In all parts of Europe the assessment of research quality is one of the key drivers in the practice of higher education: it is employed to decide whom to appoint and whom to promote; it is employed to determine what knowledge gets validated by publication (or presentation at a conference); it is employed to determine the funding which will be made available either in response to specific research proposals or (as in the UK) in response to an assessment of the research quality of whole departments. If it is 'money that makes the world go round', in the academic community at least it is research assessment that makes the money go round. And yet in the very diverse practice of educational research, especially, since this draws on such diverse forms of enquiry from all parts of the academy, it is reamrkably difficult to develop approaches to research quality assessment which satisfy ordinary demands for transparency, validity, reliability, consistency and fairness. This paper will consider why this is the case and examine some different attempts to address the problem. The paper will first examine some of the most commonly offered criteria for research quality (employed in, among other contexts the EERA initiated European Education Research Quality Indicators project), including: originality, significance, rigour, integrity and the 'international'. It will demonstrate the elusiveness of these qualities, but suggest that shred meanings may be established through people working, talking and exchanging responses in conversational communities. It will then proceed to examine attempts to identify proxies for these qualities in the form of research quality indicators (such as citation, impact factors, downloads) and discuss the extent to which 'mechanical' reading processes might be able to identify such qualities. This will refer to some of the issues raised specifically in the EERQI project. (Hence the 'impossible science' of the title Then it will consider whether different approaches based in the Aristotelian tradition of practical judgement or even in the notion of connoisseurship. (Hence the 'possible art' of the title).

This paper will include reference to policy, national and HE systems and empirical evidence but will be couched mainly in the form of philosophical conceptual analysis and argumentation.

As a philosophical contribution this is more interested in the argument than in the conclusions, but the summary above indicates that it will challenge attempts to measure quality (and especially those which attempt to do so mechanically) and will look with at least begin to put forward alternative models.