Here are some peculiar things, which are from the background to this project.

(1) Open question arguments:

Perhaps the best known is G. E. Moore's 'natualistic fallacy', which goes something like this:

If we try to define 'good' in some non-normative, or 'naturalistic' way - as, say, 'that which maximises utility' or ' that which pleases God', we have to deal with the 'open question' whether it is good to do things which match the definition. Is it good to maximise utility? Is it good to please God? The fact that these questions make sense mean that the proposed definition cannot be complete.

This barrier to defining good in non-normative way cannot be a demonstration that the concept of 'goodness' is somehow incoherent, or that it should be discarded. If we were to reach this conclusion, we would, after all, want to say that it was the right conclusion, and that therefore it was good to believe it.

The open question arguments seem to unite philosophy. As well as Moore's, there are three more - addressing themselves to truth, knowledge, and meaning.

We can always ask of a theory of truth (a theory which establishes general criteria for truth telling) whether it is true - and since it cannot be tested against its own criteria without circularity we must conclude that no general theory of truth can be constructed. But we would also want to say that this was true, so the category - however undefinable - is not redundant.

Similarly, we might ask how we reliably know any theory of reliable knowledge attribution, while being unable to dispense with the category of 'knowing'.

And we might ask what our theory of meaning (which can be used to work out what things mean) means, while being unable to dispose of 'meaning' (since, presumably, we'd want to say our questions about meaning meant something).

(2) An empirical certainty?

'We can talk' is as much a statement about the world (an empirical statement) as 'we can walk'. Yet it must be true in any language game - it must be true, in a sense, 'a priori'. This is an example of a necessary truth which is not 'analytically' true - i.e. is not a logical tautology, or a definition.

'We cannot talk' is not false because it is internally inconsistent (as an analytical falsehood would be), but because it is inconsistent with the conditions of its being a statement at all.