Thijs Niks @thijs.niks.nu · 08/09/2026I have a new version running locally that uses the Nominatim Reverse Geocoding API, but haven’t had the time yet to release it. nominatim.org/api/Reverse/ The new version will also show the map, population density, and elevation. 020
Thijs Niks @thijs.niks.nu · 14/10/2025My own version of a Midwest network looks a bit different railmapper.net/project/midw... 100
Thijs Niks @thijs.niks.nu · 13/10/2025I quickly threw the ChatGPT Midwest proposal in railmapper.net and ridership is decent at 27,000 per rail-km (2.5x Spain). The economic viability stands or falls (as always) with the construction cost assumption. With ChatGPT’s assumed $21.5m per kilometer, you get a manageable -7% loss margin. 131
Thijs Niks @thijs.niks.nu · 29/09/2025India is looking good (and indeed why I had to fix the performance issues): railmapper.net/project/indi... 121
Thijs Niks @thijs.niks.nu · 02/09/2025You can read more about how Rail Mapper works in this document: docs.google.com/document/d/1... Feel free to leave comments! (I do recommend reading it in the mobile app or on the desktop, as images are embarrassingly low quality on mobile web) 130
Thijs Niks @thijs.niks.nu · 02/09/2025By default, the tool includes a dozen example networks to give an idea of how it works. Take a look at for example: France railmapper.net/project/fran... US northeast railmapper.net/project/nort... Indonesia railmapper.net/project/indo... India railmapper.net/project/indi... 141
Thijs Niks @thijs.niks.nu · 02/09/2025A few people, like @alonlevy.bsky.social, take a different approach and try to ground their high speed rail proposals in the reality of usage and costs. Like their plan for the United States: pedestrianobservations.com/2021/03/22/h... 150
Thijs Niks @thijs.niks.nu · 02/09/2025I love high speed trains and have used them in France, Germany, Italy, England, China, Japan, United States, and Indonesia! 160
Thijs Niks @thijs.niks.nu · 02/09/2025All distances in the Northwest are under 500km, so the additional distance between Everett/Tacoma and Portland isn’t penalized but Portland does benefit from the increased passenger-kilometers. Hence the higher predicted revenue. railmapper.net/project/nort... 010
Thijs Niks @thijs.niks.nu · 02/09/2025Alon has a travel time version (2h15m instead of 500km minimum threshold?) of the model. The tool currently doesn’t take that into account, but I’m going to see if I can implement it in version 2. docs.google.com/document/d/1... 110
Thijs Niks @thijs.niks.nu · 26/08/2025Made my programming slightly less terrible and brought matrix calculation down from 10 minutes to 58 milliseconds 010
Thijs Niks @thijs.niks.nu · 23/08/2025It does look like the PPP conversion of €30 is 200k Rp, which makes 125k seem reasonable? chrislross.com/PPPConverter/ 000
Thijs Niks @thijs.niks.nu · 23/08/2025Though the apparent current plan seems set on failing, with the first red line avoiding lots of large cities (Damanhour, Tanta, Banha, Mansoura, Zagazig) for seemingly no reason en.wikipedia.org/wiki/High-sp... 010
Thijs Niks @thijs.niks.nu · 22/08/2025Nice! I have added my version of an Egyptian network: - Almost 2 billion passengers - 1.22m passengers per rail-km - $24b construction - 30% profit margin Though I expect fares should be $0.025 instead of $0.03 per km, but the tool can’t do that. railmapper.net/project/egyp... 2100
Thijs Niks @thijs.niks.nu · 20/08/2025Though quickly built once approved brainporteindhoven.com/nl/nieuws/op... 110
Thijs Niks @thijs.niks.nu · 20/08/2025Local NIMBYs obviously went all the way to the Supreme Court to try to block dedicated cycling infrastructure, so a bunch of the plans were delayed www.raadvanstate.nl/@132662/2021... 140
Thijs Niks @thijs.niks.nu · 17/08/2025Though my model comes out lower than ChatGPT thinks HS2’s own estimates are (I haven’t checked that math yet) 100
