1.5 KiB
id, title, tags, daily
| id | title | tags | daily |
|---|---|---|---|
| 2026-05-22T11:57:18-0400 | 2026-05-22 11:57:18 | 2026-05-22 |
2026-05-22 11:57:18
ConEst's Purpose Considering its Accuracy
I removed 2026-05-21_16-18-20's plot because I screwed up the error calculation. After more rigorous investigation and more recent project data (the same new bids are based on) I'm happier with this one.
!
The plot suggests that conest's role is not in decreasing cost uncertainty, because it does not appear to do so more than marginally.
I'm lead to speculate that our purpose is in determining crew loading, which is cool with me, labor is the more interesting part of construction-estimating, however our methods are not well suited to the purpose. After all, the majority of it is calculated in ms-excel. See pdi-labor-plan.
A less cautious man than I might wonder if the negative bias apparent in Bid and ConEst is indicative of systematic incentives gaming, since if it were undesirable we could change nothing of our process and simply add contingency to address it.
Our use of "WBS" is technically inaccurate. The pdi-wbs is more accurately called earned-value-management#Schedule, which is the document which defines the planned value of each task. On the other hand, project schedule is a terrible name for a specific document, so I don't blame us for stealing WBS.