Lilambd/world-pulse-daily

World Pulse Daily Census

15 public catalogues, written down every day, starting 2026-11-03 (tables added later start later). Each source shows only today (a chart, a counter, a list). Kept day after day, they become a history nobody else has.

folder	what is recorded each day
arxiv/	arXiv new submissions per category (day = 3 days before fetch)
crossref/	Crossref DOIs registered yesterday by type, top 100 publishers, licences
zenodo/	Zenodo records created yesterday by resource type, access, file type, subject
wikipedia_size/	Wikipedia size per language: articles, pages, edits, users, active users (cumulative)
wikipedia_top/	Top 100 most-read articles per language (day = 2 days before fetch)
appstore_charts/	Apple App Store top-free and top-paid 100 in 28 countries
stablecoins/	Stablecoin supply per coin (top 80, with per-chain split) and per chain
fx_ecb/	ECB euro reference rates
github_new_repos/	GitHub repositories created yesterday: total, by language, top 100 by stars (age 1 and 7 days)
ai_sdk_downloads/	Daily downloads of AI / agent / x402 SDKs on npm and PyPI, with baseline packages
hf_daily_papers/	Hugging Face Daily Papers of yesterday with upvotes
hackernews_front/	Hacker News front page top 30
stackoverflow_tags/	Stack Overflow totals, top 100 tags and 24 AI tags (cumulative question counts)
crates_io/	crates.io totals and new / most downloaded / just updated crates
docker_official_images/	Docker Hub official images: cumulative pulls and stars
summary/	rows per table per day, so a gap is visible

Sources (all public, read through their official APIs or feeds): arXiv API; Crossref REST API (CC0 metadata); Zenodo REST API; Wikimedia Action API and Pageviews API; Apple Marketing Tools RSS; DefiLlama stablecoins API; ECB rates via frankfurter.app; GitHub REST search API; npm downloads API and pypistats.org; Hugging Face API; Hacker News Firebase API; Stack Exchange API (CC BY-SA); crates.io API; Docker Hub API.

Rules: each file is <folder>/<fetch-date>.csv; rows carry their own date when it differs from the fetch date (arXiv d-3, Wikipedia top d-2, yesterday-type counts d-1). Cumulative counters (Wikipedia size, Stack Overflow tags, Docker pulls, crates totals) become daily flows by taking the difference between consecutive days. A missing day means the census did not run, not zero. No personal data is collected beyond public catalogue names (repository and app names as published). Maintained by Lilambd. Related: Lilambd/machine-economy-daily, Lilambd/world-signals.

Quick start: datasets, pandas, DuckDB

from datasets import load_dataset
fx = load_dataset("Lilambd/world-pulse-daily", "fx_ecb", split="train")
hn = load_dataset("Lilambd/world-pulse-daily", "hackernews_front", split="train")

import pandas as pd  # hf:// paths need `pip install huggingface_hub`
BASE = "hf://datasets/Lilambd/world-pulse-daily/"
summary = pd.read_csv(BASE + "summary/2026-11-03.csv")
print(summary.sort_values("rows", ascending=False))
hn = pd.read_csv(BASE + "hackernews_front/2026-11-03.csv")
print(hn[["rank", "score", "title", "comments"]].head(10))

-- DuckDB 0.10+ reads hf:// paths directly
-- rows per table per day (a gap shows as a missing day)
SELECT fetch_date, "table", "rows"
FROM read_csv('hf://datasets/Lilambd/world-pulse-daily/summary/*.csv', union_by_name = true)
ORDER BY fetch_date, "rows" DESC;

SELECT date, currency, rate
FROM read_csv('hf://datasets/Lilambd/world-pulse-daily/fx_ecb/*.csv', union_by_name = true)
ORDER BY date DESC, currency;

A runnable notebook is in notebooks/quickstart.ipynb.

Use cases

• Daily histories of catalogues that only show today (arXiv, Crossref, Zenodo, Wikipedia, app stores, stablecoins).
• Turn cumulative counters into daily flows by differencing consecutive days.
• Cross-source attention and open-source activity dashboards.

Update cadence

Daily census from 2026-11-03: one <folder>/<fetch-date>.csv per day; a missing day means the census did not run.

Citation

@misc{lilambd2026_world_pulse_daily,
  author       = {Lilambd},
  title        = {World Pulse Daily Census},
  year         = {2026},
  publisher    = {Hugging Face},
  howpublished = {\url{https://huggingface.co/datasets/Lilambd/world-pulse-daily}},
  url          = {https://huggingface.co/datasets/Lilambd/world-pulse-daily}
}

Please also credit the original sources listed in this card.

