What Claude does I assume is that instead of conducting a proper query on OSM’s data, it just does an OCR evaluation based on the rendered tiles, and tries to replicate that in a rough shape.
1 Like
I just noticed there’s a few more Wales routes missing! Here’s a link to the Transport for Wales network map.
1 Like
Using Wikidata for route is definitely the right way, but you need to be careful because different routes can use one same Wikidata ID.
1 Like
One small one: the card for Limerick to Galway says it is the only InterCity route in Ireland that doesn’t go through Dublin. That is contradicted by the Limerick to Waterford route on the site, and also by Cork to Tralee (not included but with a few direct InterCity services daily).
1 Like
There have been no Avlo trains on the Madrid to Barcelona route since last September. As far as I know that’s a permanent change in Renfe’s strategy for that line.
(I’m not convinced Avlo merited a separate route to start with - you could end up heading down a bottomless pit of slightly different services on the same routes…)
1 Like
I am working on creating a system route verification system, especially in Europe where there are more public data available. I want to use ÖBB Scotty for route verification, so that I can feel a bit safer that it actually runs through the selected route whenever I want to travel by rail.
I think when it says only Facts cross-checked against Wikipedia, it is perhaps not traced yet properly (The old AI hand drawing).
I want to at least have OSM or Operators trace + Wiki. Preferably it should be OSN/Operators trace + Wiki + Operators data such as ÖBB Scotty (They cover most of Europe accurately)
ÖBB for those who wonder what ÖBB Scotty is. Somehow the route data is accurate even for local trains in Norway…
LoL
Here a human being, writing with a physical keyboard 
.
does that for France but well beyond that. A one person project base on publicly available source, mostly OSM. Feel free to switch to another language (clicking on the flag top right).
I’ve no idea why in Brittany (western part of France) you display Paris-Brest as high-seed train but not Paris-Quimper. Both are very symmetric with high-speed trains and ordinary trains between Rennes and Brest or Quimper. In fact most double trains stop at Rennes, splitting to or joining from Brest and Quimper.
Yesterday when I checked I could see the delay of the TGV from Paris to Brest and the reason for that (a lost luggage).
1 Like
That website is very impressive and it feels like a video game which I like!
I would love have that for the whole Europe. I think it is strange that the public information for common European rail data is not very visible or accessible. I who just want to plan rail travel has to jungle with ÖBB Scotty or google maps which is not ideal.
I personally do not know to write a single code line. I read a book about chess and AI from Garry Kasparov: Deep Thinking: Where Machine Intelligence Ends and Human Creativity Begins., which gave me a good idea of how to use AI. AI is extremely good at data collection through context, while it is just stupid and it will be forever stupid. From an another website which I currently work on, I learned that feedback is the way to go for this kind of project.
For France specific I am not sure why. I am going to get that verified because I would like to know that!
I have got to thank you since you have been helping me a lot with Spain and Ireland. It is very fun to fill and repair the map, almost like playing video games again. It is very rewarding for me.
I am not sure. I would also like to know that as well. I am going to ask Claude that so that I can also know and see if I have to change something.
What I can say is that almost all hand drawing (based on geometry according to Claude) has disappeared. Now 98% of the routes should be either based of OSM or Operators data.
As he mentioned in https://github.com/nicolaswurtz/pan-gtfs-explore/blob/main/README.md:
Outil de visualisation et d’exploration des données de transport ouvertes du gouvernement français, issues de transport.data.gouv.fr.
Transparence : la quasi-totalité du code de ce projet a été écrite par GitHub Copilot (agent IA). L’humain a défini les specs, orienté les choix et validé le résultat.
He mainly used copilot to convert GTFS data coming from OD. You may ask Claude mimic that for other countries so that he could use it.
The license WTFPL is OK to create a devirative (or another project) based on that.
For other extensions, GitHub - nicolaswurtz/extras-opendata-sncf-reseau: Jeux de données supplémentaires du réseau ferroviaire national français, obtenus par transformation ou extrapolation. · GitHub could be used as base to extend to Europe for instance. The license is ODbL, fine too.
I am working on making the verification system better. I am trying to avoid the trap where Claude says that it is traced by OSM when there are too many mistakes and no links to the exactly route. I hope to be able to fix that. Besides I am going to add OSM link on the hover card so that it will be easier for users to fact check the route.
You seem to be sending Barcelona - Malaga down the conventional track outside Barcelona, resulting in two improbably sharp turns on the route. The card says “checked against the running timetable via ÖBB Scotty: our line sits 0.2 km from the shape of IRY 6020”, but Scotty displays the correct route in this area.
Moving further afield, what is your source for Sarajevo to Banja Luka? I rode this route a couple of times, but it was a long time ago, and I don’t think it is possible to travel on the full route these days. My understanding is that ŽFBH runs from Sarajevo to Maglaj, and ŽRS from Doboj to Banja Luka, with a gap in the middle. I’d love to be wrong, but…
1 Like
Just noticed something even weirder on Madrid to La Coruña. Again the card claims this has been verified.
By the way, coming back to something from earlier, it’s not strictly true that Malaga to Granada runs on already mapped tracks. The highlighted section here was built specifically to make this route possible - until recently there had never been a direct route between those two cities. But I understand if you feel this is too short to overcome the “don’t reuse already mapped lines” rule.
1 Like
I have noticed this Barcelona - Malaga line as well. It looks weird although the verification system says that it is correct. I think when I manage to link this route with OSM correctly it should display the correct section. Thanks for commenting!
For Madrid - A Coruña I managed to get this route OSM linked now, but the map is still showing the route wrongly after I checked on OSM rail graph.
I am glad that you commented on this because I was a bit unsure honestly myself. I hope to fix that and for all routes. I am working on it.
Balkan routes are very tricky to deal with. I am going to check Bosnia again.
Maybe also check where the photos on thr Bosnia routes are coming from. The main card photos are OK but the secondary galleries look like random photos of trains in and near BiH unrelated to the specific routes e.g. Ploče (Croatia) appears in both but is not part of the mapped routes.
1 Like
Thanks for letting me know! I have also noticed Nice - Ventimiglia route has wrong pictures in the secondary gallery. There is probably a bug which have not been noticed yet.
I saw that there is a route from Ploče to Sarajevo in Bosnia. Perhaps there is a route from Sarajevo to Maglaj as well (Not 100% sure though). I hope to learn more about that soon.
Some trains from Sarajevo to Mostar go all the way to Ploče, but they might be irregular or summer only. I would be more confident about Maglaj as a regular route.
This timetable looks plausible for ŽFBH - there are lots of footnotes about Ploče trains which I have not looked at in detail.
1 Like
I used the same site and I also looked on Google maps to see if they had listed the departures on the stations which they did (Not 100% accurate but they cover almost anywhere in the world)
I assume Ploče is summer service. There are several trains going to Croatia from farther north in Europe in summers, while in winters the trains go to the Alps. I think that is pretty interesting how the routes changes based on seasons.
Rest assure, there are probably many mistakes on the map which I am working on. At least they like that they do not belong there. Soon I am considering getting the whip for Claude so that it can learn a lesson;)
