The Metric Is Not The Mission: When They Still Understood Us
#goodhartslaw — Public Fediverse posts
Live and recent posts from across the Fediverse tagged #goodhartslaw, aggregated by home.social.
-
The Metric Is Not The Mission: When They Still Understood Us
-
RE: https://fediscience.org/@petersuber/116880277331727208
#GoodhartsLaw #bibliometric It was written: if a measure becomes a target, it ceases to be a good indicator. And, of course, industrial scale systems like paper mills are far more efficient in reaching it than do-it-yourself researchers' cliques.
-
"The Soviet Union had a shortage of nails during Lenin’s time. To increase production, his government started giving bonuses to the factories for the number of nails they produce. After hearing about the bonus, the factories reduced the size of the nails to produce as many nails as possible. In the end, they met the targets and got their bonuses. But the government ended up with millions of useless tiny nails." https://fronterabrands.com/goodharts-law/ #GoodhartsLaw https://freakonometrics.hypotheses.org/63816
-
Amazon built an internal leaderboard to track employee AI usage.
Employees responded exactly as rational humans do when subjected to a badly designed metric.
They gamed it.
Goodhart’s Law remains undefeated.
Technology changes. Dashboards get prettier. Babudom survives.
#AI #GoodhartsLaw #Management #Bureaucracy #Amazon
https://calamur.org/gargi/2026/06/01/when-ai-became-a-corporate-kpi/
-
Workers use AI to prove they use AI. Something something 'When measures become targets they stop being good measures.' https://arstechnica.com/ai/2026/05/amazon-employees-are-tokenmaxxing-due-to-pressure-to-use-ai-tools/ #AI #work #scifi #goodhartslaw
-
My government's favourite trick for gaming inflation numbers (also practiced by many other nations) is omitting "highly volatile" elements from the calculations. Specifically, food and energy.
Because the average person spends basically none of their money on those two items, right? And the prices of those shoot up crazily, but their prices also drop like a stone the next month, making the consumer come out even, right?
Goodhart's Law, etc, etc.
-
Pluralistic: Daily links from Cory Doctorow – No trackers, no ads. Black type, white background. Privacy policy: we don't collect or retain any data at all ever period. [Unofficial] @[email protected]Pluralistic: Goodhart's Law vs "prediction markets" (24 Mar 2026)
https://fed.brid.gy/r/https://pluralistic.net/2026/03/24/degenerated-gambling/
-
Goodhart’s Law in action 🙄:
“More! More! More! Tech Workers Max Out Their A.I. Use.”, The New York Times (https://www.nytimes.com/2026/03/20/technology/tokenmaxxing-ai-agents.html?unlocked_article_code=1.UlA.Wda2.-3Rz1wP8LBVw&smid=url-share).
H/T: Dare Obasanjo
#AI #AITokenUsage #GoodhartsLaw #Metric #TokenMaxing #LLMs #Measure #FakeProductivity #Waste #Dashboard #Leaderboard #Work
-
"Well, sure, all our sandwiches now come with shit. Shit is a very popular topping with our customers, so we thought we'd make it easy for all of you to get what you want. We determined it was too confusing for customers if some sandwiches came with shit, and some didn't, so we simplified our menu to make sure you could find exactly what you need. No, you can't order a sandwich 'but hold the shit'. After all, if you're that dead-set against eating shit, you can just order your sandwich and then eat around the shit."
This is a post about #Microsoft and #Google.
#Microslop #ShitSandwich #shit #EatShit #CoPilot #Gemini #AI #LLM #NobodyWants #bubble #GoodhartsLaw #Goodhart #metric #software #users
-
Pluralistic: Daily links from Cory Doctorow – No trackers, no ads. Black type, white background. Privacy policy: we don't collect or retain any data at all ever period. [Unofficial] @[email protected]Pluralistic: Big Tech joins the race to build the world's heaviest airplane (09 Dec 2025)
https://web.brid.gy/r/https://pluralistic.net/2025/12/09/temporarily-embarrassed-founders/
-
Pluralistic: Daily links from Cory Doctorow – No trackers, no ads. Black type, white background. Privacy policy: we don't collect or retain any data at all ever period. [Unofficial] @[email protected]Pluralistic: Big Tech joins the race to build the world's heaviest airplane (09 Dec 2025)
https://fed.brid.gy/r/https://pluralistic.net/2025/12/09/temporarily-embarrassed-founders/
-
Richard Charkin on Goodhart’s Law: The Measurement Trap
In his newest column, Richard Charkin applies the principle that when a metric becomes a target, it ceases to be a good measure to today's publishing industry.
