That would mean that a journey around a Circle Line[0], arriving where you departed, could have potentially infinite length. In fact, if you don't rule out travelling along the same stretch of rail twice, then any return journey could be of unbounded length despite not covering any net distance. Calculating the "longest train journey" then becomes a question of human endurance/longevity/wealth rather than about geography or transport.

Alternatively, if you measure "the distance travelled along the rail, without double counting" then the longest journey is probably equal to the total length of the installed rail network in Eurasia, which is an interesting statistic and a challenging optimisation problem, but hard to consider as a single journey.

[0] https://en.wikipedia.org/wiki/Circle_Line

[1] https://en.wikipedia.org/wiki/List_of_countries_by_rail_tran...

What if the longest distance meant "longest shortest rail distance between two points".

Ie, the longest track distance between any two places on Earth, subject to the constraint that there is no shorter train journey between those two points.

This seems to me to be a much more logical, and also more interesting, interpretation of the question than the one used in the article.

Wouldn't that simply be the "shortest rail distance between two points"?

Not exactly. Parent is describing the graph diameter [1].

[1]- https://mathworld.wolfram.com/GraphDiameter.html

Well, no. The shortest rail distance between two points is very very short, because you can choose the two points very close to one another.

Oh, right. Fair enough. The two points with the longest shortest-path, with no fixed choice of points.

But then you can just zigzag through Europe. That's no useful measure. You at least need a further limitation.