202. The McNamara Fallacy 

I'm sure you're already familiar with it, but McNamara's mistake is also related to Goodhart's law, which Marylin Strathern generalized as "When a measure becomes a target, it ceases to be a good measure."

I always thought it was Mac-na-mah-ra (not mc-nah-mara).

Just wanted to say that I've always loved these short video essays. Every now and then I find myself back on your channel and bingeing your content. The explanations are easy to understand and keep me extremely invested in things. Keep up the good work and I hope your following keeps growing!

I was really early and learned of a fallacy I had no knowledge of before. I'm not sure why I'm smiling...

Absolutely terrific breakdown! As an Army Psyop vet and former member of the Human Terrain System (HTS) program in Iraq and Afghanistan, respectively, I can truly relate to this fallacy being an issue at higher levels (albeit we've evolved a GREAT DEAL since Vietnam, and despite flaws, COIN fights were run with far more fluid understanding of the asymmetric battlefield and the need for some qualitative, often imperfect 'human-centric' data alongside quantitative). Afghanistan especially mirrored some of the McNamara Fallacy in terms of how we ignored the tribal system and need for bottom-up approaches to winning alongside the 'human terrain' (See One Tribe at a Time by Green Beret & MAJ Jim Gant, for ex, and MANY other papers that were sidelined by the policy makers and higher ups). We relied on superficial metrics from Kabul and the paper weight Afghan Army, blind sighted by the realities on the ground. Inside the Box thinking at its finest. This is a big reason the Taliban won.

The place where I have seen this quantitative fallacy at work, is at work. I work in aviation maintenance and managers are continuously ignoring that quality is not really a measurable quantity so simply don't know what to do about it. They don't know if we're doing well or not. Things that can't be measured get ignored, I think that is almost a universal in all corporations who really follow McNamara's philosophy which itself is a descendent of Taylorism or "scientific management". I suspect that science is likewise very limited by this effect. What kind of scientific study could make statement about the "feelings of the Vietnamese people" other than by trying to quantify it. And I'm sure they would end up with a number, even if that number was baloney. They would take a fake number before they would accept that something is unmeasurable. I think this is a direct consequence of supposed quote by Lord Kelvin that, before you can improve something, you must first measure it.

Oh gahd, please, you're pronunciation of McNamara is driving me up the wall, it's pronounced "MACnaMAra". here is a news reel from the time: ru-vid.com/video/%D0%B2%D0%B8%D0%B4%D0%B5%D0%BE-8hwRi_ciJQA.html

As the DnD nerd I am, I have mostly noticed this in people that don't like developing characters with "bad" attributes. As If having a bad score somewhere cannot possibly add to the enjoyment that is playing that character.

As John Michael Greer said: Knowing many stories (metrics) is wisdom. Knowing no stories (metrics) is ignorance. Knowing one story (metric) is death.

Can't always count what counts :/

MACK nuh MARA. mack nuh mara (mara rhymes with mare). not mack- NAMURUH. there's an academy award winning movie about him called, "The Fog of War." at least learn how to pronounce his name..

I'm reminded of one of the rules I've learned, always ask the local guy, they know more than the numbers indicate.

Very Thunk’able subject matter! I imagined a slew of conversations with friends, ¡thank you!

The key is to recognize that numbers, or quantitative data, can be objectively verified by somebody else. If your fact is "the sunset is beautiful," a half-dozen observers will give you a half-dozen different responses depending on what their concept of "beauty" is. The primary issue confusing McNamara was that he did not know how to make observations of qualia independently verifiable, and (unfortunately) that is still an issue leading to the conflation of "fact" with "truth."

You're scratching the surface, but McNamara's analysis mistakes were so sophomoric it blinds most historians and analysts even today. Collecting good data is, of course, a great place to start when trying to achieve an outcome, but even granting McNamara's incredibly simple model, what was he missing that was so, so obvious? (hint: enter any reasonable values for McNamara's stupid KD ratio model, check the results, notice anything? what does that say about McNamara's analysis skills....) (also, fyi, you're pronouncing his name wrong)

"1,2,3.." example demonstrates only that one can't make definite conclusions from 3 numbers.

It should be noted that McNamara was not simply an otherwise-reasonable person who optimized for what he could measure and ignored what he couldn't measure. He was also responsible for funneling as many as people into the war effort as possible in order to meet his self-imposed quotas. Those interested in who he really was should look into Project 100,000. I feel that culture comes from the top so it's odd to think that he had no grasp on the implication of "optimizing for dead enemy combatants" when he himself skirted the rules when he was tasked with optimizing something.

