Imagine I’ve enlisted three people to write a handbook. Amy has an advantage over the others in finding and summarising research but doesn’t write the most elegant English; Ben is best at sourcing, adapting, and captioning images; and Chris, a grammar pedant, is strongest on copy-editing. A natural way to organise them is to divide them into different roles: Amy the writer and Ben the illustrator will pass drafts on to Chris the copy-editor.
Rather than labelling the workers, an alternative – more in the spirit of the DNA computer – is to label the work. The pages of the draft get tags to suggest "This needs updating", "This needs illustrating", "This seems finished: needs review," and so on. Tagged pages go into a queue, so for example Chris tracks what needs doing by looking at the "in need of a copy-edit" queue. Pages with no tags go into a "This needs a tag" queue. Rather than working after each other in sequence, each person works, when they have time, on the next appropriate item.
Let us also introduce microattribution: each page is labelled with who has worked on it and what they have changed. On paper, this could be done with different coloured pens: in the digital sphere this could be a database of contributions. Not only will it be clear that each finished page was worked on by Amy, Ben, and Chris, but each can demonstrate what they added and none has to worry about others taking credit for their work.
This approach is more responsive than rigorous planning. Chris has a comparative advantage at copy-editing over Ben the illustrator, but Ben is still quite good at it. So if the copy-editing takes longer than expected, the relevant queue becomes backlogged, and the workers notice that Ben can vary his work to reduce that queue. He knows that his work will be credited, not assumed to have all been done by Chris. Sometimes the workers will be in a mood to use their special skills, and at other times they might want to work on something more simple and repetitive: they just have to change the queue they work on.
To extend the metaphor, we could invite strangers into this workplace to see the drafts and their tags, track what they do and add tags if their contributions are problematic. Pedants who cannot stand to see misplaced apostrophes might correct some errors in the drafts, and even if the crowd’s contributions are low-value like this, the job still gets done more quickly. Crowdsourcing does not mean that all the work is done by the "crowd". Crowdsourcing can include engaging a wider public to collaborate with professionals or to add a particular value to work they are already doing.
This is the way of working enabled by a wiki. While by definition a wiki is a site that can be edited quickly and easily ("wiki" being a Hawaiian word for "quickly"), it is perhaps better seen as a technology for organising work, simplifying what would rapidly get out of hand if done with post-its and coloured pens. Tags, categories, queues, backlogs, and user contribution records all help to break down a high-level task ("create a handbook of publishable quality") into small steps, to distribute effort across those steps, and to track progress.