i hope.
on Derrida's Cartesian madness:
that madness haunts and bears speech as its silent possibility seems similar to Levinas' signification itself as approach and proximity in that as the non-indifferent but undifferentiated moment before metonymy and which insists only in speech itself (or differentiation -- if metonymy is the moment after which carries within it the trace of contiguity in a more or less explicit way) it is itself impropriety and the madness of an eccentricity of the exposed 'self' which is called out by the approach of the other who cannot yet be distinguished.
the madness -- and death -- in signification and proximity consists in two things: the inutterability of signification itself and thus its priority to linguistic determination which situates it in the proximal relation itself in which the same cannot be differentiated from the other but in which alterity nonetheless approaches as alterity and non-indifference. the same is mad precisely because it is not itself but radically exposed to the other and yet called upon to state it'self' which can't be stated because (a) 'expression' is the silent halo of saying which escapes the sense of the said and (b) this 'self' which is expressed is not a self but an eccentric exposure to another prior to the determination of 'self' and 'other.'
thus the Cogito, or Cartesian 'hyperbole,' is called out prior to madness and reason as already mad or eccentric in that it cannot say itself. this would appear to make Derrida's Cogito the same as Levinas' Same insofar as both of them are conceived as proceeding from an impossible determination between self and other which is precisely language's prescription, or signification itself. i guess what i am implying is that the differentiation between self and other is the ur-metonymy of the moment before language while at the same time being strictly impossible actually to realize, both in the sense that it cannot be said (the 'self' as constituted is only the silent 'expression' which bears speech) and in the sense that this differentiation has no ground from which to arise. the 'real' does not consist of selves and others but populations and gatherings and proximities and contagions.
as far as writing goes it seems to me that the nonsense of metonymic differentiation-within-contiguity is the gesture of making a mark or a cut that releases the said along with the unsayable of intelligibility or exposure. 'writing' then is language's prescription. i don't know if this is what Derrida means when he places writing at speech's origin (i read that essay a long time ago) but this is my hunch going in.