Other languages

日本語

• World Pulse Daily Census は、「今日」しか表示しない15の公開カタログを毎日書き留め、履歴にしたデータセットです。
• 記録は 2026-11-03 から毎日、公式APIやフィードを通じて公開ソースから取得しています（arXiv、Crossref、Zenodo、Wikipedia、App Store、ステーブルコイン、ECB為替レート、GitHub、Hacker News など）。
• ファイル構成：各ファイルは <folder>/<fetch-date>.csv（例：fx_ecb/、hackernews_front/）。取得日と異なる場合、各行は自身の date を持ちます。
• 累積カウンター（wikipedia_size/、stackoverflow_tags/ など）は、連続する日の差分を取ると日次の増分になります。
• 欠けている日は「センサスが実行されなかった」ことを意味し、ゼロではありません。summary/ で欠落を確認できます。
• ライセンス：CC BY-SA 4.0。元のソースのクレジットも記載してください。

中文

• World Pulse Daily Census 每天记录 15 个只显示"今天"的公开目录，日积月累形成历史数据。
• 自 2026-11-03 起每日通过官方 API 或数据源从公开来源采集（arXiv、Crossref、Zenodo、Wikipedia、App Store、稳定币、欧洲央行汇率、GitHub、Hacker News 等）。
• 文件布局：每个文件为 <folder>/<fetch-date>.csv（如 fx_ecb/、hackernews_front/）；若数据日期与抓取日期不同，行内带有自己的 date 字段。
• 累计计数（如 wikipedia_size/、stackoverflow_tags/）可通过相邻两天相减得到每日增量。
• 缺少某一天表示当天普查未运行，而不是数值为零；summary/ 可用于发现缺口。
• 许可证：CC BY-SA 4.0。请同时注明原始数据来源。

Español

• World Pulse Daily Census anota cada día 15 catálogos públicos que solo muestran hoy, para convertirlos en un historial.
• Se registra a diario desde 2026-11-03, a partir de fuentes públicas leídas mediante sus API o feeds oficiales (arXiv, Crossref, Zenodo, Wikipedia, App Store, stablecoins, tipos de cambio del BCE, GitHub, Hacker News, etc.).
• Organización: cada archivo es <folder>/<fetch-date>.csv (p. ej. fx_ecb/, hackernews_front/); las filas llevan su propia date cuando difiere de la fecha de descarga.
• Los contadores acumulados (wikipedia_size/, stackoverflow_tags/, …) se convierten en flujos diarios restando días consecutivos.
• Un día ausente significa que el censo no se ejecutó, no un valor cero; summary/ permite ver los huecos.
• Licencia: CC BY-SA 4.0. Cite también las fuentes originales.

Français

• World Pulse Daily Census consigne chaque jour 15 catalogues publics qui n'affichent que aujourd'hui, pour en faire un historique.
• Relevé quotidien depuis le 2026-11-03, à partir de sources publiques lues via leurs API ou flux officiels (arXiv, Crossref, Zenodo, Wikipedia, App Store, stablecoins, taux de référence de la BCE, GitHub, Hacker News, etc.).
• Organisation : chaque fichier est <folder>/<fetch-date>.csv (par ex. fx_ecb/, hackernews_front/) ; les lignes portent leur propre date lorsqu'elle diffère de la date de collecte.
• Les compteurs cumulés (wikipedia_size/, stackoverflow_tags/, …) donnent des flux quotidiens par différence entre jours consécutifs.
• Un jour manquant signifie que le recensement n'a pas tourné, pas une valeur nulle ; summary/ rend les trous visibles.
• Licence : CC BY-SA 4.0. Merci de créditer aussi les sources d'origine.

Deutsch

• World Pulse Daily Census hält täglich 15 öffentliche Kataloge fest, die nur den heutigen Stand zeigen, und macht daraus eine Zeitreihe.
• Täglich erfasst seit 2026-11-03, aus öffentlichen Quellen über deren offizielle APIs oder Feeds (arXiv, Crossref, Zenodo, Wikipedia, App Store, Stablecoins, EZB-Referenzkurse, GitHub, Hacker News u. a.).
• Aufbau: jede Datei ist <folder>/<fetch-date>.csv (z. B. fx_ecb/, hackernews_front/); Zeilen tragen ein eigenes date, wenn es vom Abrufdatum abweicht.
• Kumulative Zähler (wikipedia_size/, stackoverflow_tags/ …) ergeben tägliche Flüsse als Differenz aufeinanderfolgender Tage.
• Ein fehlender Tag bedeutet, dass der Zensus nicht lief, nicht null; summary/ macht Lücken sichtbar.
• Lizenz: CC BY-SA 4.0. Bitte auch die Originalquellen nennen.

