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

Goodhart's Law: When a measure becomes a target, it ceases to be a good measure

en.wikipedia.org

Post reply on HN
Print view View original

71–80 of 137 posts

Re: Goodhart's Law: When a measure becomes a target, it ceases to be a good measure

#71

Post by philipkglass »

sonnyblarney wrote: post #60
cwmma wrote: post #56

Earlier quoted context omitted.

There's a quote to the extent of "the US never lost a battle in the Vietnam, but still lost the war" which is probably what they were getting at, which just honestly proves the point that America was really optimizing for the wrong measure in that war.

I very well understand there were challenges in the war - and that 'body count' and 'battles won' was not going to be a good metric. This is a valid point: we though it was going great because hey - look at the 'battles won' and 'body' count, but really it wasn't. But the US did not remotely lose the war. There is no rational basis for suggesting this - it doesn't even make sense. This is the weird delusional pop cul…

National Security Action Memoranda number 52, May 1961:

https://www.jfklibrary.org/Asset-Viewer/Archives/JFKNSF-330-...

Subject: Program of Action to Prevent Communist Domination of South Viet Nam

The US Objective: to prevent Communist domination of South Vietnam and to create in that country a viable and increasingly democratic society.

The US had not achieved those objectives 12 years later. The communists won control over the whole country a few years after US withdrawal finished. Vietnam's government is still communist today.

This is basically what I learned browsing through encyclopedias when I was a kid. I did a quick fact-check just now in case I had been misled or misremembered. I can sort-of understand saying that the communists didn't achieve victory over the USA either (because they also didn't manage to install an allied government in the territory of the USA), but claiming that the US won is weird.

Re: Goodhart's Law: When a measure becomes a target, it ceases to be a good measure

#72

Post by abiox »

minimaxir wrote: post #16
nikanj wrote: post #4

"You get what you measure" is the bane of KPIs everywhere. Measure deals closed, and you get deals with no regard for profitability. Measure profitability, and you will burn through your customer base by jacking prices and cutting quality. Etc etc.

When I interviewed for a data analyst position at a known startup, I was asked to define a target KPI for a controlled pseudo-A/B experiment, and I suggested total revenue. The interviewer replied "that's a dumb answer because revenue can be gamed; you should have said # of sales." I was too taken aback to follow up asking why # of sales can't be gamed too.

> interviewer replied "that's a dumb answer

a bit ot, but was that the actual response?

personally, stuff like this would have me ending the interview early.

Re: Goodhart's Law: When a measure becomes a target, it ceases to be a good measure

#73

Post by scarmig »

triviatise wrote: post #65
logicallee wrote: post #26

Aren't there are exceptions? For example, a measure of health might be life expectancy. If this becomes known, is it no longer a good measure? I think it's still a good measure. I mean, how can you get to a life expectancy of 100 without a healthy population?

Life expectancy is actually a really great example of the opposite. People routinely claim our health system is bad compared to europe because our average life expectancy is lower. It turns out african americans skew our metrics because they have 1) high infant mortality 2) high murder rates of young men The high infant mortality is irrespective of wealth and has a disproportionate impact on average life expectancy b…

It's not like Europe doesn't have marginalized populations. Your argument is basically "if you exclude the parts of the US population that are worst treated by our healthcare system, then ours is the best!" There is something meaningful in comparing the "median" or "modal" health experience in different countries, but "we treat black people badly compared to white people in the USA so you have to only look at white statistics" is a terrible defense of the system.

Your point about infant mortality statistics being gamed is on point, though (and I pointed it out in a sibling commment).

Re: Goodhart's Law: When a measure becomes a target, it ceases to be a good measure

#74

Post by nkoren »

njarboe wrote: post #64
nkoren wrote: post #55

Earlier quoted context omitted.

Even disregarding the potential for negative externalities etc., profit alone can still be a terrible KPI. I worked for a large engineering consultancy that was very profit motivated. They figured out a brilliant way to maximise profit: get rid of business development! Only pay for work when clients are paying you to do it. This was GREAT for quarterly profit margins, at the expense of the project pipeline. As the pi…

I would say the complexity is in balancing profit over time. Here is your contradictory force. In the end that large engineering consultancy was making zero profit and was worth nothing. Whether running a business into the ground will give you more overall profit than just selling it and putting the money into government bonds is something one can consider, but in your case was the consultancy owned by the people mak…

> in your case was the consultancy owned by the people making these short-term profit maximizing decisions[?]

It was a publicly-listed company, so that'd be a yes.

Profit is obviously over time a necessary condition for making a sustainable business. But it's definitely not a sufficient condition. I think there are a huge number of people who don't understand this.

Eg.: when I see people clambering for TSLA to make a profit... I think they're wrong. Or, at least, mostly wrong. TSLA has been unprofitable in the service of dramatically (and rather effectively) expanding both capacity and market share. Those are excellent reasons to be unprofitable. And TSLA still obviously has a lot more room to grow in both respects, so in this situation, trading profitability for growth is a good decision. (exhibit A: Amazon)

Re: Goodhart's Law: When a measure becomes a target, it ceases to be a good measure

#75

Post by minimaxir »

abiox wrote: post #72
minimaxir wrote: post #16

Earlier quoted context omitted.

When I interviewed for a data analyst position at a known startup, I was asked to define a target KPI for a controlled pseudo-A/B experiment, and I suggested total revenue. The interviewer replied "that's a dumb answer because revenue can be gamed; you should have said # of sales." I was too taken aback to follow up asking why # of sales can't be gamed too.

