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

191–200 of 223 posts

Re: The McNamara fallacy: Measurement is not understanding

#191

Post by _xnmw »

rossdavidh wrote: post #96

I believe the McNamara Fallacy is a real thing, although I would say Goodhart's Law expresses the more fundamental issue. But, the examples they give (Vietnam and Afghanistan wars) are bad examples, because in both cases the issue was primarily not that the U.S. military was relying too much on quantitative metrics, but rather that it was being asked to do a job which militaries are not good at. In neither case, desp…

I do not believe that the war was "not winnable" in Afghanistan. The Soviets managed to defeat the Taliban precursors (Mujahideen), and left behind a Soviet-propped government that lasted longer than the Soviet Union did (3 years after they left). That's as close as you can get to an example of a "win" in Afghanistan -- so don't say that the war was not winnable.

Re: The McNamara fallacy: Measurement is not understanding

#192

Post by _8j50 »

The US did not lose the war against the VC, they stopped supporting south vietnam's war against the VC and withdrew because the war became unpopular. The distinction is important because ultimately it wasn't America's war and the VC did not overwhelm the US military into surrender. It was the civilian government of the US that directed an end of their involvement. If anything, China won a cold war against the US who feared a direct conflict and caved in.

With Afghanistan, the war against the taliban government was won 15+ years ago. The US thought it could play nation-builder like with Japan. The continued involvement in afghanistan was part of the ongoing "war against terror" to reduce the chance of afghanistan becoming a heaven for terrorists again. The war against afghanistan was won, what the US lost was a prolonged occupation. If you look at history, the victor would more or less burn the invaded country to the ground and enslave its people to avoid similar long term issues, one can argue it was both self-interest and an ideology of being humane to civilians that resulted in the waste of post-war afghanistan occupation.

Regarding metrics, I am fighting this right now. I found out what is expected of me early on and worked to extract that value out of myself and support my team but in the end my struggle is being able to explain metrics that reflect a years worth of work that only capture at best 30% of what is expected of me. Either way, can't complain, management isn't easy but meeting metrics is. what I dislike is if I have good metrics and stop there, it isn't enough. Metrics will always set a maximum ceiling on quality, if management can accept that I have no problem with metrics driven performance appraisal so long as it pays.

Re: The McNamara fallacy: Measurement is not understanding

#193

Post by erosenbe0 »

jscode wrote: post #135
andrei_says_ wrote: post #132

Earlier quoted context omitted.

If you are a good analyst with a bonus tied to quarterly profits increase, can you be still a good analyst? Do you sacrifice your own monetary benefit to save a company? But then what about your family? I don’t think analysts are not smart enough or not caring enough. But institutionalized greed corrupts the integrity of everyone.

Incentives matter and PEGs want companies to grow significantly over the life of their investment. The biggest reward comes from the future sale after you’ve increased the enterprise value 10x. I’m sure that people have made poor decisions for short-term gains, but the real issue is giving immense power to someone who has never actually had to build teams, ship products, or go the extra mile to make customers happy.…

Ehh some PEGs do not care about long-term. Often they want to do anything that can get them some lump sum dividends and bonus payouts. Cost cutting combined with a recapitalization cookbook, selling assets like real-estate, and then using the proceeds and borrowed cash for themselves. Terminal price won't matter much due to either the initial windfall or because someone else will be holding enough of the bag after recap.

Re: The McNamara fallacy: Measurement is not understanding

#194

Post by aiilns »

_8j50 wrote: post #192

The US did not lose the war against the VC, they stopped supporting south vietnam's war against the VC and withdrew because the war became unpopular. The distinction is important because ultimately it wasn't America's war and the VC did not overwhelm the US military into surrender. It was the civilian government of the US that directed an end of their involvement. If anything, China won a cold war against the US who…

> The US thought it could play nation-builder like with Japan. The continued involvement in afghanistan was part of the ongoing "war against terror" to reduce the chance of afghanistan becoming a heaven for terrorists again.

I would more easily characterize US external politics as nation-destroyer than nation-builder. Even if you disregard the evidence, it's quite a stretch to believe that the situation in Afghanistan is solely due to incompetence despite US good intentions.

> If you look at history, the victor would more or less burn the invaded country to the ground and enslave its people to avoid similar long term issues

Even if this has happened with some frequency I'm not so sure it's the rule; certainly there are many exceptions. In any case, it's no argument for doing the same.

Re: The McNamara fallacy: Measurement is not understanding

#195

Post by bambax »

scubbo wrote: post #39

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...

In a documentary about the VietNam war (I think it's Ken Loach's), there's an army guy who says this:

> When you can't count what's important, you make what you can count, important.

That's the McNamara fallacy in a nutshell. Kill counts started to become the end-all be-all of the war because bodies are the easiest thing to count, but it wasn't the relevant metric.

Tragedies are often the result of people persisting in something and failing to see the big picture. Even more so when numbers seem to tell them they're right.

