Part of AtsGoesExtreme and the AtsDiary.

-----

As suggested in ExtremeProgrammingExplainedEmbraceChange, I'm using a whiteboard with a few metrics on it to highlight the things that I think are important. My hunch is that it's a great form of DictatingByOsmosis, but as with all good DictatingByOsmosis techniques, its effects are too subtle for me to say for sure.

Right now, I'm most concerned with getting developers to be TestInfected. So I'm currently tracking the number of tests that have been written. I'm also concerned with the AtsLoadFactor, since both of the developers have the potential to be interrupted a lot. So I'm also tracking the load factor. And lastly, we need to calibrate our estimates, so I'm tracking how our IdealTime estimates on tasks compare to the actual amount of ideal time spent on the task.

These metrics are tracked on a large whiteboard that's placed prominently in my office. It is the first thing you see as you look in the door. (To accomplish this, I unhooked a whiteboard from the wall in another office and leaned it against the desk right across from the door.) In the corner of the board, I've got a highlighted message that reads: "For tracking only. ''DO NOT DISCLOSE!!''" The purpose of this message is to emphasize that these numbers are ''not'' official estimates, not to try to hide information from stakeholders.

Here's the current (10 March 2000) text of the board:

-----
	 * Functional tests: 1
	 * Unit tests: 10
	 * Commitment load: '''3.7'''	'''''Committed to 3.0!!'''''
	 * Actual load: 2.8
	 * Ideal estimates: 128%
-----

See AtsLoadFactor to understand what "commitment load" and "actual load" are, and why I'm tracking them separately. "Ideal estimates" is the amount of IdealTime estimated for a task vs. the actual amount of IdealTime used. So right now our estimates are high. That's not bad, but I'd like them to be in the 100-110% range, which is why I'm tracking it.

My gut feeling is that five numbers is the most I want to track, and may be too many. The developers are well on their way to becoming TestInfected, so I will probably remove those numbers soon. I might replace them with a %-complete and current completion date for the iteration or something similar. The problem with putting a date up there is that one stakeholder visits our office frequently. He might see the current estimated date and take it as the date we actually will complete, not as an informal indication of our efficiency. So maybe I'll use "current estimated total hours" vs. "committed total hours" or something similar.

-----

As of 20 March 2000, we've been having some schedule difficulties and everybody's been working really hard. So I've rearranged the tracking whiteboard to emphasize the progress we're making rather than the problems we've had. The only negative item still on the whiteboard is the number of hours spent on non-ATS work; that's because I still see that as a risk and I don't want people to think it's okay to start taking the developers off the project again.

The current text:

	* Time lost to other projects: '''21 hrs'''
	* Load Factor: 2.6 (Commitment: 2.6)
	* Calibrated Estimates: 95%
	* Progress: 57% -> 60%
	* Tests: 101 Passing: 100%

"Calibrated Estimates" is the ratio of estimate to actual. So our actual times are slightly over our estimates, but not much. "Progress" is in two parts; the first part is the amount of work we've completed; the second is the amount of time we've used.

There's also a section the whiteboard labeled "Success Stories." Under it, I've attached with magnets all of the stories we've completed.