A destination is not always a city

Most trip planners quietly assume the destination is a city. Type a city and everything downstream works: one place to geocode, one climate record, one set of hotels, one list of restaurants. Type “Kerala”, or “Sicily”, or “Japan”, and the assumption collapses — none of those is somewhere you can book a room.

But that is how people actually describe trips. Almost nobody decides to spend two weeks in Kochi. They decide on Kerala, and work out the towns afterwards. A planner that cannot take the input people naturally give is making them do the planning before they are allowed to plan.

Bases, not centroids

So a destination resolves to an ordered list of bases: the towns people genuinely sleep in when they visit that region, in a sensible order for moving between them. A state, a region, an island or a country all go through the same step.

The obvious shortcut — take the geographic centre and search around it — produces exactly the itinerary you would expect. The centre of a region is very often a field. The right answer is a curated list, falling back to the significant cities of a country where nothing curated exists, and never to a coordinate that happens to be halfway between two places worth going.

Why the bases must come out identical every time

A long trip is composed over several passes, and each pass rebuilds the request from what was saved. If the resolver were even slightly non-deterministic — a shuffled tie, a live search — the bases of a thirty-day Kerala trip could differ between pass one and pass six, and the itinerary would relocate the traveller halfway through. So the resolver is pure: same input, same list, forever.

What the bases then decide

Once there is a list of bases, everything else lines up behind it. Places are verified per base rather than for a region that has no centre. The itinerary is composed in windows of days that never straddle a base change, so a single day is never half in one town and half in another. Stays are chosen per base. Travel between them becomes an explicit leg in the day's timeline rather than an unexplained jump.

That last part matters more than it sounds. The failure mode of region-level planning is an itinerary that reads beautifully and silently assumes a two-hour drive between breakfast and the first museum. Making the base explicit makes the drive explicit.

It is not a glamorous piece of the system. It is the piece that decides whether “ten days in Sicily” produces a trip or a list.

Put it to the test

Every claim in this post is something you can check on a real plan. One sentence is enough to start one.

Plan a tripOr tell us where we are wrong