Field Trip Finder
How we classify homeschool field trips
Every destination in Romeschool is run through a 200-line rubric that picks a curriculum-first category, infers a typical age range, and decides whether the venue actually works for a small homeschool group. The short version is below.
Curriculum-first
Categories are picked by what a family teaches there, not by what kind of building it is. A natural history museum routes to science, not museum. A history museum routes to historical. An art museum routes to art. Only when no sharper subject clearly leads does a site land in the general museum bucket — which is why the museum bucket is, counter-intuitively, the smallest of the six.
We deliberately under-weight the art bucket. Many borderline venues — historic homes displaying period art, science centers with sculpture installations — land in historical or science instead, because that’s the curriculum a family is more likely to credit.
The six categories
Every visible destination in Romeschool lands in one of these six buckets:
Battlefields, forts, historic homes, history museums — anywhere the lesson is what happened here.
Preserves, refuges, geological sites, gardens — primary instruction is the natural world itself.
Science centers, planetariums, natural history museums, aquariums, observatories, NASA centers — the lesson is a STEM concept.
Art museums, sculpture parks, studios where the work itself is the curriculum. Deliberately the smallest of the six buckets.
General-purpose museums where no other category clearly leads. Sharp subjects route to history/science/art instead.
Reserved for parks whose primary draw is recreation, not nature instruction. Most parks land under nature.
The group-capable filter
Some destinations work for a single family visit; many don’t work as a homeschool group. Group-capable is a yes/no judgment the rubric makes per site: can a family arrange a visit for a co-op of 6–20 kids without surprise? Walk-up venues with no group policy can still be group-capable. Venues that require timed entry with advance booking are usually group-capable with a phone call. The 6,594 destinations currently classified as not group-capable are hidden by default — but they’re still in the table for future use.
What we deliberately hide
We hide single-exhibit sub-features (a plaque inside a larger museum), historic markers without context, and “named locations” that are really just a building address (e.g. a regional office labelled as a science center). If a family would drive an hour and find nothing to do, that destination doesn’t make it to the visible catalog. Markers, sub-features, and stub records all stay in the underlying table — they just don’t surface in the Discover grid.
Where the rubric still gets it wrong
The 200-line rubric is run by a language model, then sanity-checked against geocoded coordinates, state borders, and a curated list of known venues. It still mis-categorizes the long tail — niche factories, observatories that double as visitor centers, museum ships that share a pier with a historic site. When you spot a wrong category, the destination detail page has a feedback button; we review weekly.