Random cities generator

How it works

Plural is the job. One name is a different page. Here the tool repeats a city draw until it has filled a count you named, then it prints the collection. The pool can be a gazetteer, a custom list, or an assembler of toponyms. The count is a setting. Replacement is a setting. Order is a setting: keep the draw order, or sort the names after, which makes a set easier to scan and destroys the original permutation.

A batch is not wiser than a single draw. It is the same construction, repeated, with one extra risk: the set can contradict itself. Two identical names in a “unique tour” are a replacement problem. A mix of real catalogue hits and invented stems in one column is a labeling problem. Fix those in the settings, not by deleting rows you dislike after you already showed the list to a group.

Length matters. A short set is a weekend sketch. A long set is a table. A set as long as the pool, without replacement, is a shuffle of the pool. You cannot emit more unique names than the pool contains. If the page promises a unique city for every row and the pool is smaller than the count, the page is wrong or it is inventing. Inventing is allowed only when you chose an assembler and you accept that those names are not a census.

A set has rules a single name does not

Decide uniqueness before you click. A load-test column may repeat. An itinerary should not, unless you meant to return. A quiz key should not, unless the quiz is about duplicates. Write the rule on the export. A spreadsheet of names with no rule is a bag, not a specification.

Name the pool: country list, world gazetteer, custom paste, or invented stems. Name the count, and check it against the pool size if you need unique rows. Name replacement: repeats allowed, or each name at most once.

Keep country and script stable across the set if this is a fixture. Export the set. The next click is a different set. Geography inside the set is not guaranteed. Names from a world list can sit far apart. That is a valid random set and a poor road trip. If you needed a cluster, filter the pool first: one country, one coast, one rail network. Randomness inside a filtered pool is still random. Randomness across the whole planet is how a “tour” becomes a list of airport codes you cannot chain.

Sorting the export is fine for a directory. It is not fine for a play order you already announced. If the group saw Paris, then Accra, then Leeds, that order is the artifact. Sorting it into Accra, Leeds, Paris is a new artifact. Keep both files if you needed both jobs.

Using the list without pretending it is a census

Copy the set into a game map, a dummy customer file, a writing table of settings, or a classroom hat. Do not publish it as “cities of the region” unless the pool was exactly that region and you said so. Do not mail real parcels to generated rows. Do not treat the set as a ranking of places worth visiting. The generator does not rank. It fills a count.

A list of random cities becomes an itinerary only after a person puts them in an order a body could travel. If two rows share a name, check whether the pool stored a country column. “Portland” can be more than one settlement. A unique-looking set that only unique-checks the string will collapse two Portlands into one and leave a real second city out. Unique on the string is not unique on the pin. When the pin matters, unique on name plus country, or on a stable id from the gazetteer.

Stop generating when the table is full enough to work. Extra batches are how a test database grows a second unofficial list nobody owns. One exported file, one rule line at the top, one owner. Tomorrow’s click is not that file unless you saved it.