Wednesday, September 19, 2007

Measuring progress

Earlier posts, in this and and another context, mentioned the difficulties found with earned-value analysis and management in a project. Essentially, some time line, that itself is a collapsed tree/graph, serves as a basis for a whole bunch of stuff that maps to the time line.

That is, we're talking an abstracted collection which overlays the activity landscape (itself an abstraction) plus provides descriptive information (some of it measurable) about the thing-in-itself that is the purpose for the project.

Model is a good word to use though there are many nuances to consider. Typically, for earned-value we need to track task progress (activity) and product completeness.

Even simple situations can be problematic depending upon several factors. An example is that it is a lot easier looking backward than forward. Yet, we've learned how to look forward in some cases, though we can try to push our extrapolation prowess too far. Also, as the cardinality of the task and product component sets rise, so to do the issues of benchmark and performance.

There are common themes that can be used for this discussion, such as those confounding some financial situations. That is, how does one mark, that is, evaluate for comparative analysis and for supporting decisions (necessary correction)? We've mentioned model; this capability is both boon and bane to the problem (and actually, the motivation for this blog). For some of the more recent financial instruments, this is not an easy task (marking to model).

Myth. That's a large subject dealing with mindsets and other phenomena that need some discussion.

Market. Unlike finance with its limited ontology and its dependence upon the dismal science, in engineering, we can actually build something and then test it. Success may very well revolve around techniques that quickly bring something to fruition which is then improved via an evolutionary scheme. This technique has found some use in software, its development driven by the requirements of the web.

Large products would need an entirely new variation on that theme. But, the notion that something must be seen in action is strongly inherent. A computer simulation versus prototype is still an open issue due to things like quasi-empiricism (will be looked at further).


03/28/2009 -- Mark to myth can be attributed to Buffet. Many claim that mark-to-market has exacerbated the current crunch (see WSJ Letters to the Editor - "Honest Accounting with Reasonable Write-Downs" Robert D. Arnott -- 3/27/2009). As said, the issues of this problem are not dissimilar from the earned-value problem of engineering. We'll revisit this issue shortly.

01/23/2009 -- Expect more effort in firming up the earned-value (and related) discussions.

06/12/2008 -- It's nice that engineering and finance have these parallels, thanks to economics being common between them, allowing more discussions of oops.

01/18/2008 -- See threads Finance as game and Wealth as measure.

Modified: 03/28/2009

No comments: