Editorial standard

Aware

Plans can be disrupted by how reality turns out. An entry says what has been announced or forecast and what it could do to a trip; it cannot say what will really happen. Knowing a risk shows you what could happen, so you can decide better.

What we cover

Only risks that are dangerous or regularly disrupt travellers and organisations:

- Extreme weather: official orange and red warnings (yellow is not covered); heat-wave bulletins at level 2 (orange) and 3 (red) from the Ministry of Health, entered by the editor; avalanche danger 4 (high) and 5 (very high) from the AINEVA bulletins; Venice high water forecast at 110 cm or more by the Comune's tide centre ("very high"; 140 cm and over is "exceptional"). - Seasonal dangers: periods an authority declares, such as a region's wildfire-danger period. Hazards that have started to hit the same areas again and again will be added once the method for establishing them is published. - Volcanic ash: the Etna Observatory's orange and red notices for aviation (Etna, Stromboli), which can close Catania airport or reroute flights. - Strikes that hit critical infrastructure: air, rail and sea transport, general and multi-sector strikes, local public transport nationwide or in the 14 metropolitan cities, national health-service strikes, and energy strikes entered by the editor. Freight, helicopter, rail-contractor and small-town transit strikes are not covered. - Protests and public order: announced demonstrations that could turn violent, football matches the national Observatory rates high-risk; entered by the editor, each citing the page that announced it.

Events that have already happened are kept as a record but not published: they belong to no calendar.

What we publish

We publish only what a named, linked source has announced, forecast or scheduled. We do not make predictions. We relay official warnings and read them for their direct impact on travel.

Every entry shows its source, the time we last checked it at the source, and its full status history. Nothing is deleted: revoked and no-impact entries stay, because they are the accuracy record.

Warnings are relayed exactly as issued by the competent authority. The travel-impact reading is ours, is shown separately and marked as ours, and the rules that produce it are published.

We describe facts, not actors. A strike entry names the sector, the proclaiming organisations and the dates; a protest entry names the place, the time and the announcement it comes from.

Our vocabulary

Horizon: seasonal (a period with a known, elevated base rate), forecast (an official warning for a defined window), scheduled (a date fixed by law, decree or announcement).

Status: announced (first seen, awaiting confirmation), confirmed (issued by an official warning service, or still listed by its source at least 24 hours later), amended (changed at source; the change is shown), revoked (withdrawn by the source, or lowered below orange; kept and struck through), occurred (window passed, verification pending), verified (outcome assessed with evidence).

Outcome: realised, partial, no impact, revoked, insufficient evidence. An outcome other than revoked or insufficient evidence always needs an evidence link or note. Verification without evidence is not verification.

The same everywhere

The website, the calendar feeds and the emails show the same entries. When a source lowers a warning below orange, the entry is withdrawn with that reason: calendar apps drop it and people who had an alert are told.

An entry keeps the level its primary source issued. When another source reports a different level, both are shown; we never raise the headline to the highest level reported.

The danger table

The warning level (yellow, orange, red) is the authority's reading of one hazard. The danger table is ours: it ranks how dangerous and disruptive an entry can be, on four criteria, each graded from A (worst) to E (mildest). You see it as a small badge beside each entry, for example "Danger C B·A·A·B": the first letter is the overall rank, then the four criteria in the order below.

Where the grades come from. Geographic spread is measured, never estimated, from the area the entry is placed on. The other three criteria start from a baseline for that kind of event (a red hydrogeological warning, a rail strike, a march), written down in the rule table on the methodology page and confirmed by a person. The baseline is not information about the specific event: where an entry uses it, the entry page says so. Marches that the source itself links as one movement share the footprint of all their places, so each one is measured on the movement, not on its own city. When we assess an event ourselves, the entry says that too.

A criterion that makes no sense for a kind of risk is left out, not graded E: a heat warning has no meaningful cost. An entry needs at least two criteria to be ranked.

The overall rank is the mean of the criteria that apply (A is 5 points, E is 1), turned back into a letter. It exists to order entries. It is not a probability and it is not the warning colour. A high rank describes what could happen, not that it will, and a low rank does not make a trip safe. What each letter means:

How we keep score

Every week the scorecard counts the published entries whose window ended (weather warnings, strikes and protests; declared periods such as seasons make no claim about disruption, so they are not assessed): how many were verified, realised, partly realised, had no impact, were revoked, or could not be assessed. Every number is published with its n: "3 of 3", not a bare 100%.

A percentage is shown only once a group has at least 30 judged outcomes. Below that it would be invented precision.

We also count disruptions that happened with no prior entry from us (misses), the median lead time between publication and the start of the window, and the median relay latency from the authority's issue time to our publication, including the time a human spends reviewing.

Nothing is excluded retroactively. If an impact rule was wrong, the entry stays and the rule changes. The scorecard is generated from the same tables the site uses.

Human review

Orange and red entries, entries without a matching impact rule, and anything our systems cannot place or classify wait for a human before they appear. Unknown hazard codes from a source are never dropped silently: they are held for review. Protests are entered by hand and reviewed like any other entry.

The exception is the official weather warnings from MeteoAlarm, which we relay as soon as they are issued and placed, exactly as issued. For those, the human review applies to our travel-impact line only: until it is checked the entry says so, and alert emails go out at once with a note that the assessment is under review (no second email follows when it is done). A warning that is escalated stays visible while its impact line is checked again.

Corrections are made on the entry itself and visible in its status history.

What DetourData is not

It does not predict terrorism, crime, accidents or anything that has not been announced: a protest appears only once it has been announced. It does not track people. It does not replace a security or assistance provider. Warnings belong to the authorities that issue them; our impact statements are general information, not advice for any particular trip.