Jim Ryan defends a form of verificationism according to which a term (e.g. "zombie") is meaningless if there's no possible evidence that would indicate whether it's being correctly applied. As Jim summarizes the core intuition: "if a person has no idea how to tell whether what he's saying is true, then he doesn't know what he's saying and doesn't mean anything by it."
I'm sympathetic to this, insofar as I think that there is an epistemically transparent component to meaning (i.e. primary intensions). That is, if a person makes a meaningful assertion, then given enough information about the actual world, they will be in a position to tell whether the assertion is true. [This is a loose version of what Chalmers calls "the scrutability of truth".] I think this captures the core of Jim's intuition. But - importantly - this proposal does not require that the given "base facts" be themselves knowable.
Consider the term 'zombie'. This is meaningful, as I know very well how to apply it for any given scenario. Give me a world where the physical base facts are identical to ours, but -- perhaps due to a lack of bridging psycho-physical laws of nature -- no phenomenal properties (i.e. conscious experiences) are found among the psychological base facts. I can tell, here and now (indeed, a priori), that my claim 'Jim is a zombie' is true of that world, comprising those base facts. What I can't tell a priori -- or, perhaps, at all -- is what base facts truly obtain, or which coherent scenario (possible world) is actual.
Jim's verificationism might be understood as starting from the above scrutability thesis, but adding further restrictions on what the contingent "base facts" of reality may consist in. In particular, he requires that the base facts themselves be epistemically accessible, at least in the limited sense that we can envisage possible evidence for and against them. But while this starts in the right place, I think the added restriction goes too far. We should merely require that the base facts be comprehensible, in the sense that someone could understand the difference between scenarios where they do or do not obtain.
Thus, for example, my inability to comprehend the difference between a world with chairs vs. one merely with atoms-arranged-chairwise, (assuming it isn't just a contingent mental block on my part,) leads me to conclude that there is no such difference to be found in reality.
However, I surely can understand the difference in positing any number of spatiotemporally isolated "universes" besides our own, even if causal closure precludes any possible evidence for their existence from reaching me in this universe. Similarly for zombies: we can understand physical properties, and phenomenal properties, and what it would be for one to hold without the other. There's no need for these (even insuperable) epistemological problems to translate into metaphysical ones. We can make sense of reality being "all possible evidence"-transcendent in these cases, so long as it remains rationally apparent precisely which base facts the posited difference consists in.
A final example: Jim uses his verificationism to support "the Humean view" that internal incoherence and factual error are the only possible reasons to revise one's values. There's no sense to be made of a fully-informed and maximally coherent view that somehow remains in "error". For what in the world is left for the idealised agent to be wrong about?
I'm with Jim on this point (and would expand it to all philosophical claims, not just ethical ones). But we don't need his full-blown verificationism to get there. It's enough to note that free-floating moral properties would not make for a comprehensible addition to the base facts of the world. (There's no moral equivalent to the zombie world, e.g. where Hitler acts exactly the same, and yet fails to be immoral.) Add to this our thesis that all truths are rationally scrutable from the base facts, and it follows that any moral truths, in particular, will be accessible in this way.
[Is that right? The argument feels a bit slippery, to me...]
All up, I think my proposed alternative captures the main benefits of Jim's verificationism, without the costs. Suspiciously incomprehensible entities may be rightly discarded, without losing the grip we surely have on familiar physical and phenomenal properties that could conceivably be arranged in undetectable ways.