Product update · September 2026
The month we opened the doors
What we shipped, what you broke and we fixed, and what we are building next. One of these every month from now on; the numbers of the launch itself are in the launch report.
Launched: Target Network 2032, first version
22 September. After an August testing party with 29 testers and 36 trial proposals, the tool opened to everyone. This is what it could do on day one.
- Launched
Draw a night train on the real rail network
Pick two cities, add stops from a catalogue of about 1 200 stations, and the route is built over the actual European rail network — gauge-aware, so a Spanish or Baltic route runs where a train physically can.
- Launched
A realistic timetable
Departure and arrival times per stop, buffer and turnaround, a day-of-week pattern and the number of trainsets it takes — in both directions.
- Launched
Twelve train concepts
From a 6-coach budget set to a 14-coach pod train or an all-private-room formation, each with places, weight, speed and crew modelled — and your own composition can be suggested from the gallery.
- Launched
A full-year evaluation
What the service costs per train-kilometre and per year — track access, energy, crew, rolling stock, stations, shunting, parking — against revenue under a stated demand assumption, the subsidy it would need, and the CO₂e it would save. Every figure with its formula and source one click away in this documentation.
- Launched
A gallery to browse and discuss
Every published proposal on a map and in a list, searchable by city, station and country, with likes, comments and a share link that previews in chat apps.
- Launched
A first demand and supply model
Supply: how many nights a week, how many trainsets, how many seats on offer. Demand: a potential you set per route, seated by traveller group and class, with a stated assumption of who would otherwise have flown, driven or not travelled — simple by design, and the first thing on the list to replace (how it works).
- Launched
Scenarios: today's network, and two operating conditions you can switch on
Every proposal is evaluated on Infrastructure 2026 — the rail network as it is. On top of that, two policy options: night trains may use high-speed lines, and optimised timetables instead of the padding night trains carry today. Each figure is shown against the baseline, so the difference is the scenario's alone. Infrastructure 2032 and price measures are prepared but not yet switched on (scenarios).
- Launched
Open from the first day
Sign in with an e-mail code or try it as a guest; English and German interface; the whole model documented here; feedback forms on every documentation page; source code public on GitHub.
Fixed and improved in the first two weeks
22 September – 4 October. Ordered by how many of you it affected.
- Fixed
Gallery loading
The gallery failed or stalled for a share of visitors in the first days; the loading path was rebuilt and the gallery now opens reliably, also on slow connections.
- Improved
Mobile devices
30 % of all requests came from a phone — far more than we expected. The builder, the gallery and the proposal page were reworked for small screens: layouts that stack, controls that are thumb-sized, maps and overlays that do not fight the page scroll. More of this is on the list below; this was the first pass.
- Improved
Gallery search and filters
Searching is now three places × two kinds — by station, by city or by country, and either via (routes that call there) or from → to (routes that call at both places, in either direction). A new distribution panel shows how the proposals spread over length, travel time, speed and number of stops, lets you set a range on each, and opens on the typical night trains preset: 500–2 000 km one way, 7–21 hours, at least 50 km/h — so the gallery shows night trains rather than people testing where the tool breaks. The same yardstick is what the launch report's figures use.
- Improved
Map display
Hovering a proposal now draws its stops on the map, a searched station stays pinned, a searched country is tinted; and proposal length and travel time are shown per direction instead of for the round trip.
- Fixed
Under the hood
A number of smaller internal fixes: stop catalogue entries, routing-graph gaps reported by you, version and deployment plumbing. None of them changed a figure you would notice; all of them made the next changes easier.
Most of these started as something you told us. Every piece of feedback — the forms on the documentation pages, the cost-factor buttons in the builder, comments in the gallery, mails to the working group — is read by the development team. We decide together what goes in and when, and the list above is the result of that for this month. Nothing is answered individually by default, but everything is counted: the launch report shows what came in.
What we are building next
October and after. In the order we intend to tackle them. Dates on purpose absent: this is a volunteer project, and the list is honest rather than scheduled.
- 1
A frontend shaped by what we saw
Input options and the builder reshaped around how you actually used it in the first weeks.
- 2
Mobile first
The first pass is in; the next makes the whole tool usable with a thumb — builder, gallery and comparison.
- 3
A place to talk
Reacting to each other's routes should be as easy as drawing one — notifications for authors, discussion visible from the gallery.
- 4
Routing fixes and stop additions
The detours you reported and the stations you searched for in vain go into the next catalogue and routing-graph update.
- 5
An enhanced demand model
The current demand assumption is the simplest one that works. The next one will be a model of who actually travels where.
- 6
A refined cost model
Your feedback on cost factors — driver hours, track access, energy — goes into the next calibration round, component by component.
- 7
More scenarios
Infrastructure 2032 and other what-if settings as selectable scenarios, so a proposal can be read under more than one future.
- 8
X- and Y-shaped proposals
Trains that split and join under way — the shape many of the best real night trains have.
- 9
A network analysis mode
Analyse several proposals together as one network: shared stock, shared stations, combined economics.
Something missing from this list? Any other idea or suggestion — send it through the feedback form under Feature request. That is where the next version of this list comes from.
Read next
The launch in numbers.
451 people, 1 321 proposals, 94 % of them crossing a border, and 924 that look like a night train could actually run them — the launch report is the wrapped-up version of the first 12 days, built from the production database.
Written by the target network working group. Versions at the end of the month: backend 0.5.x, frontend 0.1.x, documentation site as published.