There are two central, important and intractable logical difficulties that seem to face debate in the area of social policy. First, people can disagree about the important features of a social situation, both in description and evaluation. This is hardly surprising, since it is the framework wherein one structures phenomena that tells one which are significant, and no framework is logically compelling. The other way of looking at it is to argue that any explanatory (or evaluative) account (so long as it is organized and is not circular) must begin somewhere. What if we differ on some basic premise? These are not derived from any further premise (otherwise explanation would go on for ever) and cannot be empirical (since they are used to order data). At this stage argument reaches an impasse in rational discourse, both in analysis and evaluation.