openaddresses VS feedback

It boils down to the fact that the United States does not have (public domain) knowledge of every address in the country.

The USPS knows about deliverable addresses but won't give that information to the federal government because then it'd be public domain and they would lose several of their primary data moats (Zipcodes, addresses, delivery routes, for example). The Census has very complete knowledge of every address, but won't give it up because it's illegal (see Title 13 of the US Code). There is an ongoing attempt by the DOT to collect a National Address Database (https://www.transportation.gov/gis/national-address-database) by collecting information from the address assigning authorities (usually county governments), but it's incomplete and unlikely to ever be complete because of holdout/underfunded local governments.

There are several address datasets that are private (Google has a fairly complete one, FedEx/UPS probably have the most complete, TomTom, CostQuest, etc.). I started https://openaddresses.io/ to try and collect them (NAD is based off this idea) into an open-licensed dataset.

The broadband companies have records that say "this address is connected to this network, which could theoretically have this service level", but (a) they won't/can't tell you where they think the address is and (b) won't spend the time to match their address string format with the government's address because both are private data.

Finally, without the address -> location data, even if we could get broadband providers to tell us what service is available at each address, we couldn't put that service level on a map because we don't know where the address is.

----

The Markup published some work in 2022 where they used OpenAddresses to use ISP's own tools to gather per-address service offerings and put them on a map. This is what the FCC's broadband map should be doing, but can't for the above (and political) reasons: https://themarkup.org/show-your-work/2022/10/19/how-we-uncov...

This is a screenshot of all the USA data from https://openaddresses.io/

https://i.imgur.com/nHIYtap.png

It's about 80GB un-compressed. I wanted a postgres table like:

zip | integer

You might look at https://openaddresses.io bulk download, but that may not help. There's doubtless a list of all businesses in the USA somewhere, and I'll update this post if I find it. The "canonical" source would be to go to each state government and find the list of registered companies which is usually publicly available

Try https://openaddresses.io/

I would like to add addresses to OSM for New Brunswick Canada, I have read the wiki but I am still unclear on the process. It looks like https://openaddresses.io/ has data for NB, and it also looks like there was some wiki discussion about using GeoNB - address points as a reference (Dead since 2019? https://wiki.openstreetmap.org/wiki/New_Brunswick).

Magic Earth might be using OpenAddresses data (or similar), either through OpenAddresses downloads or directly from a variety of governments.

Not exactly what you described, but it could be worth taking a look at if OpenAddresses would fit the task. I haven't done much work with residential mailing data myself, but I found The Markup's detailed explainer for how it used OpenAddresses and other data sources in its recent report on internet pricing to be highly informative.

Several years ago, I heard about a neat project called https://openaddresses.io and thought there should be an easier way to meld that information to objects that are in OSM. So I wrote some software to download information from both OA and OSM and create some changetset diffs that could be loaded into JOSM and reviewed and uploaded. My project lingered for a while and I finally got re-motivated to clean it up enough for public consumption.

Go to openaddresses and download the files covering your service areas.

Even better for those of us in the U. S. of A., I found https://openaddresses.io/, which has osm compatible maps encoded with addresses in a way that simple search is easy. It includes all addresses! "22 Burbank Street, Rochester NY". Will find the address or a nearby one if it does not exist.

I stated using gofakeit's "hackerphrase" for all commit messages.

https://github.com/andrewarrow/feedback/commits/main/

hp | git commit -a -F -

hp is a golang binary that just spits out a hacker phrase. I have this aliased with the letter q for "quick" so I'm always checking in stuff with q return push done.

"zip_locations_zip_index" UNIQUE, btree (zip)

The following takes hours and hours to run:

https://github.com/andrewarrow/feedback/blob/main/location/zip.go

But you end up with a nice way to get an approximate postgis geometry for any zip code in the USA. (For free.) Anyone know of a faster way?

My favorite part of this code is buffer := make([]byte, 1). By making the buffer just a single byte, I can look for "\n" so much easier than if I was reading in 1024 bytes or 2,048 bytes etc and then I have to find that \n in the middle.

I love this list. One thing I'll add is I like to avoid using S3 or google's cloud storage AT ALL until I have > 30 GB. i.e. with google free tier of their compute engine you get a 30GB hard drive, 1GB ram, and two AMD EPYC 7B12 2250 MHz processors. So I make a fake bucket system using those free 30GB:

https://github.com/andrewarrow/feedback/tree/main/filestorag...

All I code I write is in bucket form, ready to flip a switch and use real buckets but 100% free up to that hard drive limit.

I just used a mutex here:

https://github.com/andrewarrow/feedback/blob/master/stats/me...

Multiple go routines will call this method at same time but:

https://github.com/andrewarrow/feedback/blob/master/router/p...

Where I call it I start a new go routine so nothing is ever waiting on that mutex right?

So this question https://news.ycombinator.com/item?id=35198563 prompted me to revisit the framework demo'ed here https://news.ycombinator.com/item?id=35033368

Just like reddit lets you go to what might be a 404 page /r/anything-you-want and you can then create that subreddit, I added this feature to remoterenters so you can make /rr/anything-you-want and be the mod of that sub.

To get the ball rolling with some content I made these two AI based subs:

https://remoterenters.com/rr/amazing-ai-understanding/

https://remoterenters.com/rr/writing-prompts/

And my friend "not_a_t1000" bot helped me add some interesting content. My favorite one in "amazing-ai-understanding" is about the Steering Wheel! How can it know this subtle meaning! So impressed!

Feel free to create your own new subs or add to these. Everything is open source. You can follow along with the commits to the "go on rails" framework here:

https://github.com/andrewarrow/feedback/

and the sample app here:

https://github.com/andrewarrow/remoterenters/

if anyone wants to pay for the domain, I'll set up copies of this https://remoterenters.com/ for any topic you want. Trouble is hard to compete with:

https://www.reddit.com/r/worldnews

https://www.reddit.com/r/marketing

https://www.reddit.com/r/business

for traffic and users.

paste an example search with zero results in code. Like I did:

https://github.com/andrewarrow/feedback/search?q=view&type=c...

but

https://github.com/andrewarrow/feedback/search?q=views&type=...

does have results.

I'm not sure exactly what features to add next to this site https://remoterenters.com/ but the idea is to help people find a building where their rent is actually a good deal. Because yes, many people will never get a mortgage and buy a condo or house. And a16z has a plan to give renters equity in their building + make the experience of living there 10x better.