# The launch: who we reach out to, what we say, in what order

> Kept by incident.media@web-summit.dinis, 2026-11-11, 02:37 UTC. **A draft for Dinis Cruz.** From his words of about 02:20 UTC ([the brief](../BRIEF.md)): "a nice launch of this site later today ... planning who we're going to reach out to ... maybe do an outreach email ... everything here can be public ... a lot of this needs to be done via public mediums." **A person sends every message and posts every post; nothing leaves by an agent's hand, and nothing here has been sent.** People appear only in a public professional role, with the public work that makes them relevant; each is reached through the channel they or their outlet publish themselves. No private address or number is kept. Not a warning service: follow [IPMA ↗](https://www.ipma.pt/) and [ANEPC ↗](https://prociv.gov.pt/).

## One decision first: when

Today, Wednesday 11 November, is the day of the storm: IPMA's red warnings for Lisbon run to 21:00 and the country is in an alert situation. Two things follow.

- **Nobody in an operational role is approached while the warnings are in force.** IPMA's and civil protection's press offices come last, after the warnings end.
- **A launch during the storm must not look like a warning service or like self-promotion on a hard day.** Every post below says in its first lines where to go for warnings, and what this site is not.

So the plan has two moments: a **quiet opening** this afternoon in public channels (the site is there, here is the data, tell us what is wrong), and **the launch proper** once the warnings have ended (this evening after 21:00, or Thursday morning). Dinis chooses; the drafts work for either.

## Who, in groups

Each group: who they are, the public work that makes them relevant, what we offer in one line, what we ask in one line, and the public channel.

| Group | Who (public roles only) | What we offer | What we ask | Through |
|---|---|---|---|---|
| **1. Reporters who covered Charles** | The by-lines on the coverage: [the list, with each one's pieces](../research/media/people/README.md). First those whose work the record turns on: the live desks of [Jornal de Negócios](../research/media/outlets/negocios.md), [Diário de Notícias](../research/media/outlets/dn.md), [Observador](../research/media/outlets/observador.md) and [RTP](../research/media/outlets/rtp.md); the reporters with the first articles on Day 2 ([ECO](../research/media/outlets/eco.md), [TEK](../research/media/outlets/tek.md)) and on the mayor's statement ([Jornal Económico](../research/media/outlets/jornal-economico.md)) | A public record of who reported what and when, with each piece linked and credited; a timeline of the decisions with primary sources | Is your time and by-line right? What could you not confirm that night? | The outlet's own public contact page (linked from its page here); a public reply to the open letter |
| **2. Data and visual journalists; newsroom innovation people** | The data desks and labs of the outlets above and of the international press that covered it ([Euronews](../research/media/outlets/euronews.md), [Bloomberg](../research/media/outlets/bloomberg.md)); journalism schools and researchers of news provenance | The data and its API (every file with a sha256), the method (facts, evidence and hypotheses labelled; every item with its source and the time read), the record of lead times | Use the data; tell us where the method fails; what would make it usable in a newsroom | Public posts (below); the open letter; their outlets' public pages |
| **3. Meteorologists and atmospheric scientists who publish in public** | University groups (for example Instituto Dom Luiz, University of Lisbon); forecasters who explain storms in public; national services' public accounts (AEMET posts on Bluesky) | The warnings, communiqués, airport forecasts and station readings of the storm kept hour by hour with every change; the lead-time record (when each warning could first have been known) | Check [the analysis](../incidents/2026-11-10-storm-charles/data/weather/analysis.md): what is wrong or overstated | Their institution's public page; public posts; an issue on the repository |
| **4. IPMA and civil protection (press offices)** | IPMA's communication office; ANEPC's press office; Lisbon's municipal civil protection | A faithful public record of what they published and when, sending every reader to them | Three questions each for the record ([the drafts](drafts.md)); what an aggregator should and should not do during an alert | Their own press contact pages. **After the warnings end** |
| **5. Wikipedia's editors of the article** | The editors of "Depressão Charles" (pt) and of the season article (en), **as a group, on the article's talk page**, never one by one | Sourced times for the warnings and decisions, each with its primary source, free to cite | Is anything in our record at odds with the article? Which of our sources are acceptable there? | The talk page, in a person's own name, declaring the connection to this site (owner: incident.wikipedia@, [research/wikipedia](../research/wikipedia/index.html)) |
| **6. Open data and civic tech** | VOST Portugal (a volunteers' association, active on Bluesky during the storm); the community around IPMA's open data and the national open-data portal; people who build on public APIs | A worked example of public data kept with provenance: the API, the hourly diffs of the warning feeds, the code (open source) | Reuse it; break it; tell us what a second incident would need | Public posts; the repository's issues |
| **7. Web Summit attendees and speakers** | People who posted in public about Day 2 under the event's hashtag; the organisers' media team | One page that says what was decided, by whom and when, and what is still unannounced | When did the organisers' notice reach you, and what did it say changed? (in your own words; we do not take the email itself: [Contribute](../contribute.md), C-04) And any notice the organisers published themselves, with its address and time | Public posts with the event's hashtag; the organisers' media page |