The post Richard Charkin on Goodhart’s Law: The Measurement Trap appeared first on Publishing Perspectives.
https://publishingperspectives.com/2025/12/richard-charkin-on-goodharts-law-the-measurement-trap/ -
"OKRs graphs don't make your teams more focused. Repeated conversations about the goals do."
-
-
From: blenderdumbass . org
When a measure becomes a target, it ceases to be a good measure.
Read or listen: https://blenderdumbass.org/articles/the_goodhart_s_law_of_child_abuse
#goodhartslaw #childabuse #politics #paternalism #ageism #education
-
"AI – the ultimate bullshit machine – can produce a better 5PE than any student can, because the point of the 5PE isn't to be intellectually curious or rigorous, it's to produce a standardized output that can be analyzed using a standardized rubric.
I've been writing YA novels and doing school visits for long enough to cement my understanding that kids are actually pretty darned clever. They don't graduate from high school thinking that their mastery of the 5PE is in any way good or useful, or that they're learning about literature by making five marginal observations per page when they read a book.
Given all this, why wouldn't you ask an AI to do your homework? That homework is already the revenge of Goodhart's Law, a target that has ruined its metric. Your homework performance says nothing useful about your mastery of the subject, so why not let the AI write it. Hell, if you're a smart, motivated kid, then letting the AI write your bullshit 5PEs might give you time to write something good.
Teachers aren't to blame here. They have to teach to the test, or they will fail their students (literally, because they will have to assign a failing grade to them, and figuratively, because a student who gets a failing grade will face all kinds of punishments). Teachers' unions – who consistently fight against standardization and in favor of their members discretion to practice their educational skills based on kids' individual needs – are the best hope we have:"
https://pluralistic.net/2025/08/11/five-paragraph-essay/#targets-r-us
#AI #GenerativeAI #LLMs #Chatbots #Schools #Education #Grading #GoodhartsLaw
-
The problem of AI tech companies gaming their Chatbot Arena scores
https://youtu.be/9W_OhS38rIE
#generativeAI #AIslop #AIhype #ArtificialIntelligence #GoodhartsLaw -
"Don't, whatever else you do, blame your decisions on the data. If you do you'll never get accurate data ever again."
-
"En 1920, voyant les cas de peste se multiplier, la Ville de Paris a annoncé qu’elle verserait une prime à toute personne qui lui présenterait un cadavre de rat. Entre septembre 1920 et juillet 1921, 500 935 rats ont été tués. Six mois plus tard, les autorités changeaient leur fusil d’épaule en voyant que des gens avaient commencé à élever des rats pour revendre leur cadavre."
Un exemple de la loi de #Goodhart ?
https://www.courrierinternational.com/long-format/dans-nos-archives-a-paris-le-rat-le-bol-des-habitants_210771_1 (derrière un #paywall )
-
Company year #OKRs
Decrease time to market in software development by 20%.
I am just thinking if I ever heard better textbook example of #GoodhartsLaw and #metricFixation this is not going to happen well, can’t see that huge pile of technical debt at the end of year. -
@annie @baldur Very reminiscient of Goodhart's law 👍
See also this excellent blog post by Jascha:
Too much efficiency makes everything worse: overfitting and the strong version of Goodhart's law:
https://sohl-dickstein.github.io/2022/11/06/strong-Goodhart.html
This is transferable to so many domains and #society at large!
-
Churnalism, enshittification, and a bunch of other online deterioration phenomena (plus a mention of the Potemkin news site that spread misinfo during the Southport riot).
https://www.prospectmagazine.co.uk/ideas/technology/internet/67864/dead-internet-theory-ai#internet #enshittification #churnalism #media #socialMedia #advertising #misinfo #incentives #GoodhartsLaw
-
Perfect case of #goodhartslaw
Colorado ranchers sabotaged rain gauges to falsify drought numbers, resulting in million $$$ insurance claims.
https://coloradosun.com/2024/09/08/patrich-esch-ed-dean-jagers-ranchers-fraud-insurance/
Goodhart's Law:
Every measure which becomes a target becomes a bad measure -
This obviously applies to any company and any incentive. See #GoodhartsLaw: https://sketchplanations.com/goodharts-law
-
@jburneko Kinda makes me think of #GoodhartsLaw: when metrics become targets, they stop being good metrics.
