ct (@colomct) profile picture
Hong Kong
Joined June 2010
Photo by ct (@colomct)Photo by ct (@colomct)
An excellent analysis of the recent political developments inside China, by @ananthkrishnan
4
82
2
448
23,787
This week's economic history lecture is now online (the deck of slides). Its Health + Development in economic history. Link below
3
21
127
6,743
ct retweeted
4 November 17:00 CET International Macro History Online Seminar #IMHOS 🗣️Catherine Schenk (@UniofOxford) presents 'Gaining (and losing) currency: a global database of currencies, 1850–2026' Chair: @flandreaumarc ✍️ ow.ly/lrOr50ZYU1a
2
6
1,278
ct retweeted
Open call for papers, Aging and Health. Conference to be held in Hornbæk, Denmark on June 13–15, 2027. Submit papers by 11:59pm EDT on January 14th, 2027. More information: nber.org/calls-papers-and-pr…
5
25
7,556
Correlación entre religiosidad y PIB per capita (2025)
287
766
202
7,618
1,291,744
How will high gas prices affect the midterms? It's an important question but hard to answer empirically. I've written down a plan to test part of it, and I'll publish my first results on Thursday, when most of the vote is in. 1/🧵 chdausgaard.github.io/fuel-2…
5
24
4
78
7,160
众所周知 Claude 的审美是顶级的,之所以这么强,就是因为Claude有一套自己极其严苛的内部设计规范。 刚刚Claude泄露了它内部的组件设计系统提示词和完整规范,直接把为什么它写出来的界面有一种高级感的秘密全抖出来了。 完整规范整理了出来,写前端和Vibe Coding的可以直接抄作业: 1. 核心设计哲学与流式规范 Seamless(无缝融入):绝对不能让用户察觉到 claude.ai 宿主界面在哪结束、你的组件从哪开始。 Flat(纯粹扁平):严禁使用渐变(gradients)、网格背景(mesh backgrounds)、噪点纹理(noise textures)或任何装饰性特效,表面必须干净平整。 Compact(紧凑克制):内联区域只展示最核心的视觉内容,其余解释全部放在外面的正文里。 文字归回复,视觉归工具:所有的解释性文字、描述、引言和总结,必须作为正常回复写在工具调用外部。工具输出只允许包含纯粹的视觉元素(diagram, chart, interactive widget),严禁在 HTML/SVG 内部塞入解释段落、章节标题或说明性散文(用户的字体设置只对外部回复生效,对微件内部无效)。 流式代码顺序:输出是按 Token 逐个流式传输的,代码结构必须让有用内容尽早出现。HTML 强制顺序:简短的 <style> 放最前 →\rightarrow→ 内容 HTML 居中 →\rightarrow→ <script> 放最后。 2. 基础铁律与视觉禁忌(Rules) 严禁写注释:不准出现 <!-- comments --> 或 /* comments */。 最小字号限制:严禁出现小于 11px 的字号。 严禁使用 Emoji:界面里不允许出现任何表情符号。 严禁花哨特效:不准使用渐变(gradients)、阴影(drop shadows)、模糊(blur)、发光(glow)或霓虹特效(neon effects),唯一例外是功能性的焦点环(functional focus rings)。 外层容器透明:最外层容器严禁设置深色或彩色背景,必须保持透明(由宿主环境提供背景)。 流式传输禁令:流式输出期间严禁使用 Tab 标签页(tabs)、轮播图(carousels)或 display: none 隐藏区块——隐藏内容在流式传输时不可见。必须将所有内容垂直堆叠展示(流式传输结束后用 JS 驱动的分步器 steppers 是允许的)。 3. 图标规范 禁止手搓 SVG:绝对不要自己手写 SVG 路径去画图标。 强制使用 Tabler 线性图标库:环境已预加载 5800+ 个 Tabler 图标(写法如 <i class="ti ti-home"></i>)。 只准用 Outline(线性)风格:严禁使用 -filled 填充后缀(比如 ti-heart-filled 根本没加载,渲染出来是空白)。 样式与无障碍:图标颜色和大小默认继承父元素;纯装饰图标必须加 aria-hidden="true",纯图标按钮必须加 aria-label。 常用预置图标清单:ti-home, ti-settings, ti-user, ti-search, ti-x, ti-check, ti-plus, ti-trash, ti-edit, ti-download, ti-upload, ti-file, ti-folder, ti-chart-bar, ti-calendar, ti-clock, ti-arrow-right, ti-arrow-left, ti-chevron-down, ti-external-link, ti-copy, ti-refresh, ti-player-play, ti-player-pause, ti-heart, ti-star, ti-bell, ti-mail, ti-lock, ti-eye, ti-menu-2。 4. 字体与排版规范 默认字体:全局默认使用 Anthropic Sans;仅在极少数社论感或引用块(blockquote)场景下,使用 font-family: var(--font-voice)。 标题字号与字重(全部固定 font-weight: 500):h1 = 22px h2 = 18px h3 = 16px 标题颜色预设为 var(--text-primary),严禁覆盖修改。 正文排版:字号 16px,字重 400,行高 1.7。 字重死命令:只允许用 400(常规)和 500(加粗)两档。严禁使用 600 或 700,在宿主 UI 下会显得过于笨重。 大小写规范:强制全局使用句首大,严禁使用每个单词首字母大写或全大写,包括 SVG 文字标签和图表标题也必须遵守。 严禁句中加粗:不允许在句子中间加粗文字(包括工具调用前后的回复文本)。实体名、类名、函数名一律用代码样式(code style),加粗(Bold)仅限标题和标签(labels)使用。 5. 全局 CSS 变量与强制深色模式 表面背景(Surfaces):--surface-2(纯白 white)、--surface-1(卡片 card)、--surface-0(页面背景 page bg);语义底色 --bg-{accent,danger,success,warning}。 文字颜色(Text):--text-primary(黑 black)、--text-secondary(次要 muted)、--text-muted(提示 hints);语义文字色 --text-{accent,danger,success,warning}。 边框颜色(Borders):--border(默认极细线 default hairline)、--border-strong(悬停 hover)、--border-stronger;语义边框 --border-{accent,danger,success,warning}。 字体与布局变量:字体 --font-sans、--font-voice(衬线 serif)、--font-mono;圆角 --radius(8px,更大圆角直接写 12px / 16px);内边距 --pad-{sm,md,lg,xl};间距 --gap-{xs,sm,md,lg,xl}。 强制深色模式兼容(Dark mode is mandatory):SVG 节点:使用预置颜色类(c-blue, c-teal, c-amber 等)自动适配深浅色,严禁写 <style> 块定义颜色。 SVG 文字:每个 <text> 元素必须带类名(t, ts, th),严禁省略 fill 或使用 fill="inherit"(在 c-{color} 父级内文字类名会自动适配色阶)。 HTML 文字:始终使用 CSS 变量(--text-primary, --text-secondary),严禁硬编码颜色值。 6. UI 组件 Token 与容器布局规范 容器结构与宽度:微件容器默认是 display: block; width: 100%(容器宽 680px),HTML 自然撑满,不需要额外套 wrapper div,直接写内容即可;需要上下呼吸感时在第一个元素上加 padding: 1rem 0。响应式列统一使用 repeat(auto-fit, minmax(160px, 1fr))。 组件级参数:边框:始终使用极简的 0.5px solid var(--border)(强调时用 --border-strong)。 圆角:大多数元素用 var(--radius)(8px),卡片(cards)固定用 12px。 卡片(Cards):白色背景(var(--surface-2))、0.5px 边框、12px 圆角、内边距 padding: 1rem 1.25rem。 表单元素:input, select, textarea, button, range slider 已预设样式,直接写裸标签即可(文本输入框高 36px 且内置 hover/focus;滑块为 4px 轨道 + 18px 滑块头)。 严禁使用 position: fixed:因为沙盒 iframe 视口高度是靠文档流内容自动撑开的,一旦使用 fixed 定位(比如弹窗 Modal、遮罩层或 Tooltip),视口高度会直接坍缩成 min-height: 100px(h/height 同样需显式遵守高度限制)。 弹窗/遮罩层的标准写法:必须先用一个常规文档流的 <div style="min-height: 400px; background: rgba(0,0,0,0.45); display: flex; align-items: center; justify-content: center;"> 包裹在外层充当“伪视口(faux viewport)”,再把弹窗写在里面。 交互调用:任何需要触发 Claude 思考的交互动作,统一使用 sendPrompt()。 7. 视觉模块与复杂度硬限制 五大模块(按需加载):diagram(SVG 流程图/结构图/说明图)、mockup(UI 原型/表单/卡片/仪表盘)、interactive(带控件的交互解释器)、chart(Chart.js / D3 地理与数据图表)、art(插画与生成式艺术)。 方框副标题限制:≤ 5 个单词。细节必须放在点击穿透(sendPrompt)或下方的正文里,严禁塞进方框内。 颜色数量限制:每张图 ≤ 2 组色阶(ramps)。若颜色代表状态或层级含义,必须加 1 行图例(legend);否则只用 1 组中性色阶。 水平排列限制:全宽下同行 ≤ 4 个方框(每个约 140px)。若达到 5 个及以上,必须缩小到 ≤110px,或者折行成 2 行,或者拆分为“总览图 + 细节图”。 原版Prompt见评论区⬇️
84
229
16
2,222
130,363
All right, I tried to curate a fairly representative cross-section of 13 FLIPR maps for what might happen on Tuesday night. Warning: the last 3 or 4 are deliberately weird! Here's one that came up when the polls were in the 95th percentile of accuracy, however.
507
738
268
6,226
2,100,200
卧槽!有人把中国两千多年的历史,做成了一张可以拖动的动态地图! GitHub 上这个开源项目叫《江山时序》,已经获得 1.9K Star! 从战国到现代,覆盖 18 个历史时期、164 张疆域快照。 最有意思的是,你可以拖动时间轴,亲眼看秦灭六国、三国鼎立、唐宋元明清的疆域变化。 还能查询古今地名、历史事件,甚至跟着李白、苏轼、玄奘的足迹在地图上走一遍! 以前看历史靠死记硬背,现在直接把两千多年历史演给你看。 在线体验:deanchen.me/jiangshan-map/ GitHub:github.com/DeanChensj/jiangs…
101
620
22
2,795
209,536
Really cool! "Global Pension Reform Database 1952 – 2025: From Narrative to Data Using AI" by Yudong Rao and Boele Bonthuis. "This paper introduces the Global Pension Reform Database (GPRD), the first dataset to provide systematic coverage of pension reforms across 142 advanced, emerging, and developing economies from 1952 to 2025. Leveraging artificial intelligence (AI) through a ring-fenced, locally hosted large language model (LLM), we extract and classify more than 4,000 pension reform episodes from the Economist Intelligence Unit and the U.S. Social Security Administration by type, timing, status, and generosity, enabling consistent cross-country and historical analysis. Stylized facts show that pension reform has shifted from an advanced-economy phenomenon to a global and recurring policy tool, often clustered around fiscal and macroeconomic crises. An empirical application demonstrates its analytical potential: in non-advanced economies, implemented pension reforms are associated with statistically significant and persistent declines in the public debt-to-GDP ratio, suggesting that such reforms can contribute to sustained fiscal consolidation." imf.org/en/publications/wp/i…
1
48
3
221
12,617
A must-read! "Macro: The Economic Models That Shape Our World" by Greg Kaplan. link.amazon/B063rA5eF "Macroeconomics and macroeconomic policies shape our lives in profound ways, influencing everything from the price of groceries to the likelihood we’ll have a job next year. But much of what we hear about macroeconomics―from politicians, pundits, and even professors―bears little resemblance to how economists think about policy or to people’s actual experiences. Clearly, economists haven’t done a great job explaining their work to us―until now. Macro is an inviting tour through the world of modern macroeconomics, taking readers to the research frontier―HANK, a framework that will change how we see inflation, debt, recessions, and more? In clear and engaging prose, Chicago economist Greg Kaplan demystifies how our everyday behavior connects to the biggest questions in fiscal and monetary policy. Tracing the evolution of economic thinking from Keynes to today, he guides readers through the models used to explain our economy, uncovers their flaws, and reveals what cutting-edge research tells us about managing the economy in good times and bad. These earlier models are valuable, Kaplan argues, but those that underlie much of conventional wisdom are fundamentally out of touch with the realities of how people spend and save, particularly the millions of Americans living paycheck to paycheck. The journey culminates in the HANK (Heterogeneous Agent New Keynesian) model, the current state of the art that is more consistent with reality. HANK accounts for our diverse spending and saving habits and will transform the way we think about economic policy." Gita Gopinath: "A tour de force. With rare clarity and intellectual power, this book bridges the gap between cutting-edge research and real-world policy debates. Essential reading for anyone serious about macroeconomic policy." link.amazon/B063rA5eF
3
114
696
21,185
The first release of my R package `fourthirtyfive` makes congressional district maps (2005-27, incl. those for next week's election) easily downloadable and mergeable, with hex/dotmaps, real shapefiles, descriptive names, and interactive overlays with #mapbox: shirokuriwaki.com/fourthirty…
5
63
2
463
24,873
"Gold under shifting geopolitical and financial conditions": "Using monthly data from 1990 to 2026, we find that increases in geopolitical risks – both geopolitical threats and realised events – increase gold returns, with stronger effects observed during periods of loose financial conditions." bis.org/publications/working…
5
35
2,592
ct retweeted
Comparing founders who build abroad or return home after foreign experience with comparable local founders in the same country, from Paul A. Gompers and Henry Woo nber.org/papers/w35943
3
2
11
3,834
At Econ 333 (Economics of Inequality at Policy Alternatives) at UMass Amherst, I also introduce students to the concepts and research on optimal top marginal tax rates, and how we can arrive at different conclusions based on our normative criteria.
Replying to @jasonfurman
And a preview of next week when we talk about taxes, three different perspectives on taxation. The utilitarian/Rawlsian focuses on outcomes--which implies setting the top rate to maximize revenue collections. (And doing so in sadness, not in anger.)
1
3
43
4,652
Leah Hunt, a @PennStateEcon student on the job market has written a really lovely paper on diversity and merit. Even if non-merit considerations convey no info about merit, using them in hiring can improve the average merit of hired workers. The argument is new & yet ECON 101
3
21
1
142
10,792
最近发现一个特别适合做「知识可视化」的 Skill:diagram-design。 已经有近 6 万 Stars。 它干的事很简单: 把长文、流程和关系,做成一张专业好看的图表。 对我这种内容创作者来说,主要用在 6 个地方: 01|给文章配专业插图 02|做 PPT 和分享演示的素材 03|把长内容压成一张适合传播的图 04|学一门复杂的新知识 05 |梳理项目和工作流程 06 |做工作汇报 diagram-design Skill 开源库:github.com/cathrynlavery/dia…
28
151
4
1,028
53,533
From our FirstView: "When Politics Enters the Waiting Room: Far-Right Electoral Victories Exacerbate Discrimination in Access to Healthcare" by Merlin Schaeffer, @AlessioRomarri, Dina Rosenberg, and Krzysztof Krakowski. buff.ly/uiaTkVa
8
2
28
2,232
Spatial JMCs: Each year, I compile a list of spatial-economics job-market papers. To make sure you're on my list (& save me some work), please reply with your info in the following format: Firstname Surname (School) - JMP title - homepageURL [Trade JMCs: reply to other tweet]
15
33
109
9,458
Econ PhD job market candidate database is live for 2026-2027 - 678 candidates from 86 institutions so far - this year aiming to add all Bschool Econ adjacent PhDs - filter by field/institution and JMP title/abstract - JMCs can update their own records - connect Codex/Claude Code to the database as you please Dm/comment as new schools post their JMCs and I’ll add them, or if you have issues econ.now/candidates
I'm making a database of every 2025-2026 econ PhD job market candidate: econ.now/candidates - 18 institutions and 307 candidates so far - Filter by institution/field/advisor - CVs, JMPs, links to candidate site Will be adding every finance/econ candidate over coming weeks!
12
52
7
354
60,694