"Social-computing" for content creation
making extreme-effort content plausible

The content this site envisions, is currently dauntingly expensive to create. Long term, the answer to "how can we afford it?", is tech change.

Technology matters. Consider teaching scale. With improving tech, a laboriously created 1957 book, and 1977 movie, could now be high-school projects. With comparable effort, we could now do much better.

The cost of content creation will decline dramatically with the maturing of "crowd-powered systems" (aka "social-computing", aka "human-computation" - as of 2012, the name is hasn't settled yet).

What are they? Start with complex process flows, containing both human and computer components. As in a corporation. But online, with intense machine learning and statistical oversight. Permitting you to decompose problems, and assemble processes, much more powerfully and flexibly.

With the potential to profoundly alter content creation costs. As with systematic reviews, below.

Related areas

Systematic reviews

A systematic review exhaustively summarizes and integrates research literature to address some question. Most commonly in medicine. They are very expensive. Doctor time is scarce and costly. Reviews are quickly out of date. Updating them is expensive, and "freshness" targets are missed by years. Current work includes doing machine learning, with selective expert use, to prefilter literature with negligible false negatives. Future work includes using a mix of experts, of varying expertise and cost.

Other fields would benefit from systematic reviews. They're more common in medicine, because of cost, and their high value there. If costs drop dramatically, use can spread. To education. And content creation. The bottleneck for the kind of content I'd like to see, of scare and expensive experts, looks a lot like the challenge facing systematic reviews.

Research-paper blog posts

Creating blog digests of research papers with an unskilled crowd. CMU?

Grading exams

Candy machine feeds students to grade CS 101 exams. Umati (CHI'12 paper). Hartmann, Berkeley.

Order-10 redundancy is sufficient to exceed expert performance.


A word processor that uses paid micro-contributions to aid writing tasks such as text shortening and proofreading. Soylent, Bernstein, MIT.


Organizing crowds to quickly answer questions (for the blind with an iphone camera), and to appear as a single "person" you can talk with. Bigham, Rochester.

Imagine students being able to 411 a virtual teacher, research librarian, or domain expert. On demand. With plausible cost and pragmatics. And content being able to depend on it.


The subfield doesn't have a good survey paper yet (as of 2012-11). I usually suggest people start with Umati, to see the concepts in play.

Michael Bernstein

Crowd-Powered Systems [PDF]   (PhD Thesis).

Björn Hartmann

Jeffrey P. Bigham

Human computation: a survey and taxonomy of a growing field [PDF]   (2011) Ben Bederson.

less surfable...

CMU Social Computing Lab

Center for Evidence Based-Medicine @ Brown.

Spun off from Tufts (2012). Medicine is where the money and thus progress is, but they're very interested in applying these techniques to other sciences, education research, history, etc.

Related topics

Making the creation cost plausible.
Because if you can't afford to build it, its properties don't matter. Possibilities include: a motivatingly big payoff; better infrastructure with existing incentives; crowd-sourcing tech; and oriented incremental steps.
Crowd-sourcing content creation.
Existing communities, incentives, and tech, might support much greater collaborative content creation.
Wikipedia opportunities.
WP has untapped potential for improving science education content. Research groups have wonderful content which isn't online solely because no-one has asked. And a more intensive use of Talk, User, and project pages might permit more creative and correct content creation, without changing WP:No_original_research policy.


Does a good intro to the area exist yet? Any surveys? Else, sketch some key concepts, like expertise from redundancy. Link to fun papers for better surfing. Once the cost page is live, move the cost intro there.

Page history

Mark it up No JS?
Email me

Fb share