Wednesday, 14 December 2011

Automating value judgements

In my role as a professor of marketing and technology, I fully expect to have to assess and grade students. That is part of the lot of a teacher of any kind, from kindergarten onwards (although, arguably, the assessment may take different forms over the years).

As a conscientious educator, I am keen to ensure all the hard work of the students is also fairly and fully assessed.  There are possibly those, as rumour and legend would have it, who throw their papers down the stairs and then start grading those on the top step as 'A's, those on the 2nd step as 'B's and so on down to 'F's on the bottom step.

Tempting though it might be, I am determined to be as paperless as possible.  So far I have saved at least 700 sheets of paper in just one course (assuming double-sided printing) through not printing the submitted assignments.  And throwing a laptop or iPad down the stairs to achieve the same effect just wouldn't do. Think of the splinters.

However, grading 2000+ word assignments is time consuming.  If you do it properly, it involves reading each assignment at least twice, possibly more. Comparing papers to ensure the comparative grades in the class are fair.  Composing specific feedback (100 words minimum? 200 words optimal?) to help the student understand their grade and improve for the future.

Group work needs assessment and, by the nature of group work, involves fewer actual papers to read and assess. But it removes the individual assessment and is therefore subject to slacker students benefiting from the hard work of other team members.

There is no getting past it.  Detailed, objective and rigorous individual assessment is essential.  But how to do it efficiently?

My first thought was to build the most advanced AI computer in the world and get it to grade the computers.  As soon as the patent comes through I'll let you know.

My second thought was back to throwing everything down the stairs.

My third thought was carrying on as I have been and grading at 2am until I get through them all.

So my final thought seems the only viable option, other than me pulling an 'all nighter' or 7.  If Amazon, and every e-commerce site worth its salt, can tap into the crowd to assess and vote on how good (or not) individual books are, surely that would work for individual assignments from students?

All one would need would be a simple website, such as Scribd, where the PDF documents of assignments could sit, visible to the world.  The world at large could then read the assignments and grade them. Once a critical mass of assessment is reached (such as 500 votes), the average is taken as the final grade.  Space for comments would provide the feedback.

This is, in fact, I am certain, genius. There are only two obstacles to overcome:

  1. Getting that critical mass of people to read and vote on all the assignments in a short timespan; and
  2. Ensuring the students and their families do not vote for them.
If, therefore, you ever see such a business model come into existence, remember you read it here first.  Until then, it looks like I have an 'all-nighter' ahead of me....





Top image: digitalart / FreeDigitalPhotos.net