← Work

A transport atlas

Booking sites need both ends of a journey and a date. waypoints.now answers the question that comes before that one: where can I get to from here, and what shape is the journey?

Problem: You cannot search for a destination you have not thought of yet
Scale: 26,034 crossings · 2,125 terminals · 115 countries
Role: Architect, builder and editor
Status: Live · documented 11 Aug 2026

waypoints.now homepage: the headline ‘How to get there, without changing’, a search field, and three statistics cards. — The atlas states its own size and its own limits in the first paragraph.

A polar chart centred on Athens. Every place reachable without changing is plotted by true bearing and distance, in rings at 335, 928, 1,816 and 3,000 km, coloured by mode. — Athens: 436 places reachable without changing, plotted by true bearing and distance. The shape of the reachable world from one city, which no timetable can show you.

The problem

Every transport search assumes you already know where you are going. Type an origin, a destination and a date, and you get fares. But the decision that matters happens earlier — when someone is standing in a place, or committed to one, and wants to know the shape of what is reachable. That question has no interface. Booking engines cannot answer it because their index is built for pairs, not for reach.

The second problem is quieter. A comparison site showing one result is not showing you a choice, and a crossing with no published duration is usually dropped rather than declared — which makes the network look tidier and smaller than it is.

What I built

An atlas rather than a search engine. Every city, terminal, port and operator gets a page showing everything it reaches, so the unit of navigation is reach rather than route. Crossings are compared door to door across sea, air, rail and road, and where the data has a hole, the hole is named on the page.

The Greece country page: 987 crossings between 82 ports, with cities ranked by how many destinations each reaches, shown as horizontal bars. — Greece: 987 crossings, 82 ports, 63 cities — each city measured by what it reaches, not by how important it is.

The comparisons are the part a booking site structurally cannot publish, because they are unflattering to the operators it sells for:

119 of 593 crossings sold as fast average under 45 km/h.
9,270 of 13,628 operator crossings have no rival at all.
1,579 crossings are untimed — and listed anyway, rather than quietly dropped.

A single route page, Anafi to Thira, listing every operator with a duration bar, and marking one operator’s crossing time as not held. — One row per operator and speed. Where a crossing time is not held, the page says so rather than estimating.

The decision that shaped it

The atlas holds no fares and no live departure times, and says so on every page. That looks like a weakness and is the reason the thing can be trusted: a planning aid that guessed at times would be wrong seasonally, silently, and at scale. Naming the gap costs a little credibility once. Filling it with a guess costs all of it, later.

What it demonstrates

Reorganising a dataset around the question a person actually has, rather than the question the existing tools were built to answer — then publishing the parts that do not flatter the industry the data came from. The same shape recurs in timesys.pro, where the unit is a jurisdiction rather than a date.

Visit waypoints.now →