This paper is a report of my efforts to find out what Proto-Austronesian (PAn) might have sounded like, both in terms of its phoneme system and its stress patterns. This may seem a somewhat fanciful venture, but I believe that the phonology of a protolanguage must be reconstructed as the phonology of a real language if we are to properly tackle the task of subgrouping its descendants. This is especially true of a protolanguage which, like PAn, sits at the apex of a genetic tree. I have called this "an outsider's view" because I have worked with speakers of none of the languages which are central to this paper, yet my interest in Oceanic languages and Proto-Oceanic has ensured an interest in high-order protolanguages, and especially the language atop the tree.