Killing normality with a Cohen real (Q1924656)

Many consistence results in topology require information about which topological properties of a space are preserved after a specific forcing extension. For example, the solution to the normal Moore space conjecture can be viewed in this light: if a space is not collectionwise normal and ``sufficiently many'' Cohen or random reals are added, then the space is not normal. So it was known that if one adds enough Cohen reals, normality need not be preserved. Watson asked if there are spaces whose normality is not preserved when just one Cohen real is added. Grunberg, Junqueira, and Tall established that such a space would have to be a Dowker space (a trigger word if ever there was one). The underlying reason is roughly that the unseparated closed sets in the extension can be written as a countable union of ground model closed sets. This paper provides the missing converse implication, interestingly, that every Dowker space canonically gives rise to a space whose normality is not preserved. There seem to be two key steps. The first is to start with the underlying reason from Dowker that if \(X\) is a Dowker space, then \(X\times (\omega+1)\) is not normal. Namely, that if \(\{D_n\}_n\) is the decreasing sequence of closed sets witnessing that \(X\) is not countably paracompact, then \(X\times \{\omega\}\) cannot be separated from \(D_f= \bigcup_n (D_n\times [0,f(n)])\) for each unbounded \(f\in \omega^\omega\). The second idea is, naively, to strengthen the topology of \(X\times (\omega+1)\) just enough to have that \(X\times \{\omega\}\) is separated from \(D_f\) for each \(f\), but in the generic extension, \(X\times \{\omega\}\) will still not be separated from \(D_g\) where \(g\) is the new Cohen real.