Named people beyond by-lines (individual meteorologists, researchers, editors): none is listed yet, because we have not read their public work closely enough to say why each one. That is [a task](../contribute.md) (C-10) and the next round of this page.

## Where a public post can reach them: Bluesky

From incident.social@'s read at 02:24 UTC on 11 November ([research/social](../research/social/index.html): accounts and reasons in its accounts.json). Organisations only; no individual's account is listed, and we do not look up reporters' personal accounts.

| State on Bluesky | Accounts (organisations) | What it means for the launch |
|---|---|---|
| **Posting about the storm** | VOST Portugal (97 posts; a volunteers' association, not an authority), Agência Lusa (40), Público (9, verified), AEMET (4, verified), Euronews (4, verified), Met Office (1, verified), Reuters (1, verified), Bloomberg (1, verified), Lisboa Para Pessoas (1) | a public post that mentions the record of who reported what can be seen by these newsrooms; VOST and the open-data community are reachable here |
| **Present, silent on the storm** | Web Summit (silent since 10 Nov 17:36), Météo-France, ECMWF, EUMETSAT, AP, The Guardian, Metropolitano de Lisboa (last post in March), Renascença (last post in May), Meteored | not a channel for this launch |
| **No account found** | IPMA, ANEPC, the government, the Câmara, RTP, Observador, Expresso, Diário de Notícias, Jornal de Notícias, CNN Portugal, SIC, ECO, TSF | reach them by their own sites' public contact pages, or on the platforms they use (LinkedIn, X, Facebook: not readable by us, so a person checks) |

So Bluesky and Mastodon suit the quiet opening (open data, civic tech, the wire, two newsrooms); most Portuguese newsrooms and all the authorities are not there, which is why the open letter by email, and LinkedIn, carry the launch proper.

## The open letter (a page on the site, and the email)

To be published at its own address and sent by Dinis from his own address to the public contact of each group. Plain text, short.

> **Subject: Storm Charles, as a public record: tell us what is wrong**
>
> Hello,
>
> On Tuesday a storm closed a day of Web Summit in Lisbon. While it happened, a small team kept a public record of it: every warning IPMA issued and when, every decision (the government's, the city's, the organisers', the transport operators'), and who reported each one first. It is at **storm-charles.promptugal.news**.
>
> What it is:
> - **A timeline of decisions**, each with its primary source and the time we read it. Facts, evidence and hypotheses are labelled and never mixed.
> - **A record of lead times**: how long before each warning took effect it could first have been known.
> - **A media watch**: 200-odd articles from 16 outlets, each linked and credited, and who was first on eleven key facts.
> - **Data for anyone to reuse**: every file is served as JSON with its hash.
>
> What it is not: a warning service. For what to do in a storm, follow IPMA and ANEPC. We make no forecast of our own.
>
> Who runs it: I do, Dinis Cruz, with a team of AI agents that I direct. The agents read public pages, write the record and the code, and sign their work; I decide what is published and I send every message, including this one.
>
> What is not finished: some outlets do not let a machine read them, so they are missing, not ignored. Captures of the pages we cite are being built. Parts of the site are rough. One outlet's terms forbid automated reading and we stopped reading it; if yours do and we missed it, tell us and we stop.
>
> What I am asking: look at the part you know. If your by-line or time is wrong, if a fact is mislabelled, if our reading of a forecast is naive: say so, in public if you can. The page **/contribute.html** lists thirteen things a reporter, a specialist or a reader can do right now, and how. Corrections stay visible, with credit as you prefer.
>
> How to answer in public: open an issue on the repository (linked from the contribute page), or reply to the post this letter is attached to. I read them.
>
> Thank you,
> Dinis Cruz

## Short public posts

Each posted by Dinis from his own account. The first line of each says what the site is not.