> interviewer replied "that's a dumb answer a bit ot, but was that the actual response? personally, stuff like this would have me ending the interview early.

That was one of the tamer criticism of my answers during my interviews that day.

I wasn't extended a job offer, but would likely have not accepted it if I had.

Re: Goodhart's Law: When a measure becomes a target, it ceases to be a good measure

#76

Post by waltherg »

davidmanheim wrote: post #27

This is probably where I should mention my two longer articles that discuss this in depth; https://www.ribbonfarm.com/2016/06/09/goodharts-law-and-why-... https://www.ribbonfarm.com/2016/09/29/soft-bias-of-underspec... As well as my more recent paper categorizing the different ways it occurs; https://arxiv.org/abs/1803.04585

Since you’re here, what does the following property stand for in your second paper?

s ∈ AifM(s)/gec

Re: Goodhart's Law: When a measure becomes a target, it ceases to be a good measure

#77

Post by jingleheimer »

notEvenOnce wrote: post #23

(cough! cough! "scrum") (cough! cough! "agile") (cough! cough! "daily stand-ups")

I always felt that agile processes were a means to give non-answers to performance measurement in order to allay the consequences of that measurement and allow developers to get on with the business of making software.

How big is the task; 10 story points

How long will that take; depends on our velocity

Can it be done by this date; we're not allowed to plan that far ahead. You can put it in the 'icebox'. We might get to it eventually.

Re: Goodhart's Law: When a measure becomes a target, it ceases to be a good measure

#78

Post by ranger207 »

sonnyblarney wrote: post #57
Semirhage wrote: post #43

Earlier quoted context omitted.

And FYI the US handily and clearly won the Vietnam war; the South Vietnamese insurgency was defeated and the North Korean Army was obliterated. That’s a serious, if delusional claim that requires a loooooot of support. “Handily” in particular seems like a ridiculous way to state it, even if you only look at US casualties.

No, it's not remotely delusional. What's delusional and troubling is the degree to which this war is completely misunderstood in pop culture. The US handily won the Vietnam war, full stop. In classical terms is was a clear and unambiguous outright victory. Think of Korea for example: a stalemate, whereas the US obliterated their oppponents in Vietnam. The US hardly lost a single battle and losses in the war are consi…

>The US handily won the Vietnam war, full stop. >In classical terms is was a clear and unambiguous outright victory.

The only measure of success is whether or not you accomplish your objectives. The US did not accomplish its objectives in Vietnam of keeping a US-friendly government in power. Regardless of tactical measures of success, the US failed in its strategic objectives. I agree that popular culture has not accurately portrayed the Vietnam conflict, but saying that the US won in Vietnam is completely untrue.

Re: Goodhart's Law: When a measure becomes a target, it ceases to be a good measure

#79

Post by dredmorbius »

abiox wrote: post #72
minimaxir wrote: post #16

Earlier quoted context omitted.

When I interviewed for a data analyst position at a known startup, I was asked to define a target KPI for a controlled pseudo-A/B experiment, and I suggested total revenue. The interviewer replied "that's a dumb answer because revenue can be gamed; you should have said # of sales." I was too taken aback to follow up asking why # of sales can't be gamed too.

> interviewer replied "that's a dumb answer a bit ot, but was that the actual response? personally, stuff like this would have me ending the interview early.

And what is your gentle phrasing for that action?

Re: Goodhart's Law: When a measure becomes a target, it ceases to be a good measure

#80

Post by stcredzero »

sonnyblarney wrote: post #51
stcredzero wrote: post #28

Earlier quoted context omitted.

That seems to happen in all wars. Aircraft downed and tanks destroyed in WWII, for example. There were occasions in the Vietnam War where regions were declared to be free fire zones, and everyone within was to be killed, down to the last child and granny.

I'm sorry but this is a serious (and arguably offensive) falsehood. A 'free fire' zone is an area wherein there is no requirement for soldiers to coordinate heavy arms fire with other units via HQ etc.. This is a very reasonable policy in jungle fights where units can come under ambush at any time, i.e. "We're in an ambush and we're about to die, can you ring up HQ to see if we can shoot back?" But the notion that US…

I'm sorry but this is a serious (and arguably offensive) falsehood.

I'm sorry, but I will show you to be mistaken below.

But the notion that US forces would just designate an area where everyone including civilians are to be killed is an outright and offensive lie.

Then you should have an issue with the Ken Burns documentary. I think I got the terminology wrong; "free fire zone" means something else. However, the practice I am referring -- that of designating everyone in a certain place as an enemy, then killing everyone within -- did happen.

https://en.wikipedia.org/wiki/My_Lai_Massacre

From Ken Burn's "The Vietnam War" Part 5 narration: (00:44:43,393 --> 00:44:47,428)

In the summer of 1967, Tiger Force was sent to the fertile Song Ve Valley. The entire population had already been herded from their homes and crowded into a refugee camp. But some had come back to resume the farming they had always done. The valley had officially been declared a free-fire zone, and Tiger Force's officers took that literally. "There are no friendlies," one lieutenant told his men. "Shoot anything that moves." Over a seven-month period, they killed scores of unarmed civilians. Among their victims were two blind brothers; an elderly Buddhist monk; women, children, and old people hiding in underground shelters; and three farmers trying to plant rice. All were reported as "enemy... killed in action."

Post reply on HN