Cool new working paper on why and how cable news threw gasoline on the culture war fire Link in reply

中部民間團體今舉行記者會，以「廚餘無過，過在人禍」為題，強調與政治藍綠無關痛。痛批台中市政府防疫失控、行政怠惰，導致垃圾與防疫雙重危機。民團共同要求市長盧秀燕、農業局長、動保處長與環保局長立即下台，為防疫失能負起政治責任。 台中市長盧秀燕今天在市政會議說，市府用最高的標準面對非洲豬瘟，因為這個豬瘟病毒非常頑強，在世界各地都造成畜牧災難。因此面對外界的質疑以及需要精進的地方，市府團隊會即時檢討改進與調整。 主婦聯盟環境保護基金會台中分會會長耿明誼指出，非洲豬瘟並非天災，而是市府長期行政怠惰與管理無能造成的人禍。疫情爆發，不僅引發防疫破口，也讓垃圾與廚餘問題全面浮現，前端稽查失靈、後端掩埋超載，導致整個城市環境治理系統崩潰。 她批評市府防疫決策混亂，疫調資訊不實、擅自清消、未與中央協調，「內控極差、行政失能」。 呼籲立即啟動隨袋徵收、推動廚餘資源化，從制度面改革廢棄物管理。 好民文化行動協會理事長林芳如痛批，市府防疫資訊錯誤、隱匿真相，更動搖民眾對政府的信任。她呼籲檢調應依刑法第130條「公務員廢弛職務釀成災害罪」追究高層責任，並要求盧市長公開道歉。 彰化縣環境保護聯盟研究員林政翰指出，台中每日約有254公噸熟廚餘原用於餵豬，禁令後全數改掩埋或焚化，他警告，廚餘含水量高、熱值低，焚化不完全易產生戴奧辛等汙染物，侵蝕爐體並加重空污。建議應擴大推動黑水虻處理、堆肥及社區簡易回收設施，取代「大撒幣」式的廚餘機補助。他批評議員提案編列百億預算送廚餘機給市民「荒誕至極」 「台中環保局的決策像父子騎驢，被罵就轉彎！」爭好氣聯盟發起人許心欣批評，掩埋覆土不實、焚化後又引發空汙疑慮，顯示環保局缺乏專業能力。她要求市府公開焚化爐戴奧辛檢測數據，否則只是「唬弄市民」。 台中市食農教育發展協會理事長呂木蘭說，廚餘無罪，應視為再生資源，可轉化為肥料、飼料或發電能源，若市府用心規劃，就能兼顧循環經濟與環境永續。「市長不救豬瘟卻忙著購物節，這樣的施政令人蒙羞。」 台灣中社代理社長李永明強調，防疫應回歸科學，台中市府延誤通報、違反程序，導致全國防線失守，應公開疫調流程與資料，向市民與業界道歉。中台灣教授協會也指出，防疫與藍綠無關，但台中市府多次違背防疫原則，必須嚴厲譴責。 后里青年郭瀚陽則直言，縣市合併後資源傾斜嚴重，非洲豬瘟爆發後，后里成了廚餘掩埋與焚化的代罪羔羊。「我們要的是能負責任的政府，而不是掩耳盜鈴的市政團隊。」 一名返鄉青年肉品承銷人表示，疫情導致豬農與肉攤全面停擺，但補助僅限合法登記業者，許多年長小農被排除在外，陷入無收入窘境。他呼籲中央與地方檢討補助機制，納入基層農戶並建立「廚餘再生回收鏈」，以高溫滅菌與分流追蹤確保防疫安全。 台灣護樹協會理事長張美惠指出，台中市政府像一輛零件的車子，失控又失能，不知道要把台中帶到哪裡去？她對盧市長失望，認為她是一位好媽媽，容忍團隊部署的錯誤，但卻不是一位好市長，令他失望。 各團體強調，從防疫到環境治理，問題不在廚餘，而在人禍。他們呼籲市府誠實檢討、公開資訊、承擔責任，並以制度性改革防止悲劇重演。 - 本篇裡面的團體，過去有跟市府同聲的，也有一直以來是站在民間監督的腳色，顯示的是跟市府友好的部分團體也跳出來譴責市府。 打個比方，其中護樹團體代表就是護樹老娘本人，連她都出來講話了ㄎㄎ udn.com/news/story/7325/9116…