**LinkedIn**

> Not a warning service: for Storm Charles, follow IPMA and ANEPC.
>
> On Tuesday the storm closed Day 2 of Web Summit in Lisbon. While it happened, I ran a team of AI agents to keep a public record of it: what was warned and decided, by whom, when, and who reported it first. Every item has its source and the time it was read; facts, evidence and our own analysis are labelled and kept apart.
>
> Some of what it shows: IPMA's red wind warning went out at 07:18; the first report we could read came 47 minutes later. The organisers' page calling off Day 2 is dated 19:09; the first articles appeared 32 minutes after. About half of the storm coverage we could read is one news agency's wire, published by many outlets.
>
> It is unfinished, and some of it will be wrong. That is the point of publishing it: if you covered the storm, forecast it, or managed it, tell us what we got wrong. The data is open.
>
> storm-charles.promptugal.news · how to correct us: /contribute.html

**Bluesky** (300 characters)

> Not a warning service (follow IPMA and ANEPC). A public record of Storm Charles in Lisbon: every warning and decision with its source and time, and who reported it first. Built by a team of AI agents I direct. Unfinished; tell us what's wrong. storm-charles.promptugal.news

**Mastodon** (500 characters)

> Not a warning service: for Storm Charles follow IPMA and ANEPC.
>
> A public, sourced record of the storm in Lisbon: what was warned and decided, by whom, when; who reported it first; the lead time of each warning. Facts, evidence and hypotheses labelled. Open data, with hashes. Built by a team of AI agents I direct; I send every message myself.
>
> It is unfinished. Corrections welcome, in public: storm-charles.promptugal.news/contribute.html
>
> #StormCharles #DepressãoCharles #opendata

**A reply-sized post for reporters** (to accompany the link to an outlet's page here, only where the reporter's outlet invites public replies)

> Your coverage of Storm Charles is in a public record of who reported what and when, credited and linked: [the outlet's page]. Is the time right? Corrections welcome.

## The order, and the time of day

All times UTC (Lisbon is on UTC). Dinis decides each step; nothing is automatic.

| When | Step | Why then |
|---|---|---|
| Before anything | Dinis reads this page, the open letter and [who we would like to talk to](who.md); edits; says yes or no to each group | a person sends |
| Before anything | The checks: the front page loads on a phone; /contribute.html opens issues; the numbers in the posts match [the tables](../research/media/ranking.md) at the time of posting | numbers move every hour |
| Wednesday, early afternoon (quiet opening) | The Mastodon and Bluesky posts; the open letter published as a page | public channels first; small audiences that tolerate unfinished work; the storm is the news, so the post is short |
| Wednesday, afternoon | Open data and civic tech (group 6); data journalists (group 2) by public posts | they can use the data while it is live |
| **Wednesday after 21:00, or Thursday 08:00 to 10:00 (the launch)** | LinkedIn post; the open letter by email to newsrooms' public contacts (groups 1 and 2), each with a link to its own outlet's page | the warnings have ended; newsrooms plan the next day's follow-ups in the morning |
| Thursday morning | Meteorologists and scientists (group 3) with the analysis page | station data is in; the comparison with the forecast is possible |
| Thursday, after operations stand down | IPMA and civil protection press offices (group 4), with the three questions each | not during the event |
| Thursday | Wikipedia talk page (group 5), by a person in their own name; Web Summit's media team and attendees (group 7), on Day 3 | Day 3 runs on Thursday if the organisers' aim holds |
| Friday | A follow-up post: what people corrected, with credit | corrections stay visible |

## What a person must check before sending

- That the site still says what the letter says it says (the counts, the times).
- That no message goes to an operational address during the warnings.
- That The Portugal News (and any outlet whose terms forbid automated reading) is approached, if at all, as a request for permission and not as "your coverage is in our list": it no longer is.
- That every reply is published only as its author agrees.