Norvig, Wiliam, and Selinger - keynotes from ALT's 2007 Conference. Plus useful interview with Norvig.

[Updated 13/9/2007 with links in notes section to BBC and Custom PC interviews with Peter Norvig, on 22/9/2007 with this link to a post by Ian Harford quoting in full Norvig's  reply to his question about user involvement in "ranking" web sites, and on 4/10/2007 with this link to Theorising from Data, which is a long presentation about how Google extracts meaning from data.]

You can access the full keynote presentations from the 4-6 September 2007 Association for Learning Technology* Conference as large files that can be viewed within Elluminate from the ALT-C 2007 web site. More open format versions due over the next week or so.

What I found most interesting was the dovetailing between Dylan Wiliam's absolutely gripping talk and Peter Norvig's more laid-back one.

Both concentrated in part on what can improve the progress that learners in formal education make, without increasing costs.

In different ways they each identified the key as "better formative feedback"; and I was left feeling that one particular enabler of progress would be some sustained collaboration between i) academic and industry researchers in the online testing domain, and ii) the engineers behind systems like Google's own translation tools. The reason I think this is that each are concerned with drawing out semantic meaning from large amounts of complex multifaceted data; and I would love to be a fly on the wall at a seminar involving people from both these worlds.

Two other aspects of Norvig's talk that struck me hard were this diagram from Adrian Sannier, of Arizona State University, which illustrates the rationale (on our partner's development curve not our own) behind ASU's decision to make very extensive use of Google Apps (for example Gmail) for its learners (and staff?), rather than providing its own:

Source: Peter Norvig's keynote presentation at ALT-C 2007

and this explanation of the practical meaning of the 2 standard deviations of performance difference that Benjamin Bloom found between learners taught normally, and learners given (unaffordable) individual tuition. (Use of mastery learning produced 1 standard deviation of difference.)

Source: Peter Norvig's keynote presentation at ALT-C 2007

The conclusion to Norvig's talk was that most education should be:

• centred on engaging, real-world projects;
• explored in teams;
• organised so that teachers are facilitators and can point to theoretical knowledge when it is needed  ("which is less than you’d think").

Finally that "Different students learn differently. But let them figure it out from the world full of information, don’t try to create materials ahead of time."

Note. I spent some of the time in the run-up to and during ALT-C organising media interviews with a tolerant Peter Norvig; and with luck these will appear during week beginning 10/9/2007 in Computing, on the BBC's Digital Planet [MP3 file of 10/9/2007 broadcast], as well as in Custom PC, which I am linking to despite its vile pop-up adverts. I also arranged a brief interview between Norvig and Rhona Sharpe (who with Frances Bell is newly appointed as a co-editor of the ALT Journal). When and if the latter is published it will have some news about a Google service that is imminent.

* I have half-time employment with ALT.