Wow. From @nberpubs -

For lecture 2 I will teach you all about hyperbolicity and how it is critical for analyzing discrete-time dynamical systems: piped.video/watch?v=hzrM_4xS…

Excited that our paper on markup estimation is out in Econometrica @ecmaEditors. Much enjoyed working on this with @Basile_G and @MorzentiG. Big thanks to everyone for feedback over the years+our (anonymous) referees+editor @ChadJonesEcon who all really helped shaping the paper

In my new book, The Science of Second Chances, I talk about the strong research evidence that access to SNAP benefits reduces recidivism for adults coming out of prison, because it helps them make ends meet without resorting to crime. I also talk about the even stronger evidence that access to SNAP benefits for young kids reduces their future criminal justice involvement, because good nutrition is important for brain development. (One hypothesis for why nutrition is so important is that it helps protect the brain against exposure to common toxins like lead. Because lead and calcium both bind to the same receptors in the brain, consuming more calcium in early childhood can crowd out this harmful toxin and both protect from and mitigate the detrimental effects of lead exposure. That exposure would otherwise result in more impulsive and aggressive behavior in the teen and adult years.) Many are talking about the current interruption in SNAP benefits as a moral issue. But it is also a public safety issue. If we do not get families the food they need, we will pay for it for years to come, in the form of higher crime rates. My team at @Arnold_Ventures has been working to repeal state and federal bans on SNAP benefits for people with criminal records, which reduce food access even in "normal" times. Such bans are counterproductive. For more on this, please consider pre-ordering my book! (It comes out in February.) If you're a policymaker who would like more info on this evidence, please get in touch.

🔥 New Series! Learning GPU programming through Mojo puzzles - on an Apple M4! No expensive data center GPUs needed. No CUDA C++ complexity. Just Python-like syntax with systems performance. First video just dropped: piped.video/watch?v=-VsP4kT6… #Mojo #GPUProgramming #AppleSilicon

鏡週刊這三篇人物報導很有意思，剛好展現了三種不同的 AI 使用方式，從極度警戒到職涯教練都有。 第一個是啟明出版社發行人兼總編輯林聖修，他本身有工程師背景。第一次使用 ChatGPT 是為了尋求合約爭議的建議，但後來意識到機密資料可能會被拿去做其他用途，於是當天就把帳號刪掉了。他認為，AI 應該幫助人類處理瑣碎、重複、無聊的工作，讓人類能「好好做事」。 第二個是政大 AI 中心的助理教授吳怡潔。她利用大型語言模型分析鄭南榕《自由時代》週刊共三百多期雜誌，節省了大量時間。不過她也在課堂上提醒學生，不要過度依賴 AI，以至於一旦離開它就無法寫程式。這究竟是使用 AI，還是被 AI 附身？她主張跨領域學習，不分文理組都要會用 AI，但更要知道哪裡該踩煞車。 第三個是小說家陳又津。她會要求 ChatGPT 觀察自己的工作日誌，幫助她調整出適合身體狀況的工作節奏。她建立了自己的 AI 團隊，讓 AI 協助處理瑣事、寫作與撰寫學術論文，但批改作業仍由她親自完成。由於成長背景，她兒時長期不敢、也不太會問問題；不過面對 AI，她第一次感覺「可以問到飽」。陳又津認為，AI 其實是一面鏡子，會依據使用者輸入與訓練內容展現能力。如果使用者沒意識到自身的盲點，就容易陷入鬼打牆的狀態。 看完這三篇報導，我開始思考自己到底是哪一派。 可能是一邊警戒，一邊依賴，也一邊試著找到與 AI 共處的節奏吧。

Making Interpretable Discoveries from Unstructured Data: A High-Dimensional Multiple Hypothesis Testing Approach. arxiv.org/abs/2511.01680

What were the causes of South Korea's export miracle? My new paper w @philippbarteska, @straightedge, @seung_econ explores a major overlooked factor: U.S. military procurement during the Vietnam War. We bring new evidence to the Q of how geopolitics shapes development. 🧵:

Today I taught a PhD class on the economics of AI. In doing so, I drew this picture on the board of my current understanding of what I called the "how good is AI" literature (aka the productivity impacts of AI). I thought I'd write up a long version of that discussion - at the very least for my own notes. The basic point I made is that answering this question requires an understanding of the crazy multi-dimensionality of the problem that dramatically shapes the questions you ask and the answer you get. A few key dimensions: a) Tasks -- this one is obvious. How useful AI is depends on what specific task is being considered. Every day, we get a paper along the lines of "I had AI help with task X" (call center work, consulting, graphic design, writing ... name it) -- and here is what I found. b) Human + AI OR Human vs. AI AKA is AI augmenting or replacing humans? -- The value of AI in helping a human achieve something is not the same as the AI doing the task itself. Yet, we often conflate the two. Economists and social scientists often focus on the augmentation test, while CS people are mostly comparing humans with AI. For example see this nice paper from Serina Chang showing how the two are not the same (aclanthology.org/2025.acl-lo…) c) Point vs Systems integration -- Early papers have mostly been about isolated tasks divorced from the specific job in artificial settings. Even "real" field experiments have focused on work like call center work that is atomistic. However, one could imagine answers we get are very different when tasks are embedded with jobs and jobs are embedded within organizations. This point is what is driving the divergence between "AI increased productivity by a gazillion percent" in studies and "95% of all AI implementations fail" narrative from real enterprises. Surely the value of AI depends a lot on this unit of analysis. d) Which AI and How AI: CS researchers are much better on this than economists, but the basic idea is that there is no "one" AI system and results clearly depend on which AI we are talking about. And here, its not just about which model you look at (GPT 3.5 vs 4o say), but also how these models are prompted and going forward the extent to which they are daisy chained in agentic systems, or indeed fine tuned or modified in some deep way for specific tasks. Just because an AI system sucks at something out of the box, doesn't mean that with some work it can't be made to improve. Or not. We'll never know unless we stop treating AI like a monolith. e) Which Human? -- This is one point on which we have made good progress but still a lot remains to do. Clearly the value of AI relative to a human or augmenting a human depends a lot on who the human is. Past work has found a flattening of the expertise curve in routine tasks -- but a mental model where AI's productivity effects are removed from which humans we are comparing them to would be the wrong mental model. Corollary, that human's incentives might matter as much as their capabilities! Now imagine asking the value of AI question and blowing it up in terms of each of these dimensions: Tasks X Augment/Automate X Point/System X <every AI model out there> X <every human out there> it becomes pretty clear that we are only getting started in terms of understanding the societal/productivity implications of this thing. And thats even before the put the LLMs inside robots and let them loose on the world. Grad students rejoice! <HT: I feel like I channeled my inner @random_walker here - so this is inspired by his writings on AI snake oil and related topics!>

你有聽過白話字 Pe̍h-ōe-jī 嗎？

U Toronto's "Mathematical Expression & Reasoning for CS" PDF: cs.toronto.edu/~shaharry/csc…

"it's Python, you do anything and it allocates" How true is this? I modified CPython to print when it allocates an integer object Then added numbers in a for-loop 100k times My terminal got spammed with 101006 allocations Why? Let's explore the internals of CPython:

If you have been curious about which topics are increasing in popularity in econ, @paulgp has a new tool for you! paulgp.com/econlit-pipeline/… This lets you search over AER and AEJ's + 30k NBER WP! Very neat!

Is this the end of handwritten math? Introducing Lean. ~~~ Ank Yog. piped.video/0QZI_m8WZ0Q #ITP #LeanProver #Math

we are working on a Rust-based ETL server that can stream your Postgres database to S3/Iceberg (and other databases like BigQuery & ClickHouse) 100% open source, and designed so that it can be embedded in any Rust server

Perhaps the future of hiring is not AI, but a dating app... In ~10 days my fantastic co-author @JSockin presents our paper “Interviews” (with @ellliottt) at @nberpubs. We show candidates walk away when an interview feels “too easy”—they assume the firm would hire anyone.

Robust fuzzy clustering for high-dimensional multivariate time series with outlier detection. arxiv.org/abs/2510.26982

I'm excited to announce that my book of selected writings, Coexistence and Other Fighting Words, will be out on December 10, 2025! Preorders are available on Amazon or my publisher's website, Political Animal Press. For a taste of what's coming, see bayes.cs.ucla.edu/DPF/Table-…, but please start with "Why I Wrote This Book" ucla.in/3A9BirT.