Thanks to John Udell for introducing me to screencasting, or "Making movies of software." I watched something about heavy metal and wikipedia and was totally hooked.

I've had it with powerpoint presentations of my research project, so instead I give you: totally unscripted, off the cuff, pre-beta and using a cotton wool ball in place of a real microphone filter.... a late night screencast introduction to concern maps in Visual Studio 2005. Download the screencast (wmv, 3.2mb).

I tweaked my RSS today to include full posts in the body. Apologies if this plays havoc with your aggregator.

On a related note, after making the change I re-built my weblog (asp.net) app using csc rather than VS.Net and the assembly weighed in at half the original size.

London, Feb. 19.

A couple of people used my Concern Map tool for the first time last week. Both of them (I said a couple) recommended creating an intermediate screen between the diagram view and the source code view. So that's what I implemented today.

The first image shows a concern map and result list, displayed after double clicking on a concept node (orange rectangle). The second (below) is a detail of the result list.

The screen shots above come from double-clicking the node labeled 'Read operations', associated with the code query: //methodCall[@type='System.Data.Common.DataAdapter'][@name='Fill'] ("match any call to a method Fill on type DataAdapter"). Two matches are displayed in the tool window to the right.

My System.Windows.Forms is rusty, and EnvDTE80.Windows2.CreateToolWindow2 is not all it is cracked up to be, so this took a lot longer than you might think.

Read The Wisdom of Crowds yesterday. The basic premise is that a diverse group will arrive at a better average answer to all sorts of problems than the best person in that group. It also highlights the dangers of small and/or homogenous groups for decision making, particularly when they are dominated by some authority figure. I wonder how many people are trying psychology experiments at work based on this?

Which reminds me... of a psychology experiment I took part in as an undergraduate for extra credit. The experimentor was testing a group of about 40 of us for "ESP", judged by our ability to stare at him for 5 minutes and then draw the shape that he was thinking of on a sheet of paper. He would then describe the shape, and see who was close. A couple of people were close, and one person nailed it. We then did something similar with numbers, and again someone got it right. The experimentor was suitably impressed, made some comments to that effect, and then solicited feedback from us on the process and what we thought of ESP in general. The whole experiment took about 25 minutes and we then shuffled out of the room to get back to... doing whatever undergraduate arts students do.

So what was this three card monte of an experiment about? I'm pretty sure the guy was trying to see if anyone was willing to correct him. Statistically it was likely that somebody would guess his shapes and numbers (and I'm sure he had some confederates just in case). But nobody did. Was he onto something, or should he have just tried engineering students instead?

In other news: unexpected web searches bringing people to this web site:

• colonel kurtz
• ixxys bagels
• katz diner

I packaged up the Concern Map editor and sent it to a few (hopefully) interested people today. It is pretty reliable, although the install is nasty - lots of manual steps.

I also extended the Code Query capability to include fields and properties. I can now write something like

//field[@name='Foo']
or
//class[@implements='System.Collections.IEnumerator']//property[@name='Current'].

I've got a positive response from people who have seen the screen shots, so I plan to make something available online when I get a proper installer written.

My concern map editor is progressing rapidly. I have now integrated the visual language with the IDE and code query functionality, so users can interactively navigate from model to associated code.

The editor currently allows users to use three different sorts of queries (called selectors) - regular expressions, file names and code queries (essentially an XPath over a represention of .Net source code). I'll probably add in some sort of desktop search and a richer code query (more like AspectJ) when I get some time.

Map, with four concepts

The same map, with selectors attached to some of the concepts

Double clicking a concept evaluates selectors and loads matching documents