With all the completion discussion, it occurs to me that, if we REALLY want
more accurate completion figures, perhaps we should have a way of weighting
tasks. For instance:
[_] %0 Repair receiver with broken dial string
[_] Remove cabinet screws and covers
[_] Restring dial string
[_] Retune for frequency accuracy
[_] Replace covers and cabinet screws
Yes, I know receivers don't have dial strings any more, but I needed a
gruelling, gruesome task that cannot be further broken down, and replacing a
dial string on an old capacitor tuned receiver is just such a task.
In the preceding, skinning the covers takes 2 minutes with an electric
Restringing the dial string takes an hour.
Retuning takes 5 minutes
Replacing covers takes 4 minutes.
So the total job takes an hour and eleven minutes, of which an hour is the
restringing task. It might be misleading for the overall completion to go
from 25% to 50% when one restrings the dialstring. More accurately, it should
go from 4% to 90% or something like that.
If VO incorporates task difficulty weighting, and I'm not for a minute saying
it should, that feature should be installed in such a way that it in no way
changes VO for the person not using the feature.
As long as we're on the subject of weighting, be aware that large trees are
weighted the same as small trees. In other words:
In the preceding, checking the single leaf from oneleaf would boost completion
to 25%, whereas checking a leaf from the threeleaves tree boosts it only to
8%. The way I work, I break things down to a level of obviousness, implying
that for me, all leaves are of approximately the same difficulty, always
assuming that I've fully decomposed them. Our current percentage completion
algorithm values an item with 50 leaves in its tree as equal to an item with
All that being said, I value the robustness and performance of VO much more
highly than a mathmatically sound feature enhancement, so unless the changes
I discussed are both simple and universally applauded, please forget I ever
discussed them :-)
On Wednesday 28 September 2005 07:59 am, Steve Litt wrote:
> Hi all,
> As long as we're on the subject of weighting, be aware that large trees
> are weighted the same as small trees. In other words:
That's why I think that a fraction would be a better representation. The
fraction numerator would be the number childless, tasks completed and the
denominator would be the total number of childless tasks.
www.noels-lab.com Chips, firmware and embedded systems
www.vimoutliner.org Work fast. Think well.