tppiotrowski

Happy to field any questions about the project.

Related project I did in 2014 tried to do this. I was a web developer so used the web audio APIs to trigger a recording when the decibel level exceeded a certain value. I was living in a big tent in my friends back yard in Sydney at the time and was convinced it was airplanes coming into SYD that were waking me up at 4am but never really captured conclusive evidence because my laptop battery couldn't make it through the night :)

Maybe we need a journal completely dedicated to replication studies? It would attract a lot of attention I think.

This is a huge leap from the previous global dataset, which contained about 1.7 billion buildings

I use the monthly release of Overture Maps building dataset and the last one had around 2.3 billion buildings iirc

[1] https://docs.overturemaps.org/blog/2025/11/19/release-notes/

I remember thinking there's a small arbitrage opportunity in countries that don't have pennies and nickels. In NZ I believe stores round to the nearest decimal (.02, .01 => .00 and .03, .04 => .05). They lose on some sales but gain on others. However, they don't round if you use a CC.

Here's one for the FIRE folks: if it rounds up, use a CC and if it rounds down use cash. Use all those pennies you save using your CC to retire 3 minutes early.

I think what Claude generated code is still missing is the feeling of "I learned something from reading this code. Reading this code made me a better programmer". We're not there yet. That's why we have forgiveness for the scripts/tests/etc Claude writes - those are purely for utility. However, reading Claude code must make us feel like we're getting smarter, not dumber.

My experience using Claude code is that it's excellent at React/React Native and because apps are mostly a View layer the consequences of bad "vibes" are less far reaching.

Protomaps does this already. They do a planet build each day and offer a download for free:

https://docs.protomaps.com/basemaps/downloads

I think mbtiles are being phased out for pmtiles because no DB required and can be served from static storage like R2/S3 (with a worker but hopefully the worker part goes away and they support byte offset requests soon)

Fair. I used Phoenix because it's the only published city shading plan I've come across. It could just be that arid/hot climates are the places trees are needed. The study linked was in Houston, Texas.

The Phoenix report is valuable because it provides lessons that should be avoided going forward: change the laws so property owners are not liable if a tree outside their business hurts someone, don't plant a tree if you can't irrigate it, work with local residents to plant and water trees to save on labor and increase success, etc.

If there's other municipal shade reports I'd love to read them. Helping people find shade is what I do for a living. [1]

[1] shademap.app

Planting trees is hard. Phoenix AZ has a tree planting program called the Shade Phoenix Plan and they publish public reports.

Watering and maintenance are a big cost. Iirc it's about $1000 to plant a tree. $100 for the tree and $900 for the irrigation and labor to plant it. In the first 10 years of the program 2/3 of the 106,000 planted trees were removed due to accidents, storms, not enough water/aging. [1]

[1] https://www.phoenix.gov/content/dam/phoenix/heatsite/documen...

There are a few other reasons that aren't baffling:

roots destroy water pipes and heave sidewalks and pavement. They can interfere with power lines as well. :(

Counterexample:

https://x.com/levelsio/status/1730659933232730443

The YC crowd seems to think the only worthwhile businesses are worth $1 billion+ otherwise why bother

He was told it would take $10,000 to fix, and the representative also let slip that all Tesla repairs are covered by warranty for A YEAR - meaning he could have got his suspension fixed for free in January 2024

They bought it December 9, 2022. What am I missing?

Edit: or was it fixed on his last visit of June 2023 and then was eligible for replacement?

I agree with this. Any law that's not universally enforced: speeding, jaywalking, tax audit, etc is a tool for political persecution.

Sorry about the burnout. Sounds like you've got skills and I'd encourage you to explore something smaller. There is a path as a solopreneur. I do sun and shadow modeling using publicly available datasets [1]. My customers are gardeners, permaculture, hunters, fishermen, photographers and also real estate prospectors but they're people not big orgs or banks. It feels good to work on this level and personally answer emails and questions. I don't make much revenue but I like the grassroots path. Maybe you'd find it rewarding as well.

[1] shademap.app

Why does this mostly affect the US? I've been abroad most of the year and eggs don't seem overly expensive.

Another one: No "Sign in with your Google account" popup covering the upper right corner of the page

You can train ML to recognize vegetation in satellite photos and further train it to estimate its height +/-3 meters

https://github.com/facebookresearch/HighResCanopyHeight

The fact that it's self reported makes me wonder how close to the actual number it is. Also, how did they choose their sample.

Edit: Link to original paper https://d1wqtxts1xzle7.cloudfront.net/32982174/Turner_-_Appe...

True. Argentina doesn't have DST and I just realized living in the northern hemisphere made me miss the fact that the sunlight hours chart is concave in the south

https://shademap.app/@-50.35203,-70.98027,4.13034z,173600823...

I generate these type of charts [1] focused on the daylight hours so it was a surprise to see a concave shape instead of a convex one. Awesome way to validate these computer generated charts with captured physical data.

[1] https://shademap.app/@52.39941,4.88468,11.49849z,17360064872...

Yes. There are planet wide DEM (Digital Elevation Model) datasets which record ground level elevations but no planet wide set for DSM which includes built structures and vegetation.

As someone who has researched DSM availability across the globe, Google's Solar API is a top contender. Other option is government LiDAR surveys but the coverage, file formats, projections, etc are all fragmented. I think it would be great for the mapping community to create a world wide DSM map tile dataset similar to the ground elevation tile dataset that contour lines and 3D terrain views are generated from. Maybe someone is already working on this?

In the article they show areas where their approach can generate DSM although this is just the potential areas and not the areas where data is already available. :(

Sounds incorrect. I think the insurance rate is based on the value of the vehicle and SUVs and trucks are much more expensive than compacts and sedans.

Lighting conditions seem to check out:

https://shademap.app/@46.05225,8.19394,11.38803z,17341770702...

In “The Gospel of Relaxation,” James applies psychological insights to everyday concerns. Limit introspection: don’t become a prisoner of morbid feelings. To feel brave, act brave. To become cheerful, smile and laugh.

Today known as "fake it till you make it"

Or my personal version: "Action creates motivation, not the other way around"

I'm currently in Pokara, Nepal and Google maps has definitely hallucinated a lot of streets that don't exist (computer vision false positives I presume) This makes Google useless for routing because it sends you into dead end alleys and driveways it thinks cross through to the next road.

Organic Maps (OSM client) on the other hand is much more accurate and supports offline routing.

This mechanism is the source of fear for foreigners who believe they will be the victims of a mass shooting if they visit the USA.