-
Thinking of something like “#GoodhartsLaw of Identity” which is like “when a measure becomes an identifier, it ceases to be a reliable identifier.”
It’s a sort of wave function collapse. Like maybe once we measure and document something, it ceases to be inherent to the thing and becomes a spoofable number, password, etc.
Is there already a term for this?
#DigitalIdentity -
XKCD: stop being so accurate about how strategists and senior managers think. It's creeping me out.
-
Können wir Entwickler:innen-Produktivität messen? Und wenn ja, sollten wir das auch? Was sind die Seiteneffekt, die eventuell übersehen wurden.
Das wird in der kritischen Auseinandersetzung mit einer McKinsey-Studie zum Thema sehr informativ vom @ewolff betrachtet.
Große Empfehlung Meinerseits.
#SoftwareArchitecture #metrics #development #coding #GoodhartsLaw #dora #space
-
“Such is the terrifying power of Goodhart's Law: you can create your own system and game it too.”
I had never heard of Goodhart’s Law, but I recognize this behavior!
h/t @campegg
https://www.experimental-history.com/p/how-to-drive-a-stake-through-your
-
@hotdogsladies Mazel Tov. Back in 2008 I discovered the nascent Fediverse, and - being what in those politically incorrect days was called a "freetard" - I thought "Brilliant! Now I can try doing that thing Merlin does on Twitter."
A long series of lame <140 char gags later, I ultimately decided I should stick to being funny unintentionally. But I'd found a lot of good people.
Relationships beat engagement metrics every time. #GoodhartsLaw
-
Some things to bear in mind when you see a news article about #SharePrices rising or falling.
Share price is, nominally, a function of how well rich people expect a company to do.
#GoodhartsLaw states that "When a measure becomes a target, it ceases to be a good measure". High share prices long ago became a target, and so have nothing to do with how well a company is doing; at most, they are now a function of how rich people expect them to change. #SelfFulfillingProphecy
-
CW: Long thread/26
#GoodhartsLaw holds that "When a measure becomes a target, it ceases to be a good measure." Give an MBA within HCA a metric ("get patients out of bed quicker") and they will find a way to hit that metric ("send patients off to die somewhere else, even if their doctors think they could recover"):
https://en.wikipedia.org/wiki/Goodhart%27s_law
Incentives matter! Any corporate measure immediately becomes a target.
26/
-
This may not be any kind of profound insight, but I just realized that part of the AI alignment problem is just Goodhart's Law.
https://en.wikipedia.org/wiki/Goodhart%27s_law says that any measure that becomes a target ceases to be a good measure.
It's an incentive mechanism: for example, you are running a doctor's clinic or hospital. You measure your revenue (because I'm presuming you're doing this in the US...) and then you make a target out of it.
Then the staff has an incentive to order unnecessary tests, or code a patient's condition as more serious, or so on. The incentive is to hit the target and not necessarily address the underlying measure -- or, the reality (health of your patients) underlying the measure.
Meanwhile, folks are worried about AI alignment, and some powerful, superintelligent AI system that can act in ways that harm humans.
So we want to measure how well-aligned an AI system is. So you...ask it. Maybe you take its responses and do further training. But the concern is, you could just train the model to *say* that it's aligned, while actually it's not.
(A similar thing happened with the image classifier that was supposed to distinguish dogs and wolves -- and it did so wonderfully! Until we realized that it wasn't distinguishing dogs and wolves, but more so whether the photo was taken indoors or outdoors: https://hackernoon.com/dogs-wolves-data-science-and-why-machines-must-learn-like-humans-do-41c43bc7f982 )
In other words, the analogy -- in SAT test-like phrasing:
Goodhart's Law:
measure :: target
and AI alignment:
AI is aligned with human interests :: AI declares itself to be aligned
