; the lines nested within it are leaf genes. Whereas a line/leaf gene encodes variations of characters; the canto/parent gene encodes variations of line arrangement. These variations are summarized in the table (above).
As you can see, most writers prefer the same arrangement of lines (1 2 3 4) as you. And everyone has the same two lines (1 2) at the beginning of the canto. Or rather, variations of the same two lines; because, although all writers have line 2, for example, that line has variations (as we saw previously).
But those other lines (b a c b, and h b) in the other arrangements, they are considered to be essentially different from 3 and 4. And none of them exists in your own text.
Unfortunately, none of this information is especially helpful in making a selection. The current tools are really only useful for recombining simple lines and phrases. Other techniques will be needed to recombine higher orders of variation, e.g. of cantos, paragraphs, and whole chapters. We'll come back to this point at the end of the demo.
For now, suppose you decide in favour of the last canto variation (above). Then you could select it, and press the Copy button, and it would merge into your own text.
 |
Notice that the variations inside the lines are unaffected, as far as possible. You still have what you chose earlier for lines 1 and 2.
The third and fourth lines remain a mystery. They are empty. But they are genes, so we can search for variations, and fill them in. In future, the tools might automatically do this, using common choices or best guesses; but for now, you would have to do it manually: Simply target one of the empty genes (as shown below); then select and copy — all as described previously.
 |
That completes the demonstration. As you can see, the text was not greatly improved. Hopefully we can try again later, when the tools are better developed. As they stand now, the tools are not quite adequate for this type of application.
The most obvious problem is selection of higher order variations. The allele table is useful enough for comparing among single lines, but it works poorly for larger, composite structures like stanzas, paragraphs, and such.
At first, this might seem to be a problem of visualization: How to show higher orders of variation, at a glance? But actually, the allele table is adequate for this task. For example, it does accurately show the variations of line arrangement in a canto. The problem is not so much the visualization, but the underlying information. Knowing the precise variations of line ordering does not help. Structural details of a variation, at any level, are insufficient criteria for the selection of one variation over another; something more is needed.
Even at the lowest level, a similar problem occurs. Although it may not be obvious from the demo, partly because it is a translation, when choosing a particular line variation over another, the variation alone is often an insufficient guide. It often happens, for example, that other parts of the text, at other loci, will bear strongly on the choice.
The general problem seems to be that the recombinant tools focus too narrowly on isolated structural elements of the text, and are blind to their stylistic and semantic interconnections. If the tools could show these interconnections somehow, it would be easier for the writer to choose meaningful variations, at any level; because a tentative choice of one variation would lead to another in another part of the text. In this manner, the writer could traverse the text and recombine interrelated structural variations as a cohesive set. Altogether, the set of interrelated variations would transcend their structural details, and perhaps embody a stylistic or semantic whole.
Work is under way to develop techniques of harvesting the necessary information from the gene pool. Tools will be developed that allow the writer to discover and navigate through broader patterns of interconnected variation. Ultimately, the goal is to enable the selective recombination of pan-structural, semantic variations, such as literary plot, or character. A first attempt at a solution is...
...a design sketch, unfit to post. But it did lead to the idea for the current, paired-regions approach; and that approach obsoletes the system described on this page.