Peer teaching website

From OLPC
Revision as of 02:19, 23 December 2006 by Sj (talk | contribs) (peer teaching)
(diff) ← Older revision | Latest revision (diff) | Newer revision → (diff)
Jump to navigation Jump to search

A peer teaching website could be a shared website for peer evaluation and accomplishment that many different applications, courses, and tests could reference.

This page outlines a framework for multiple distinct websites and communities, each organized around some topic. The topic might be a domain of knowledge (math, writing, gardening, etc), or might be geographic or organized around just about anything.

Individual domains

In this domain there are a series of tasks presented, at increasing levels of difficulty. For instance, a botany website might have tasks like:

  • Take (and upload) a picture of a plant around your house
  • Take a picture of both a deciduous and coniferous tree, and label each one
  • Do the colored water carnation experiment (cut the stem in half, etc)
  • Estimate the ratio of trees to people in your neighborhood
  • &c.

Generally the tasks are organized serially, and you go through them one by one. In some cases the task might take some time to complete (e.g., actually grow a plan), so serialized tasks wouldn't be good -- there should be multiple tasks at some levels.

Each task results in the student reporting something on the website. In some cases it might be the answer to a problem (e.g., on a math-related website, you would give the correct answer). In other cases it might be evidence that you did something (take a picture of the carnation). In still other cases it might be a story about the task -- a log, or just a report. Or even just a claim that you did as the task requested, and the site trusts you not to lie.

The evaluation is always done by another member of the site. This is part of what allows for flexible tasks and flexible goals, because it is always a person doing the evaluation. Even when automated evaluation is possible, the website doesn't do anything to support that. The website *can* provide source material for the child to work from, and that material can be self checking (primarily dependent on the activity). But there is no specific support for this.

As a result, incorrect answers or confused attempts at a task are likely to show up on the site. This is a feature. This gives an opportunity for other children (probably older children) to give feedback and help, and try to correct the problems. Short of claiming to have responded to a task, a variety of these help mechanisms should be allowed -- each task should have a talk or forum page attached. Task-related chat should be encouraged, including possibly chat rooms or simply IM-style chat.

Once a student has completed a task they can see all the other solutions to the task. They have also raised their level on the system. The direct mapping of levels and tasks would be a local policy for the website -- there may be a combination of score received from tasks, tasks that are blockers to completion, and non-task website participation. There is no single score attached to levels (unlike RPGs), but instead each level has its own criteria.

As a child goes up in level on the system they can check tasks below them (perhaps with a buffer to ensure mastery, e.g., a level 6 user can check level 4 problems). At higher levels you can also create new tasks for other users. Peer review may take place for some of these actions, particularly creating new tasks (checking for accuracy alone is important enough to require peer review). At lower levels children can annotate tasks that they've solved to provide links to extra learning material that might be useful, or perhaps hints if they are not too leading.

Not absolutely everything will be based on peer evaluation. For instance, it might be required to get past a certain level that you actively evaluate a certain number of tasks, or provide help on some level. This can be scored in an automated way, though ultimately the aggregated data from that automated scoring should be evaluated and more importantly acknowledged by a real person.