Is your use case "I'm a tourist in New York, who would like to learn more about my whereabouts." or "I live in New York, and I would like to be surprised with historic facts about places in my community", or what?
Most projects aren't going to work, so that can be one of the primary goals, mainly to find out whether or not that's the case, without too many compromises.
So most of the time to be realistic you have to try real hard for it not to work quite early on, or that could be something where you end up wasting too much time on a hopeless effort.
Some people never quit, and it can be difficult to know when to mothball something or give up altogether.
Mothballs can be your friend, especially when there are a greater number of promising projects than could ever be fully completed anyway.
Progress can surprise you though, but more often than not many things will take an unexpected downturn before you ever see a case of an unexpected upturn.
"Would it be fun" you can test by faking it, like curating it by hand for a small specific area (e.g. a popular street in a popular part of a popular city; or your hometown; or somewhere that appeals/you want to visit).
"Can I make a general algorithm using publically available data to do this for everyone, for everywhere, automatically" is the technical hypothesis. It seems like a really challenging problem! But because it is clearly defined and scope-limited, I feel it's more straightforward than the first one.
And to fix your ranking algo you need to understand places more deeply. The core of your product. You mentioned yourself that ranking by page views was flawed.
First of all, it seems that you can probably categorize places into a distinct set of categories. Historical landmarks, locations of past events, locations of ongoing or regular events, buildings for large businesses or government, metro areas (cities, counties, states) etc. Figure out what you actually think is interesting.
Probably don’t include a link to Manhattan on your map next to a local business for example.
Choose a city you’re not familiar with and figure out if the top links are interesting to you. Tweak your algorithm and check another city. Simple isn’t always good, the world is messy and you’ll have a lot of edge cases.