LEEEEEROOOOOOOOOOOOYYYYYYYYYYYYYYY.....jennnnnkins.

You’ve probably had this question a lot over the years, but do you find any free time between your work and this YT channel?

GDP as a metric for prosperity seems abit archaic. I guess welfare and happiness can can be only be quantified to a degree but it would make alot of sense for governments to employ these criteria to understand and evaluate policy and its impact

Numbers in war don'd matter you fail to break the will of the enemy.

This is very informative insight into why the US lost the Vietnamese war. I mean, the lack of an anthropological approach is mind boggling, as Sun Tzu and Aeneas Tatcitus discussed the need of support from the civilian population to win a war. I can imagine the Global War on Terror in places like Afghanistan and Iraq enacting the McNamara fallacy when it comes to underestimating the effectiveness of suicide bombings giving increasing support of guerrilla factions.

This fallacy might also lead you to a related type of fallacious thinking: The streetlight effect. This is a type of observational bias that occurs when people only search for something where it is easiest to look. Since numbers are easiest to quantify, you’ll stick with numbers even though other (harder to reach) information would help you more. The name refers to an old joke: A policeman sees a drunk man searching for something under a streetlight and asks what the drunk has lost. He says he lost his keys and they both look under the streetlight together. After a few minutes the policeman asks if he is sure he lost them here, and the drunk replies, no, and that he lost them in the park. The policeman asks why he is searching here, and the drunk replies, "this is where the light is" en.m.wikipedia.org/wiki/Streetlight_effect

An over educated geek; He, as are you.

It took me quite a while to appreciate the number of available layers in the opening pun.

You can make an argument for anything with statistics

Thanks for your short and concise videos, so helpful

sorry, if you can not pronounce his name correctly, bye

See I always have issues trying to make decisions with qualitative data myself, now maybe that's a matter of scope since it's much easier to choose a new apartment based on quantitative metrics (Bedrooms, time to work, maximum speed of internet connection, etc) than qualitative (Do I *really* like this balcony more than the other?) and that would be different if I were making decisions of consequence like McNamara was where other peoples opinions are of heavy import but still...know I'm guilty of this fallacy in my thinking.

Mac-Na-Ma-Ra. kids today...

I find this hard to balance with anecdotal evidence and absolute subjectivism. Among the most widespread and misunderstood sayings is "correlation does not mean causation", however a significant enough correlation always means prediction, no matter whether one causes the other, both are caused by the same thing, or they just share temporal or geographic proximity. But I've seen it used too many times to discredit highly rigorous, meta-analytical findings by anecdotal evidence or just flipping the table and deciding truth is either subjectivist or no-one can really know what's going on. I'm afraid if the McNamara fallacy would be wide-spread it would lead to similar misinterpretations. So a really key question is, how do you balance avoiding the McNamara fallacy without falling too deeply into uncertainty and subjectivism? After all, certain findings hold more ground than others, and "sticking to the objective numbers" will give you the most informational bang per investigative buck.

Lol love that Leeroy Jenkins reference in the end

Better go get some mcdonALDs

A big problem with the obsession with numbers is that as soon as human valuation comes in calculation is impossible. By what rational methods can one determine the value of oranges vs staples?

Models are of human creation and thus come with their own limitations. The fact that he believed that his models were the final say in everything is just sheer arrogance.

I see where quantification helps. Like if I'm a runner, and I want to get faster, I can try different training approaches, and compare performance and see what works. But there are many variables, and some can help and hurt, like losing body fat can help run faster, but it can also cause a "distorted body image" where you push your fat down below a healthy level, and other body systems start breaking down, until performance gets worse. So I ignore a less quantifiable measure like "How do I feel today?" with a single number "How much do I weigh?" or "How low is my body fat today?" If you look at the compromises professional athletes make to reach peak performance, they may successfully optimize the variables, so pushing body fat lower, but not too low, when there are clear health problems. And maybe there are measures of "health" that can give warnings, like low testosterone perhaps. Then at least you can act with "eyes wide open", putting some level of health at stake for a goal but acknowledging it may not be healthy for long periods, so you can cut back when a big competition is over.

I think the best approach is to define the targets in very layman human sentences and then see which metrics should be measured to quantify them, not the numerics themselves be the target.

Hocus pocus, I'm a dip-LOD-icus.

How To Lie With Statistics - great book!

0:15 Hey. Just found your channel and thought I should inform you … his name is pronounced MAC-nah-mehr-ahh. The stress goes on the first syllable.

A great explainer video, just a note it's an Irish name pronounced like "Mac Na mar A" I am from Ireland and every time you say it, I am like "What the heck" lol