Martin (1992: 570):
Probably the main theoretical weakness as far as modelling [the system of genres that constitute our context of culture] is concerned lies in interpreting both system and process as what might metaphorically be referred to as "negotiation". This stems in part from a fundamental weakness in the dynamic modelling of the exchange; but it projects from there onto difficulties in dealing with tenor (the ways in which interlocutors treat in status, contact, affect), with genre (the interplay through which participants consummate, frustrate or abandon a genre) and on to ideology where tension among coding orientations vies with power, deprivation and systemic inertia to engender evolution. As Halliday comments in Thibault (1987):
I would interpret the power relations in a particular situation, when we represent that situation in terms of field, tenor and mode, by building into our representation that fact that the situation may be different things for different interactants. The total picture is obviously going to bring in all angles; but in any typical context of situation in which there is a power relationship of inequality, then the configuration embodied in that situation is different from the way it is seen from either end. This means, of course, that the register that is being operated by the interactants will be bifurcated, although we may choose to characterise the register of the situation as a whole by building in both strands. (1987: 620-1)
It remains to develop ways of building in "both strands" that show how text negotiates with system, and different systems with each other; lacking a model of this metasystemic dynamism, contextual theory remains dangerously incomplete.
 The main theoretical weakness, as far as modelling "the system of genres that constitute our context of culture" is concerned, actually lies in Martin's confusion of context potential (context of culture) with language sub-potentials (genres/text types/registers), as explained in previous posts. There has since appeared an entire book devoted to promoting this misunderstanding: Genre Relations: Mapping Culture (Martin & Rose 2008).
 This confuses potential (system) and the instantiation of potential (process) with interpersonal semantics, the system–&–process whereby interlocutors enact intersubjective relations as meaning. Note that Martin's argument is that it is this interpretation that is the main theoretical weakness as far as modelling genre systems is concerned.
the main theoretical weakness
[[interpreting both system and process as [[ what might metaphorically be referred to as "negotiation"]] ]]
Identified / Value
Identifier / Token
 The claim here is that the source (cause) of this theoretical weakness — the (mis)interpretation — is a fundamental weakness in the dynamic modelling of interpersonal semantics.
a fundamental weakness [in the dynamic modelling [of the exchange] ]
Identified / Token
Process: relational: causal: reason
Identifier / Value
 The claim here is that the theoretical weakness — the (mis)interpretation — extends to difficulties in modelling tenor, genre and ideology.
from there onto difficulties [[[in dealing with tenor … genre … ideology … ]]]
Process: relational: spatio-temporal
 This misunderstands Halliday. The Halliday quote is concerned with the question of modelling contexts in which two interlocutors see the same situation as different configurations of field, tenor and mode. These are the two strands that Halliday refers to.
 There can be no such "negotiation". Text and system are the same phenomenon viewed from opposite poles of the cline of instantiation.