Português

• World Pulse Daily Census registra todos os dias 15 catálogos públicos que só mostram o hoje, transformando-os num histórico.
• Coleta diária desde 2026-11-03, a partir de fontes públicas lidas por suas APIs ou feeds oficiais (arXiv, Crossref, Zenodo, Wikipedia, App Store, stablecoins, taxas de referência do BCE, GitHub, Hacker News etc.).
• Estrutura: cada arquivo é <folder>/<fetch-date>.csv (ex.: fx_ecb/, hackernews_front/); as linhas têm sua própria date quando ela difere da data de coleta.
• Contadores cumulativos (wikipedia_size/, stackoverflow_tags/ …) viram fluxos diários pela diferença entre dias consecutivos.
• Um dia ausente significa que o censo não foi executado, e não zero; summary/ deixa as lacunas visíveis.
• Licença: CC BY-SA 4.0. Credite também as fontes originais.

한국어

• World Pulse Daily Census는 '오늘'만 보여 주는 15개 공개 카탈로그를 매일 기록해 이력으로 만든 데이터셋입니다.
• 2026-11-03부터 매일 공식 API나 피드를 통해 공개 출처에서 수집합니다 (arXiv, Crossref, Zenodo, Wikipedia, App Store, 스테이블코인, ECB 환율, GitHub, Hacker News 등).
• 파일 구성: 각 파일은 <folder>/<fetch-date>.csv (예: fx_ecb/, hackernews_front/)이며, 수집일과 다를 경우 각 행에 자체 date가 있습니다.
• 누적 카운터(wikipedia_size/, stackoverflow_tags/ 등)는 연속된 날의 차이를 구하면 일별 증가량이 됩니다.
• 빠진 날은 조사가 실행되지 않았다는 뜻이며 0이 아닙니다. summary/에서 누락을 확인할 수 있습니다.
• 라이선스: CC BY-SA 4.0. 원본 출처도 함께 표기해 주세요.

العربية

• World Pulse Daily Census يدوّن يوميًا 15 فهرسًا عامًا لا يعرض كلٌّ منها إلا بيانات اليوم، ليحوّلها إلى سجلّ تاريخي.
• يُسجَّل يوميًا منذ 2026-11-03 من مصادر عامة عبر واجهاتها البرمجية أو خلاصاتها الرسمية (arXiv وCrossref وZenodo وWikipedia وApp Store والعملات المستقرة وأسعار البنك المركزي الأوروبي وGitHub وHacker News وغيرها).
• بنية الملفات: كل ملف هو <folder>/<fetch-date>.csv (مثل fx_ecb/ وhackernews_front/)، وتحمل الصفوف حقل date خاصًا بها عندما يختلف عن تاريخ الجلب.
• العدادات التراكمية (wikipedia_size/ وstackoverflow_tags/ …) تتحول إلى تدفقات يومية بطرح كل يوم من اليوم التالي له.
• اليوم المفقود يعني أن التعداد لم يُشغَّل، وليس صفرًا؛ ويُظهر summary/ الفجوات.
• الترخيص: CC BY-SA 4.0. يُرجى أيضًا نسبة الفضل إلى المصادر الأصلية.

हिन्दी

• World Pulse Daily Census ऐसे 15 सार्वजनिक कैटलॉग को हर दिन दर्ज करता है जो केवल आज दिखाते हैं, ताकि उनका इतिहास बन सके।
• 2026-11-03 से प्रतिदिन, सार्वजनिक स्रोतों से उनके आधिकारिक API या फ़ीड के ज़रिए दर्ज किया जाता है (arXiv, Crossref, Zenodo, Wikipedia, App Store, स्टेबलकॉइन, ECB विनिमय दरें, GitHub, Hacker News आदि)।
• फ़ाइल संरचना: हर फ़ाइल <folder>/<fetch-date>.csv है (जैसे fx_ecb/, hackernews_front/); जब तारीख़ fetch-date से अलग हो, तो पंक्तियों में अपनी date होती है।
• संचयी काउंटर (wikipedia_size/, stackoverflow_tags/ आदि) लगातार दो दिनों का अंतर लेकर दैनिक प्रवाह बन जाते हैं।
• किसी दिन का न होना मतलब उस दिन जनगणना नहीं चली, शून्य नहीं; summary/ से अंतराल दिखते हैं।
• लाइसेंस: CC BY-SA 4.0। कृपया मूल स्रोतों को भी श्रेय दें।