General observation. But there's lots of tours of all kinds, and even locals go on them including walking tours.

ok, bsky sucks - just read the rest on linkedin:
www.linkedin.com/feed/update/...

Math: I'm looking for the LLM to correctly sum the values from the sub-regions and not (e.g.) double-count by both zip and CB. I'm also looking for reasonable answers in the tighter range of 500K-1M

Iteration: followup: "can you do better by breaking this down by areas and gathering real data about those areas? For example, census data by zipcode, CB, or other region?" I'm looking for analysis+data. CB="community board" (well known in NYC + has public data), e.g. lnkd.in/eVtfGJbC

Reasonableness: a quick initial answer is fine but it should be somewhere between 200,000 and 1,000,000 - Manhattan is well documented to be in the mid-1Ms and below CP is a sizable percentage.

Language and context: LLMs should correctly guess that "central park" means NYC, "population" means residents, not tourists/commuters and "below" means south and not underground (!).