Proposing to deprecate railway=razed and railway=dismantled

General talk Tagging general discussion
, ,
354

I don’t know about instead of OSM, but any real railroad track that deserves to be mapped in OSM also deserves to be mapped in OpenHistoricalMap, without question. OHM wouldn’t be a real historical map if we were to decline extant features: there would be surprising gaps when looking at the past, just because something happens to remain up to the present day. OHM also doesn’t care that the Academy considers many things to be mundane and historically insignificant; otherwise, if we did, the famous features would be oases in a terra incognita. That said, OHM is uncomfortable with features that lack dates and sources, because anyone can just make those up, and there is a preference for public domain data when possible.

OHM currently has 258,297 miles (415 689 km) of railroad tracks, roughly equal to what OSM had in January 2008 and only 14% of what OSM has today. Of that total distance in OHM, 69% or 176,953¼ miles (284 778 km) is missing end_date=* tags, indicating that the tracks remain intact up to the present day. But a significant share of the mapped tracks predate the Beeching Axe, peaking in the 1930s.

A line graph of total railroad trackage by decade from the 1800s to the 2020s. Coverage plateaus between the 1810s and 1840s before raising steadily, peaking in the 1930s and declining slightly before the dramatic declines of the 1960s and a stabilization in the 1990s.
A line graph of total railroad trackage by decade from the 1800s to the 2020s. Coverage plateaus between the 1810s and 1840s before raising steadily, peaking in the 1930s and declining slightly before the dramatic declines of the 1960s and a stabilization in the 1990s.1372×1014 70.9 KB

Although the rise and decline of trackage coverage roughly mirrors the real-world historical progression, there’s still plenty of room for expansion (including horse-drawn wagonways, which might be an interesting tagging discussion). A look at geographic coverage in QLever also reveals very large areas for improvement:

A heatmap of railroad track coverage throughout the world, showing hotspots in Colorado and the Midwestern United States, France, Central Europe, and Scandinavia, South Korea, New South Wales, and especially on Null Island (due to a bug in QLever).
A heatmap of railroad track coverage throughout the world, showing hotspots in Colorado and the Midwestern United States, France, Central Europe, and Scandinavia, South Korea, New South Wales, and especially on Null Island (due to a bug in QLever).1920×1073 80.6 KB

These statistics are extremely crude, because just like OSM in 2008, we haven’t gotten around to double-tracking and quad-tracking many railroad corridors yet. Regardless, the raw numbers obscure the real value of the OHM database: rich metadata about events, images, and sources can accompany any feature, and chronology relations can draw attention to non-spatial relationships between features. With this extra context, users can come to understand railway infrastructure a little deeper than by viewing a map that only gestures to the past.

I always feel ambivalent about discussions like this. On the one hand, I appreciate the opportunity to shamelessly plug OHM. OHM needs help from the OSM community to accomplish its goal of telling geographic history. But on the other hand, it saddens me to see all this energy spent rehashing an old debate, debating or defending marginal features in OSM. We could be spending all this energy more meaningfully by mapping things that still need to be mapped, in either project.

8 Likes