The Quandary of xorlo
Interesting. Actually trying to work within misterism -- rather than sniping at it for being non-traditional, etc. — explains many things that seemed just arbitrary before. For a big example, the change in the meaning of {lo PA broda} and hence of {PA1 lo PA2 broda} seems just to satisfy a whim. But, if {lo broda} stands for Broda (brodakind, Mr. Broda, whatever) then PA2 could — in the tradition only be the size of the whole of set of broda, a meaning alreaady shown to be pretty well useless. The useful notion is the size of some subset of broda, whose members .... . But that is not directly available in this new notion, which does not consider the progeny of Mr. Broda quantitatively. But the same result can be achieved by going after Mr. PA-Broda, the parent of all bunches (or whatever word you want here) of PA broda including especially any of current concern. Of course, we are usually interested in a particular one and {lo PA broda} allows for more than one such bunch, but that is usually handled (as it has always been) by context. Going from there, [PA1 lo PA2 broda} will of course be PA1 bunches of PA2 broda, quantifying over the things whose parent lo PA broda is — just as {PA lo broda} does.
I am not sure about this, but it seems that the way that a predicate, {brode}, say, usually applies to {lo broda} is that Broda intersects Brode in such a way that the corresponding sets have members in common. The intersections are member by member, that is each broda involved is a brode: disjunctive distribution, for a phrase (which looks at the brodas rather than Broda, but I am not so deep into this that I can find suitable words here — the point being that the intersection need not be specified in terms of members). Presumably, {loi broda} also refers to Broda but now the predication is by bunches, not individuals, disjunctive collection. When we get down to individuals — typically by quantification — the collective/distributive distinction has to be made in some other way, which we still lack (but then this problem is in traditional Lojban as well outside of descriptions — and in them, in the defining phrases), so there is no loss here.
>From the point of view of Lojban ontology, the question floats around whether Broda, as a separate sort of thing, needs to be used. That it does seems to arise out of the problem with intersections or overlays or however it is put. If Broda were simply a set or a bunch, intersections would have pretty clearly to be defined member by member (set intersection or encompassing work that way) and that would screw up the use of {lo broda} — the reference to Broda — in "opaque contexts". It would be nice, if we are to use Broda, to have some better idea of its nature (axioms, say, or at least a number of truths, even if not guaranteed sufficient to characterize things completely). We know (I think) that {suo broda cu brode} implies {lo broda cu brode} and that the converse does not quite work (those "opaque contexts" again at least). But surely there are more things of this sort to have laid out all in one place.
It would also be useful to work out the various kinds of predications and which need to be marked in what contexts (and get the marks for them, of course). In connection with that, the exact nature of general claims, for which {lo broda} seems a natural expression. Actually, given {lo broda} the problem is more to give explicit expression to non-general forms, though this may be handled by restricting to {lovi broda}, say or by the modal parallel of {su'anai}. With expressions other than {lo} and quantified expressions (i.e., with the {le} and {la} series and pronouns), predication must be assumed to be either individual or, in any case, non-disjunctive. This is probably also true for restricted {lo}. That being said, only individual, distributive and collective need marking, as disjunctive can be assumed where it might apply. And, of course, one of these could also go unmarked, even in precise speaking, as the default form (individual — but that is relatively rarer? — or distributive). All of these — and something in the general/specific area — are needed also in the traditionalk language.
Since Broda has to exist in every world, whether or not there are brodas in that world (else {mi nitcu X} fails when there are no Xs), care needs to be taken in the predications to be sure that they are always drawing on the right set of instances (or whatever): We don't want {lo pavyseljirna cu blabi} to be true in a world where there are not unicorns, at least not for other than definitional purposes (and that might be better handled with some other expression).
Since Broda now appears to be formally indistinguishable from brodaness ({lo broda}as an individual from {le ka ce'u broda}), the objection that xorlo involves an expansion of the Lojban ontology no longer holds: with suitable tickling of the interpretation of disjunctive predication (for loci rather than chunks) we can make do with the traditional ontology (which is, Lord knows, already bloated enough for a metaphysically neutral — yeah, right! — language).
So there is no killer objection to xorlo. There are any number of objections to the way in which it has been presented over the years and especially in the official proposal. But these are excusable as partly due to the fact that some of the terminology and distinctions have only recently become available and partly to the inherent difficulty in laying out the system in the way it seems to have been conceptualized because of the discussions which led up to it.
What is left is just that the system is a radical departure from what seems to have been the pattern of Loglan-Lojban for its first half-century and that the change has been for little practical purpose and even involves some loss (for us logicians at least) even greater than those that fell out in the shift from Loglan to Lojban (the basic comparative nature of many adjectives, for example). As noted elsewhere, the pattern similarlities of {lo} and {le} are broken, after having been reinserted only a few years ago. And the zipfiness of some descriptions has been diminished slightly. These have been accompanied by no obvious gains and, indeed, with the loss of the old foundation of expressions in the cases rather than the abstract generalities. We have gained the legitimization of the move from {mi nitcu lo broda} to {lo broda zo'u mi nitcu by} but this does not seem to be much of an achievement: we already had the move from the old form, {mi nitcu tu'a lo broda} to {su'o da zo'u mi nitcu da} and the present system does not allow — any more than the previous did — the move to {da poi broda zo'u mi nitcu da}. Nor are any of these moves available for {le broda} or {la Brod} in the present system any more than the old. And, of course, in the old system we could get {tu'a lo broda zo'u mi nitcu by} anyhow. In addition to which, the new system leaves places where these peculiar restrictions apply unmarked in use, tempting one to try in opaque (well, at least still very cloudy) places these inferences which work in non-opaque places. I take this to be a net loss (or it would be if people had used the correct forms in the old days: what they use is now correct but misleading whereas the old forms were simply incorrect).
Let's see. What else is claimed as an advantage for this system? It makes generalities easier to say even in careful usage. But, of course, it makes particularities correspondingly more difficult to say in careful usage. And, I suspect, we deal more with particularities than generalities.
The metaphysics involved is, of course, dispensable. It was never the schemes most favorable point, being generally muddled by trying too many different explanations at once rather than separating out various interpretations or, better still, laying out what happens and leaving the nature of the things that make them happen open (there seem to be at least two stories and more than likely four about what lo broda is individually but they have at most heuristic value — and some of them damned little of that).
In sum, while xorlo is apparently a feasible system and one that does not depend upon some shaky metaphysics of one sort or another, it changes a number of things for no real reason at all. A couple of barely visible changes in what can be done better comes at the cost of losing a fairly significant distrinction and doing some minor screwing around with Zipf's law. I just don't yet see how the game is worth the candle here — aside, fo course, from legitimating a major part of the available. And even that must deviate in only very minor — and virtually automatically correctable — ways (assuming that the shifts here cover all the deviance).