rustywasm

Just added LA County. I spot checked a few places I know, seems to be pretty accurate.

I added LA County data, and checked it against where I group up and where my mother and father grew up. Seems reasonably accurate so far.

Yep, that was my misconfiguration of the CloudFlare cache. Hopefully fixed now.

My mom (who is 97) has a vivid memory of seeing FDR in 1935 -- the whole extended family took the the Red Car from Compton to downtown to see his motorcade. She remembered the electric trams worked really well, especially for folks without much money.

Thank you! Great link!

And here is one that is just fun to watch because of Huell Howser --

Tallest Building | Visiting with Huell Howser https://www.youtube.com/watch?v=5ziEDhCPjOQ

I think the best intro to LA is Cadillac Desert by Marc Reisner (it's really about the entire West and its relationship to water, but I think the first few chapters explain LA better than many other books.)

Yes, great point. What was torn down, burned, destroyed, etc. along the way isn't here. This is the surviving city, not the city as it was.

Exactly this -- This is more like the construction date of buildings that have survived. Everything demolished, burned, or replaced is invisible.

Probably should put that front and center.

The waves are partly real and partly me.

LA genuinely built in bursts. The 1920s and the postwar decade each account for over 230,000 buildings still standing, and the 1930s and 1970s onward are much thinner, so even a perfect per-year animation would surge and pause.

But really it's for performance -- I group buildings into 5-year bins and fade each bin in over its span, rather than popping each building at its exact year.

A per-year filter on a million polygons re-tessellates every visible tile, which was too slow to animate, whereas fading pre-built layers is reasonably fast.

Glad you found your house!

Did not think of that - yes that is a really interesting idea. Totally different way with interacting with data. I'm going to ponder that...

Yes, this is incredibly similar to the pipeline I ended up with -- https://github.com/kfarr/zoningviz/blob/main/HOW_IT_WORKS.md

The other thing I learned is that some of the data sets use a destructive insert, so unless someone is doing a data snapshot data elements can be lost. (The vendors charge extra time-series option.)

Yes, good point. The LA City data was rich, but the other communities have data in other places. I will see what I can do.

One the experiences for me that never gets old is coming in to LAX (flight path East to West) and the sprawl just goes on forever. The other crazy thing that 60% of the population speak a language other than English at home. So I think it has one of the best options for cuisine in the world, because it's all there (somewhere!)

This started as a data puzzle. There is a lot of public data but it's hosted by for-profit companies and the data models don't seem to be coherent (probably lots of good reasons for the data models being the way they are, so no criticism, and it's wonderful they are public.)

So for LA City, they use Socrata (owned by Tyler Technologies -- NYSE: TYL) to host and publish open data. LA City also uses Esri's ArcGIS for geo-based data.

So behind the scenes there is my code that builds a parquet database of * MyLA311 service requests * Permits * Construction Enforcement * Crime * Parcels Id with Polygons * Lidar

So that part was a typical data-munging exercise, and it was fun & interesting to me to learn how public entities try and deal with IT budget cuts and service the community, etc. (For example, 311 requests for homeless issues are transferred to another department, so they all look they are closed immediately, but I never could find the agency data that ultimately resolved that type of request.)

So I have a pile of data that is organized in a way that makes sense to me, but I could not build an interface that I liked.

So I asked Claude to create a fly-over visualization to see what it could do with this data, and it blew my mind.

I'm a life-long coder and I'm old, so it's an interesting experience to see how good these models and coding harnesses are, and how quickly they are improving.

As to what makes this HN worthy? I enjoyed playing with the interface, and there might be more people than just me who love LA (warts and all) and its history and architecture. So it's interesting and I'm not asking for anything in return, so for me that satisfies the "HN Show" criteria.

The deeper question for me right now is what does hacking mean when the coding is done by these systems? It's a different kind of hacking, and its makes things possible that impossible before, but it's a million miles away from 6502 assembly.

Good idea! Let me work on that...

I joined the county's 2020 lidar building footprints (LARIAC) to the assessor roll, so every one of the 1,129,558 buildings inside the City of LA has a footprint, a height, and a year built for 98% of them. (All public info, but kind of a pain to grab.)

The page extrudes them in 3D and plays the city filling in from 1880 to now. Colors are by decade, or switch to height and fly to downtown. Hover any building for its address, year, height, and footprint. Some things I found: the 1950s, not the 1920s, is the biggest decade still standing; the Valley visibly snaps into existence after the war; and downtown's 150-foot height limit shows up as a hard ceiling until 1957. Zoomed out it draws a 100 m grid because a citywide view can't carry a million polygons, and heights are exaggerated below street level zoom (the corner label says by how much).

Stack: DuckDB spatial for the join, tippecanoe to PMTiles, MapLibre fill-extrusion, all static files on R2 with no server. City of LA only, since the parcel polygons I pulled stop at the city line.

Happy to answer questions about the data or the rendering.

[Just a fun project, no commercial relevance.]