Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

The McNamara fallacy: Measurement is not understanding

mcnamarafallacy.com

Post reply on HN
Print view View original

31–40 of 223 posts

Re: The McNamara fallacy: Measurement is not understanding

#31

Post by peteradio »

jscode wrote: post #23

Quick story: I was the CFO for a company that sold to a private equity group (PEG). I took over as the CEO as the founders retired, leaving me to deal with the PEG. It quickly became apparent that the PEG managers looked at everything through the lens of an Excel spreadsheet. These guys were brilliant attorneys and analysts but lacked experience building businesses and managing teams. Ultimately, they couldn’t add mu…

I worked at a place with a lean 6 sigma certified specialist who towards the end of the companies doom effectively had the lead engineer cleaning out molding machines to track down every last tiny molded part that over the course of several years of continuous running had flung outside of its target. Same guy told me if the coke machine ever stole my change that he'd help me get it back from the vendor.

Re: The McNamara fallacy: Measurement is not understanding

#32

Post by jancsika »

jscode wrote: post #23

Quick story: I was the CFO for a company that sold to a private equity group (PEG). I took over as the CEO as the founders retired, leaving me to deal with the PEG. It quickly became apparent that the PEG managers looked at everything through the lens of an Excel spreadsheet. These guys were brilliant attorneys and analysts but lacked experience building businesses and managing teams. Ultimately, they couldn’t add mu…

> The moral of the story is that people with analyst mindsets play an essential role in our economy, but sometimes giving those people power over large organizations can have disastrous consequences.

I'm gonna cosplay an "analyst mindset":

1. Need to measure costs and benefits of slashing benefits/pay.

2. A benefit-- slashing benefits/pay allows us to hit some obvious financial goal

3. A cost-- Uh oh, I don't yet know how to reliably measure any of the costs.

4. Good analysts don't take action without measuring.

5. I'm a good analyst.

Conclusion: I cannot take the action of slashing benefits/pay

The only way to make it work is to add a step "3b: cherry pick metrics for the costs of slashing benefits/pay such that the phony metrics justify the decision management already wants to make of slashing benefits/pay." But now we've shifted from "analyst mentality" to "the mindset of the little Beetle-like bureaucrats described by George Orwell in 1984."

Re: The McNamara fallacy: Measurement is not understanding

#34

Post by ProAm »

eumoria wrote: post #22

McNamara basically admits this himself in the documentary The Fog of War: https://en.wikipedia.org/wiki/The_Fog_of_War It's worth a watch but it's very soft on him and his role. Still a very good documentary.

Great documentary. I felt it was a person trying to come clean and ease his conscious on his death bed.

Re: The McNamara fallacy: Measurement is not understanding

#35

Post by hummusandsushi »

dempedempe wrote: post #13

This article is good, but not great - the author only gives one example of how quantitative-only reasoning can be bad (the example of the poppies). The other "example" is just the US military lying. There are also no specific examples of non-quantitative reasoning that, if ignored, would be damaging. I feel like the Wikipedia article does a better job explaining this: https://en.wikipedia.org/wiki/McNamara_fallacy Al…

The webpage appears to be made to serve as a warning to data-driven businesses to not fall into the same perverse set of incentives that McNamara created, to encourage business managers to diversify their accounts of the success of their business beyond just the quantitative narrative.

Re: The McNamara fallacy: Measurement is not understanding

#36

Post by thaumasiotes »

aeturnum wrote: post #21
aeternum wrote: post #2

From a game-theory POV, all wars are caused by misunderstandings / partial information. If each side really knew the true strength of the other side, it would be clear which would win and therefore actual war is illogical and unnecessary.

This seems untrue. There are also genuine uncertainties about the future: things that neither side knows but will impact the outcome. Because of that, there is always a range of outcomes for a conflict, and so both sides would still have an incentive to carry out the war. Consider, for instance, if there are two countries (A and B) looking at conflict with perfect information. Given their relative strengths, the war…

> Consider, for instance, if there are two countries (A and B) looking at conflict with perfect information. Given their relative strengths, the war will cost both $1 billion, but A will win and reap $1 billion + $1 in plunder. In your model, B will always surrender and give A $1

In the perfect information scenario, B will surrender more than $1. B is better off at any tribute level up to $1,000,000,000.

Re: The McNamara fallacy: Measurement is not understanding

#37

Post by thaumasiotes »

asdff wrote: post #14
crazy1van wrote: post #11

Earlier quoted context omitted.

> If each side really knew the true strength of the other side, it would be clear which would win This is too simplistic. Often wars are heavily influenced by things other than just the strengths of both sides. For example - What if the ground hadn't been soaked from days of rain at Agincourt and Waterloo? What if the Germans hadn't held back their armor reinforcements for so long on D-Day?

They'd still lose. Wars are fought not from individual battles but from logistics. Modern military's don't even engage if there isn't a lopsided power imbalance favoring their success.

It only takes one side to engage.

Re: The McNamara fallacy: Measurement is not understanding

#38

Post by facorreia »

paulpauper wrote: post #5

I don't understands what the fallacy is. It is that an overreliance on data can overlook factors not in the data? Why is that a surprise. You would also have to show that not relying on data would generate better results.

My understanding is that the fallacy is cherry-picking a few data points that are easy to measure and then claiming that these are the most important data points, and then optimizing execution for moving them in the "right" direction. This is described in the metaphor of "searching under the streetlight". The key to this con is to act confident and to challenge any objections as lacking supporting data.

Re: The McNamara fallacy: Measurement is not understanding

#39

Post by scubbo »

Surprised not to see Goodhart's Law[0] referenced here - "When a measure becomes a target, it ceases to be a good measure". Not the same concept, but a related one (as is the Cobra Effect[1], of which the poppy-field burning is an example)

[0] https://en.wikipedia.org/wiki/Goodhart's_law [1] https://en.wikipedia.org/wiki/Perverse_incentive#The_origina...

Post reply on HN