This apparently ordinary station on the Berlin U-Bahn is remarkable for two reasons, one visible and one not visible at all.

The first is fairly apparent. The typeface of the station name and the brown tiles give a clue, though the full splendour of the orange ceiling doesn’t really come across in the picture. This is the 1970s in unabashed glory.

The second reason is not apparent at all. It is that as well as being an underground station, this is also a nuclear shelter. In the concourse above and in the tunnels, massive doors are ready to swing into place. Just by the ticket machines, a small discreet door in the wall leads to interlocking steel doors, a decontamination area, and then a warren of rooms and passages providing everything necessary to keep three and a half thousand people safe for two weeks, before emerging into the post-nuclear dawn. Trains were to be parked at the platforms to provide extra space, with bunk beds four high on the platform itself.

It is extraordinary, impressive – and a complete folly. It formed part of a civil defence network  for Berlin which in total provided 20,000 places for two million inhabitants. It was a gesture, not a defence system.

From our omniscient viewpoint 35 years on, it is easy to ask why on earth anybody thought this was a good or necessary idea. It’s fairly clear that it was a gesture, but a gesture to whom, signifying what?

Of course it didn’t look like that at the time. It never does.

This is a particularly stark example, partly because of the speed with which the perceived threat of the cold war dissipated. But it is very far from being unique. Options may be appraised, costs and benefits assessed, future proofing added and gold plating removed, but still solutions emerge from the paradigm within which they were created. Perhaps it was once obvious that the building of a new station should be seen as an opportunity to build a shelter as well.

None of us will make exactly that mistake, of course. But we are no more immune to follies of policy and delivery than our predecessors were. The difference is only that ours are not yet visible to us: it will be apparent in a few decades – perhaps sooner – where we are going wrong.  That’s not much good of course: if we are going to fail, much better to fail early, in time to do something else instead.

If the only problem with the Pankstrasse shelter were that we can now marvel in its pointlessness, there would be little point in worrying about it. There is a strong case for saying, though, that this is a folly which could have been avoided; that the problem was not the absence of information but its interpretation. Knowing (or thinking) that provides no guarantee that we can do better, but it does suggest that innovation by simple extrapolation may not be the best way of designing policies for the long term.

They still test the doors which seal the tunnels once a year. And the ceiling is still orange.