Thijs Niks @thijs.niks.nu · 17/08/2025That’s a much larger ridership increase than the 2x previous travel time elasticity analysis would suggest pedestrianobservations.com/2022/06/05/i... 110
Thijs Niks @thijs.niks.nu · 17/08/2025My model estimates that increasing the speed (and roughly halving the travel time) would increase ridership by about 2.5-3.5x railmapper.net/project/hs2-... 110
Thijs Niks @thijs.niks.nu · 16/08/2025My ~200,000 constant comes from trying to match your original NEC example. I basically took the same model, changed the populations and then tried to match the overall ridership by adjusting the constant. 110
Thijs Niks @thijs.niks.nu · 16/08/2025If you use the cities menu (button at the bottom of the screen on desktop), you can add coordinates while in manual mode 110
Thijs Niks @thijs.niks.nu · 14/08/2025Do you happen to have an idea why Siemens thinks they can build 2000km of rail in Egypt for just $8.7 billion? That seems very cheap. www.dw.com/en/egypt-sig... 100
Thijs Niks @thijs.niks.nu · 14/08/2025I added a high speed rail example network for Nigeria, which is a beast! 1.35 billion annual passengers, 4000 km network, 340k passengers per rail-km, +9.9% profit margin (despite 16% bond rate) railmapper.net/project/nige... 231
Thijs Niks @thijs.niks.nu · 14/08/2025Why would anyone want to stop at Ashford? Because they can’t get slots at St Pancras? 310
Thijs Niks @thijs.niks.nu · 13/08/2025One of the really interesting things to me is how adding a direct São Paulo to Belo Horizonte line is a money loser and increases the network loss margin from -35% to -53% 100
Thijs Niks @thijs.niks.nu · 13/08/2025Using $0.12 per km fares results in a much more reasonable -35% loss margin (compared to -100% before) 200
Thijs Niks @thijs.niks.nu · 13/08/2025The config changes shouldn’t be problematic, though changing the population radius (which comes with a confirmation dialog) can be slow and tricky for larger networks 010
Thijs Niks @thijs.niks.nu · 13/08/2025I used Google Flights, one month out. But doing that again just now gives me a higher fare (52/436 = 0.12) than yesterday (35/436 = 0.08) 100
Thijs Niks @thijs.niks.nu · 13/08/2025My spreadsheet version of the Japanese network does use a -0.5 price elasticity 120
Thijs Niks @thijs.niks.nu · 13/08/2025Correct, price elasticity has not been included yet docs.google.com/document/d/1... 130
Thijs Niks @thijs.niks.nu · 13/08/2025These are the networks I have estimated before in spreadsheets (with the obvious missing one being China) 110
Thijs Niks @thijs.niks.nu · 13/08/2025Yup! Though they would still be fine at 2x the French cost even. Just not 10x. docs.google.com/document/d/1... 030
Thijs Niks @thijs.niks.nu · 13/08/2025It currently doesn’t, but I hope to add price elasticity (-0.59) in the future. My spreadsheet does already support it. docs.google.com/document/d/1... 030
Thijs Niks @thijs.niks.nu · 11/08/2025It uses the GHSL population data we discussed before, so you can model networks anywhere in the world and get consistent results. x.com/nerd4cities/... 100
Thijs Niks @thijs.niks.nu · 11/08/2025Just noticed the 335m passengers. That’s a lot! Nice. And 100k passengers per rail-km is definitely profitable territory if construction costs are not outrageous 200
Thijs Niks @thijs.niks.nu · 10/08/2025Thanks! Raising it from 0.8 to 1.0 does increase the Northeast Corridor ridership estimate from 116m to 210m per year. It also increases the French estimate from 120m to 158m. Neither is outrageous, but using 0.8 does seem to produce results closer to your own. 110
Thijs Niks @thijs.niks.nu · 10/08/2025Yeah, that makes sense. The tool estimates what the cost would be if they were to build a fast network from scratch today: railmapper.net/project/germ... Though unclear if they could do that for €50m per kilometer (which is what I currently use as the average cost): railmapper.net/project/germ... 110