It doesn't matter that the numbers are right if we're looking at the wrong numbers.

Re: The McNamara fallacy: Measurement is not understanding

#196

Post by davidhowlett »

greiskul wrote: post #178
Andys wrote: post #164

Earlier quoted context omitted.

Heh, I wonder if he would apply that to science as well?

Science advances in strange ways sometimes. Plate tectonics started basically with the anecdote of "south america really looks a lot like it fits into east Africa", but wasn't believed at first due to no hard data. Anedoctes are not data, but when you have some with no data, instead of ignoring it you should take it as a signal to go search if there is data to support it. Great scientists and mathematicians usually h…

I believe you mean West Africa.

Re: The McNamara fallacy: Measurement is not understanding

#197

Post by tremon »

bb88 wrote: post #86
jacobr1 wrote: post #46

Earlier quoted context omitted.

Having dealt with 2 PE exists, rarely is the proposal something as upfront and silly as slash everyones pay. That probably does happen for a company being restructured in the red, but the more subtle actions tend to be things like: * Comp bands for are now targeting p50 averages rather than p75 or top of market. So you can't close new hires that are going competitors. And you can't give raises to your top performers…

Cash is still King. The more cash on hand, the easier it is for the business to survive in an economic downturn, and the more dividends and stock buybacks can happen for the investors. Software engineers from a CFO perspective aren't any really different than plumbers or carpenters. It's just labor. Getting a cheaper rate on labor is far more beneficial to the company than say making sure it's employees are happy. If…

Software engineers from a CFO perspective aren't any really different than plumbers or carpenters

Plumbing systems are extensively documented and well-understood. The equivalent situation in plumbing would be that the entire layout of the city's sewage system (or similarly, the construction plans for the company's headquarters) is only kept inside the heads of two senior developers, likely because the Powers That Be have decided that keeping accurate documentation is not worth the cost. Preventative maintenance and quality control are impossible, because nobody knows which pipes have been laid first and are the most at-risk from leaking.

Point being: everything is just labour. That doesn't mean software "engineering" should be allowed to fall so far below the baseline we expect from other engineering professions.

Re: The McNamara fallacy: Measurement is not understanding

#198

Post by tremon »

dwaltrip wrote: post #137
milesvp wrote: post #85

Earlier quoted context omitted.

I worked at a company that started to go through the "you can't improve what you don't measure" phase. In general it was good for the org I was in, but I used to have to remind management that there's a corollary to that saying, which is: you necessarily improve things you measure at the expense of the things which are difficult or impossible to measure. This seems to be a hard one for some types to truly grok. A com…

It's fascinating to me how many smart people fall into this trap. Can they quantify their love for their partner? No? Well, I guess they have to give up any efforts to improve the relationship.

"If you spend 10% less on birthday presents for your children, will that make them love you 10% less? How can you justify spending this much on presents?"

Re: The McNamara fallacy: Measurement is not understanding

#199

Post by _8j50 »

aiilns wrote: post #194
_8j50 wrote: post #192

The US did not lose the war against the VC, they stopped supporting south vietnam's war against the VC and withdrew because the war became unpopular. The distinction is important because ultimately it wasn't America's war and the VC did not overwhelm the US military into surrender. It was the civilian government of the US that directed an end of their involvement. If anything, China won a cold war against the US who…

> The US thought it could play nation-builder like with Japan. The continued involvement in afghanistan was part of the ongoing "war against terror" to reduce the chance of afghanistan becoming a heaven for terrorists again. I would more easily characterize US external politics as nation-destroyer than nation-builder. Even if you disregard the evidence, it's quite a stretch to believe that the situation in Afghanista…

> Even if this has happened with some frequency I'm not so sure it's the rule; certainly there are many exceptions. In any case, it's no argument for doing the same.

The only time where there is nation building is after permanent colonization or assimilation into an empire. The US did pretty much burn the country to the ground, my argument is they should have left right after. Same with Iraq. If theu harbor terrorists again, give them round two but worse. Let the people self-determine and fight for whatever government will prevail. They shouldn't have worried about what happenes after they left so long as they made the price of harboring terrorists clear.

Re: The McNamara fallacy: Measurement is not understanding

#200

Post by projektfu »

erosenbe0 wrote: post #190
mahkeiro wrote: post #182

Earlier quoted context omitted.

Just look at Intel to see a company that was managed by the accounting people. Great numbers for a few years, large dividend, etc, but now they have to spend billions and billions to put the company back on track and give it a future.

Disagree. Intel has had continuous dividends for at least 30 years. Few giant American companies have been as consistently successful as Intel. I can think of only Microsoft, one or two energy conglomerates like ExxonMobil, a telecom or two like Verizon, and maybe a bank like Goldman Sachs.

Abbott Labs, CR Bard, Deere & Co.

Something about staying in your lane while increasing opportunities in that lane seems to be a good long-term formula.

Post reply on HN