Re:Where has all the productivity gone?
by Jtsummers on Saturday July 31, 2021 @12:14PM
(#28019402)
You Can't Improve What You Can't Measure Disease
YCIWYCM is a great way to kill productivity and performance improvement efforts when taken too far. Sometimes, you've just got to go with your gut. The amount of effort needed to make measurements, especially for less tangible knowledge work, can kill productivity overall.
Often, the measurable parts are at the conclusion of a particular effort, the total time it took to get there and the value of the output. But it's not necessarily possible to meaningfully measure the internal process of reaching it when (again, knowledge work specific) half the time may be spent thinking about the thing. How do you measure that? How do you indicate progress when you're one "aha" moment away from success but can't force one to occur? Synthesis, what most programmers do, the combining of ideas to form a new theory [Naur1985] is hard to do mechanically. It can be done, for a lot of work, but not necessarily for everything. But because it's not a purely mechanical process (in contrast to a lot of other business processes or physical manufacturing work where you have more mechanical work and less synthesis) it's hard to identify useful metrics and collect them.
When the work is more mechanical ("We've done this before, but...") then you can start finding more places to measure and gain useful insights. But the more novelty that occurs, the more you need to collaborate with each other or spend time in thought, the harder it is to find useful measurements.
YCIWYCM is a great way to kill productivity and performance improvement efforts when taken too far. Sometimes, you've just got to go with your gut. The amount of effort needed to make measurements, especially for less tangible knowledge work, can kill productivity overall.
Often, the measurable parts are at the conclusion of a particular effort, the total time it took to get there and the value of the output. But it's not necessarily possible to meaningfully measure the internal process of reaching it when (again, knowledge work specific) half the time may be spent thinking about the thing. How do you measure that? How do you indicate progress when you're one "aha" moment away from success but can't force one to occur? Synthesis, what most programmers do, the combining of ideas to form a new theory [Naur1985] is hard to do mechanically. It can be done, for a lot of work, but not necessarily for everything. But because it's not a purely mechanical process (in contrast to a lot of other business processes or physical manufacturing work where you have more mechanical work and less synthesis) it's hard to identify useful metrics and collect them.
When the work is more mechanical ("We've done this before, but...") then you can start finding more places to measure and gain useful insights. But the more novelty that occurs, the more you need to collaborate with each other or spend time in thought, the harder it is to find useful measurements.
[Naur1985] https://pages.cs.wisc.edu/~remzi/Naur.pdf