Stringency (RE: 12 problems)
Natale, Robert C (Bob)
bnatale at lucent.com
Fri Jan 17 13:06:45 CET 2003
Hi,
jak said:
> Whoops! Cancel that analysis. If the number is "months per RFC",
> then it actually showed an *improvement* between 1995 and 2000.
Yeah, I had a hard time following your analysis in several respects.
In any case, what I think these numbers show is that we are almost
surely placing too big a workload on our "system" (by which I mostly
mean the *people* involved in quality assurance):
Year RFCs Average 'C/B' % Change % Change % Change
Delay RFC Count AvgDelay 'Productivity'
---- ---- ------- ----- --------- -------- --------------
1993 31 5.5 0.177 baseline baseline baseline
1994 80 8.4 0.105 158.06 52.73 -40.82
1995 61 10.5 0.172 96.77 90.91 -2.98
1996 96 12.4 0.129 209.68 125.45 -27.20
1997 103 14.4 0.140 232.26 161.82 -21.20
1998 147 19.6 0.133 374.19 256.36 -24.85
1999 157 19.8 0.126 406.45 260.00 -28.92
2000 158 24.2 0.153 409.68 340.00 -13.67
The "% Change" figures are relative to the 1993 "baseline" numbers.
So, while the "workload" has increased by over 400% and the delay by
well over 300%, the "productivity" index has remained relatively constant.
Of course, we all know there are factors missing from this analysis.
For one, we'd need to know whether the resources available to the
IESG, IAB, RFC Editor, Secretariat, etc., have increased proportionally
to the workload. I'll assume the answer is "no" here. Also, the
definition of "Productivity" here is very gross -- without factoring
in some meaning of "output quality", I don't consider it a very
meaningful measure.
I am sure all of that (and much more) can be debated in a variety of ways.
I am not one to shy from qualitative positions: What I see in these
numbers so far is (1) a system that is probably generating more work than
it has resources (especially qualified people) to handle well and (2) a
relatively constant number of qualified people working very hard to keep
the system afloat.
The numbers do suggest that workload growth may have topped off, but
I would say they are at a level that is significantly too high. Trying
to solve this problem by adding more resources *might* help, but the
root (I believe) is in reducing the workload on the front-end.
Cheers,
BobN
More information about the Problem-statement
mailing list