Below was the first message in a string of exchanges about Burn-up
charts on the Silicon Valley Patterns Group mailing list. You can read
the entire thread from our archive with the following search link:
On 3/29/03 Russ Rufer wrote to the Silicon Valley Patterns Group:
I wanted to continue our conversation on inverting the Burn-down chart.
I'm attaching a sketch of our Burn-up chart. The idea to flip the chart
on its head came out of a conversation at the Silicon Valley Patterns
Group. ("Burn Up Chart.jpg" will also be on the files area of the
Silicon Valley Patterns Group mailing list).
I was expressing a problem I saw with Burn-down charts: it mangles two
measures so that neither can be seen clearly. During the course of a
project, both velocity and the expected feature set change. The way a
burn down chart is drawn, unless one of these is held constant, you
can't make inferences about the other.
Phil Goodwin introduced the idea of drawing the line from zero upwards
to show progress. I think I was the first to add a moving feature bar
to represent the movement of the goal the developers are chasing
(although Phil may have had this in mind when he started to explain his
idea). Throughout the release, extend the expected feature bar to keep
it current. Then as expectations change, alter the height of the bar
the team is trying to clear.
As Phil points out, when the target is to reach 0 (in the burn-down
chart) you psychologically think of working faster/harder to get there.
With the burn-up chart, it's easy and natural to think of cutting scope
to bring the target into reach.
Development velocity is now just the slope of the progress line, and
changes in velocity are changes in slope.
The resulting chart seems much more useful. You can see both of the
measures independently now. Assuming yesterday's weather, it's also
trivial to estimate how much closer you'll be to your (moving) target
after subsequent iterations.
Who else do you think we should be sharing this idea with?