black_pignouf wrote:Maybe I'm stupid, but I don't understand the 1st graph.
Shouldn't the automation curve fall down to zero in order to save time, or at least represent less work than the original task?
Shouldn't the two curves cross somewhere?
If automation isn't worth it in theory, it sure won't in reality.
QPhysics wrote:Top graph, "Work on original task" should have been a horizontal line.
As drawn, the automation does nothing for saving time at any point in the process.
Hm, a few things:
- I think part of this is a weakness of labelling: "automation takes over" is not marking a line, it's marking an event, the drop of the two work lines.
- Work done by the automation isn't listed on the graph.
- Here's one I disagree with, but it looks like in the second graph the function for the "work on original task" is work needed, with the dashed line indicating only part of it getting done, and the vanished line indicating none of it getting done.
With those givens, the "theory" graph matches my experience with automation, where the "work on original task" tapers off and is then eliminated by the automation, and the "writing code" is constant.
Now, there's also a concern with what to automate, how to automate it(as BAReFOOt said), and thinking it through up front; I think of this more as addressing strip 974, where if the problem is up now, then solve it by hand before working on the general case.