Terence comments on Levi's comments on my comments on Latour's modes. (How's that for a referential chain?)
On the 'prohibition of category mistakes arising from illegitimate crossings,' it is not all crossings that are said to be 'bad.' Far from it - modes are co-dependent. What use is REF without REP? Were there nothing of the latter there'd be nothing to 'bring back.' Likewise, TEC cannot function without beings of REP to hybridise. In the habit chapter he gives the example that if a microphone goes out during a pastor's sermon then it takes an engineer to fix it and that's a REL/TEC crossing. Every mode such as REL, LAW, POL, etc. is related to a specific form of HAB and various METs and NETs - these uber-modes. The prohibition is on crossings that do violence to the modes and seek to erase one or the other of them. So, there's clearly a conservatism or conservationism inherent in that but it's not crossings that are illegitimate per se, only certain forms of crossing that try to state one mode in terms of another, thereby erasing the former (at least that's my understanding).
That said, he's altering his argument subtly as the book progresses. In the habit chapter he's now able to say that DC isn't all bad. It's now the result of well ordered processes; it's an appearance and appearance has its own mode of existence, HAB (a mode itself pluralised by the fact that every other mode instantiates it differently). Even the appearance of being 'critical' early on in the book vis-a-vis the 'evil genius' of DC was, in fact, just another mode to be asserted positively. An addition to the pluriverse, not a subtraction from it.
What is this kind of ontological optimism lacking? It doesn't prevent us from questioning or interrogating modes - indeed, the whole endeavour seeks to lay them bare, to de-habituate them and open them to appreciation (i.e. appreciation meaning an assessment of value). It isn't 'uncritical' in that sense; it is de-habituating. It doesn't stop us proposing new modes that transform the other modes by their entry into the ecosystem (indeed, this is highly encouraged). It doesn't stop us from combining modes in new ways, provided that we do not attempt to define one mode in terms of another. It doesn't necessarily mean that e.g. an ethnographer working with a church group must be essentially preoccupied with REL. That group is a confluence of all kinds of modes and cannot even begin to be understood mono-modally. Fieldwork is essentially polymodal, that much is clear. Nor does it stop us from extending e.g. the REL mode to situations that are not conventionally 'religious.' Indeed, as I mentioned before that mode is defined so abstractly that it could apply to almost anything.
However, this ontological optimism does tend to encourage us to 'keep our distance,' to let others just get on with their modal adventures unperturbed; to live and let live, so to speak. And this brings me back to what I said about chapter 1 with Latour exoticising the other in order to construct it as that which must be absolutely respected - 'respected' to the point where it seems unclear as to how we could even engage with it (a common problem, in fairness). An anthropology of almost any church group (to rejoin that example) would surely demonstrate that most Believers tend at least as much towards a kind of confluence of REF and DC overlaid with a hazy mysticism of HAB, speckled with flecks of MET, as they are to the translations described by REL. That Jesus really was the son of God, that there are indisputable moral certainties and that the Bible 'proves' this in a very real and referential sense (pointing to the book to make a point, quoting chapter and verse to prove a point, etc., everything points towards reference!) is fundamental to these forms of life.
The assertion that these are really just crossings (however legitimate) and that REL is the true essence of religion is not a merely descriptive, empirical assertion - it is a proposition, a declaration of value, an appreciation of value. It is an assertion that the proper form of religion is concentrated around the REL mode with crossings across that mode only serving its practical, institutional needs. It is a value judgement. And this is where Latour must depart (and if he does not do so then be deported!) from his self-styled evidentialist empiricism and reveals his true moral-political-philosophical colours.
These are not just neutral descriptions of modes as they really are but the distillation and rearticulation of the values of the Moderns - a kind of saving of the Moderns, an identification of what is valuable in their metaphysics, a flushing out of the poison and a reconstitution of the valuable materials as a carefully designed, rigorously controlled 'menu of values.' Like a kind of tapas menu (if you permit me the analogy) where some items are necessary (the bread and butter modes) and you're encouraged to try everything but some items are optional. All items, however, have been carefully crafted from many materials - an appropriately heterogeneous network. It is a gourmet restaurant, the cuisine is often stunning in its subtly and ingenuity but it can hardly pass its wares off as raw - nor would it want to. Everything is highly refined and, indeed, this is what makes the whole endeavour valuable; that is the chef's skill. No mediations, no value. Essential Latourism.
Long story short, the value menu is propositional, not simply descriptive. It may be based upon experience both as a concept and in terms of some fieldwork but that doesn't change the fact that its menu items are highly philosophically refined. Latour's rhetoric (for we really must distinguish it from the substance of his argument, there's no other way to read him) often implies an earthy, concrete empiricism-qua-evidentialism that is promised but never really delivered. The empiricism-qua-trajectories-of-experience is perfectly valid - he channels some experiences and not others. No problem there. But he should be more direct in his representation of his endeavours. His skill is great but he sometimes tries to pass it off as magic. This must not be permitted. A habit, an appearance of the worst sort.
REL isn't 'how religion works' (though I can accept that it may be part of how religion works), it's the part of religion that is asserted to be most valuable, the most appreciated. A kind of mereological judgement (and therefore critique since judgement connotes critique) - looking into the assemblage, picking out the most valuable part according to your own predispositions and making this the essence of the thing, whether it is functionally essential or not (that is an open question - a lab without REF is surely unthinkable - it'd be a total failure - but the same may not be true of religious groups and REL).
By doing this across a range of domains one ends up with a system of essential values that can be debated with others and their interpretations of what's essential and what's not. That's the way around the impasse of the rhetoric of description/empiricism vs. the reality of proposition/judgement. Take them for what they are and then they can be properly debated; indeed, this is the condition of possibility for diplomacy.
The essences are not idealisations. REL is not the idealisation of religious practice, it is one part of it - a part raised up and honoured by its identification. The mereological judgement extracts and abstracts this one component and raises it to the level of essence. Not because that's necessarily how things work but because that's how things work best - or so it is claimed.
If we take this approach to the modes then things open up and are pluralised again. If we ignore the continual, nagging suggestion that these modes have been rigorously derived from experience of the way things are and simply accept that they are derived from particular, narrow portions of experience that have been abstracted, translated and submitted for debate and scrutiny then, personally, I find little more to object to on this issue as long as we're also able to say that e.g. REL isn't confined to religious practice but is simply typified by it; that religion is the archetypical instantiation of it but does not 'own' it.
If we can assume all of that (a double dislocation) then I don't see what is to stop someone (a) detaching REL from religious practice in observing it in other kinds of situations and (b) insisting that REF is the appropriate mode for exploring fundamental questions of cause and reason, that the piousness of REL is sometimes an appropriate form of translation but it's nothing to base a life around and that religious practice is for the most part a crossing of FIC/HAB with some thoroughly misunderstood MET thrown in for good measure all wrapped up with an often dangerous ribbon of POL and LAW. That doesn't define any mode in terms of any other; its a perfectly legitimate diplomatic argument. It doesn't deny the reality of REL, it simply recognises that religious practice is neither determined by that mode nor is that mode exhausted by religious practice.
Apologies to anyone who managed to read all of that. I'm still working this out as I'm writing it! There must be a shorter way of making the point but that's all I've got for now.