Something I've only really noticed over the last couple of years is how difficult many people find it to 'have ideas', whereas I spend all my time trying not to get sidetracked with chasing harebrained 'how about this...' schemes... Maybe it's because I used to spend my days a child reading books by Edward de Bono and Tony Buzan?
Whatever...
A couple of the techniques I use more or less unconsciously are 'idea accelerators' and 'idea collider' (think: particle accelerators ;-)
The idea accelerator (or meme accelerator) is related to reductio ad absurdum - take an idea and run with it as far and as fast as you can (the important thing is to keep momentum up). This is a bit like a longitudinal brainstorm, where you just have to keep pushing a meme forward, rather than bringing in completely new ideas.
The idea collider (or meme collider) just requires you place two ideas/techniques/applications together (I try to hold them in the same breath...) and see what comes of them. Cross-fertilisation of ideas, sort of...
Anyway, anyway, there are some things that have been clunking around in my mind for some time that are just begging to be fused in some way, but I haven't been able to make time to think them through at all, let alone completely. So I'm hoping that by posting the germ here, I'll commit some dog walking/commute thinking time to the matter as a result...
The trigger for the post, by the way, is the Dell Idea Storm, a Digg like clone where users can "submit ideas for a new Dell product or service" and then vote on each other's ideas a la Digg. [I need to sort out an accented font, don't I?]
Anyway, anyway - the germ of the collision is this: take one prediction market framework (something like Betocracy for example), add possible successful course predictions in place of the bets and combine it with something like the search engine referrers stats from a web analytics package (like Google Analytics for example) to feed automated bets into the system...
Ok, ok, the thinking is garbled, and I'm not totally sure how the prediction market bit would work, but the Digg/Dell Idea Storm model shows the way forward. In an OU 'what courses should we be providing?' sense, imagine something like the following:
Example 1:
Example 2:
Example 3
As Example 2, except that votes are only cast if the proposed course would be returned as one of the top three (?) ranking results if it were included in a live/real search.
Result: a more robust version of 2, assuming people only click through the top few hits!
Example 4
As Example 2, except that votes are only counted if a student clicks through to request further materials (or even better, actually clicks through th register) on a real course after using search terms that would have turned up the proposed course (maybe as one of the top three (?) ranking results if you want to tighten things up further) if it were included in a live/real search.
Result: predicted potentially popular courses that people might actually register on float to the top.
What I'm leading towards here is the idea of 'invisible vote casting' (predictive marketing?!) used to help identify courses that might appeal to students ont he basis of their course descriptions.
I suspect the approach could also be used as a basis of forecasting the number of students who might register on a proposed course (search powered forecasting!)?
I'm still not clear on how the prediction market variant would work though! ;-)
Posted by ajh59 at February 21, 2007 02:17 PM