I've further improved my global traffic / congestion / road quality formula. Times start from the OSM speed limits, then every road gets scaled by three things:

- traffic, guessed from population density: 1 + 1.09·d/(d+5000), where d is people per km². Basically nothing in the countryside, about 2x in a dense city centre

- bendiness, on fast roads only: 1.45 + 3.16·B/(B+406), where B is degrees of turning per km. Turns out nobody does 60 on a twisty B-road

- junctions: 0.64·(1 + 0.84·J/(J+5.35)), where J is side roads per km

I checked it against Mapbox's typical Tuesday lunchtime traffic on ~2,500 routes in 33 countries. It's usually within about 10%. Still a bit optimistic around London though.

Even though the stereotype of the Spanish Inquisition being religious fanaticals, they did not prosecute witchcraft because they correctly considered it mere superstition.

Data is the NOAA Storm Prediction Center tornado database, 1950 to 2025, lower 48 only. Each dot is where a tornado touched down. I dropped every year onto the same Jan to Dec calendar (Feb 29 counts as Mar 1), so each date shows 76 years of tornadoes at once.

The white dot is the average position of every tornado within a week either side of that date. Around Christmas it sits in central Mississippi, by late July it's up in northern Missouri. That's about 530 miles.

Few things I didn't expect:

- May 24 is the busiest date of the year, 702 tornadoes over the 76 years

- April, May and June are 54% of all tornadoes

- in Dec to Feb about 4 in 5 hit south of the Kentucky/Tennessee line

- in July half of them are north of 41°N. In January it's 2%

Disclaimer: I also run a free map site with live quakes, volcanoes and hurricanes if you like this kind of thing: https://www.liveatlas.earth/north-america/united-states-of-america/

https://www.thedailystar.net/supplements/international-day-the-girl-child-2026/news/bangladeshs-child-marriage-paradox-4295326

Very confused

https://brilliantmaps.com/spread-of-islam-map/

I redid the Mexican ancestry map that gets shared online that uses some old study. I did this one the using DNA-based estimates from the Mexican Biobank study published in Nature in 2023. The study genotyped 6,057 people from 898 locations across all 32 states.

The ancestry percentages were inferred from genetic data, not racial or ethnic self-identification. I reviewed other genetic studies for context, but all the plotted state values come from the same study. I made the map and charts in Python.

One important limitation: these describe the study sample, not population-weighted averages for each state, so rural localities and Indigenous-language speakers were deliberately oversampled.

Here the source: Sohail et al. (2023), Nature, Supplementary Table S3.

The 1921 election was the first election Canadians saw since the end of the Great War, the ruling Tory party was reduced to third place with the new Progressive party taking the official opposition in their first ever election. If you have any questions feel free to ask and I’d be more then happy to answer.

The 1962 election saw the Progressive Conservative government of John Diefanbaker reduced from the largest majority government in Canadian history to a weak minority government which would fall not even a year later. If you have any questions feel free to ask, I would be more than happy to explain.

THIS IS MY PERSONAL OPINION

Edit: i forgot to color my country honduras into something else.