On Dow's solution of Bell's problem (Q5929011)

Murray Bell has asked if any compact 0-dimensional space with a closed \(G_\delta\) mapping onto \(\beta N\), must the space itself map onto \(\beta N\). This paper provides a particularly elegant and simple proof that this need not be so if the continuum hypothesis is assumed. The question remains open in ZFC. The mention of the reviewer in the title is that he had a particularly inelegant and complicated proof of the same result. The author's example is obtained as follows. Let \(C\) denote the Cantor set and let \(f: \beta(N\times C) \rightarrow \beta N\) be the natural map sending \(\{n\}\times C\) to \(\{n\}\). By CH and Parovichenko's theorem, there is a compactification \(\gamma N\) such that \(\gamma N\smallsetminus N\) is equal to \((N\times C)^* = \beta(N\times C)\smallsetminus (N\times C)\). Use the mapping \(f\) on \(\gamma N\smallsetminus N\) to define a quotient (or adjunct) space of \(\gamma N\), call it \(\delta N\), so that \(\delta N\smallsetminus N\) is homeomorphic to \(N^*\). However, because of the map \(f\), no infinite subset of \(N\) is \(C^*\)-embedded in \(\delta N\), hence \(\delta N\) cannot be mapped onto \(\beta N\), while \(\delta N\smallsetminus N\) is a \(G_\delta\) which does map onto \(\beta N\).