Hacker News Reader

Comment by lukejduncan | Hacker News Reader

As an example, if you search for Dearborn, MI on AirBnB it takes you to Detroit instead https://www.airbnb.com/s/Dearborn--MI--United-States/homes?p...

That's an interesting choice. I wonder if this is even intentional or a bug?

View on HN

I find that lodging searches by place name can sometimes be sort of tricky. For example, I probably don't want to search all of Manhattan and certainly not all of New York City. On the other hand, while I personally am pretty familiar with neighborhoods, to the uninitiated the long list of neighborhood filters offered up by Expedia is probably pretty intimidating and useless. At that point, you probably just do a map-based search but it's hard to say what the ideal approach that can be handled algorithmically is.

Map based searches are good, but one step further would be to allow the user to geofence the locations they are looking for. When I search for locations on Google, I get a map view that spans from Brooklyn across Manhattan and into New Jersey and the results tend to be scattered across those areas when all I want is Manhattan, including areas somewhat outside of the current view.

What you really want is to say/type "Hey Google. I have a conference at the Javits and I want to stay somewhere within a 20 minute walk near some good restaurants. And I don't want to pay more than $300 per night." You pull back to what you really want to do and you realize just how horrible and limited the current systems are.

For sure!

Agreed. Is that a problem that can be solved and what keeps companies from doing this?

I assume the usual answer. $$. To do a really good job, you need to program in local knowledge that probably isn't easily encoded in algorithms.

Manhattan is actually pretty easy. In many parts of Manhattan it really is mostly about just drawing a reasonable bounding box.

But for a random location, getting to a result that you'd get conversing with an intelligent assistant familiar with the area is going to be tough. You have to apply a lot of your own judgement and, even then, I definitely mess up now and then with cities I'm unfamiliar with.

And most people pay for low prices. Not incrementally more sophisticated recommendation engines.

Yeah I'd say that hurts the people who want to stay there too. If I search for a specific town I want places near the town not places near the biggest hub.

Dearborn appears(?) pretty close to Detroit, and probably has a much wider range of housing options.

Reply on news.ycombinator.com