The answer is is I don't know, but I'll speculate anyhow. Here's another case where it seems that a common syntactic pattern is a grammatical confusion:
(link) The thing is is that it all depends on the graphic card's drivers.
(link) The best thing is, is that they are spill proof.
(link) The worst thing is, is that a client believes them.
(link) "The important thing is, is that it is for all the right reasons—for Columbia College and its students," Schlossberg said.
(link) The amazing thing is is that this data can now be represented as a vector (gradient vector).
However, this time I think it's different. In the case of the extra that that I wrote about yesterday, I argued that people are indeed just confused: those are production errors. In the case of the extra is that I'm writing about today, I believe that people are not getting confused, but are producing phrases that are grammatical -- in terms of a non-standard grammar.
I could be wrong, in either or both instances. And some people might argue that it's misguided to make a qualitative distinction between production errors and grammatical modifications. But let's go on a bit anyhow.
Why do I think these are the result of a non-standard conception of English grammar, rather than just a faulty implementation of standard English grammar?
Well, for one thing, many of the examples are way too short for the speaker or writer to be befuddled by length and complexity. For another thing, I know from experience that (at least some of) the people who use double-is constructions don't see any problem with them, on reflection, whereas I suspect that (at least some of) the perpetrators of double-that constructions would see them as errors on careful reading. (Though this is a weak argument at best, since it might only show differences in degree of prescriptive awareness or obedience).
But the main reason is that it's easy to see how to make double-is constructions grammatical. One obvious idea is to treat them as variants of structures like:
(link) What the thing is, is not cowardly, but profoundly and detestably wicked.
(link) What the result is, is that when the carb gets hot, almost all of the clearance at the shaft is taken up by expansion.
(link) How serious the problem is is less important than how serious it feels to them.
These are fully grammatical in standard English -- the first one is a quote from the esteemed prose stylist G.K. Chesterton, from an essay that is well worth reading on other grounds.
This analysis is not without problems. As far as I know, every instance of a double-is construction has a plausible (fully grammatical) alternate with an overt wh-word; but the mapping doesn't always work in the other direction.
Thus I think you could get a non-standard version of the second example above:
The result is, is that when the carb gets hot, ...
but not the first one
???The thing is, is not cowardly, but profoundly and detestably wicked.
So it's not good enough just to say that [the ADJ N is] can sometimes be a noun phrase, roughly equivalent to [What the ADJ N is]. Still, this might work if we insist on the right kind of interpretation for what, which is surely different in the two sentences contrasted above.
The construction doesn't require a clause introduced by that -- it can be a bare sentence
(link) And the key thing is, is we have Elizabeth back right now.
or some kind of indirect question, as in the title of this post, or this internet example:
(link) To restate Rabbi Nachman's point about fear, life is a narrow bridge and the important thing is is not whether you are afraid; the important thing is that you choose to try to cross the bridge.
It seems to be optional -- at least, there are lots of other examples like the one just cited, where the double is phases in and out:
(link) And we've got some problems. One problem is there's misinformation about these cards. Another problem is, is that people -- they feel like it may be too complicated.
There are some examples where the complement (of the second is) is just a nominal:
(link) The answer is is verse 22.
or an infinitive
(link) I think the answer is is to have Thread B not terminate but rather have the Thread A delegate release the Mutex for Thread B when bytes are receieved.
or some other sort of element:
(link) Let’s say the answer is is no.
There's a theory of this presented in the following paper, which I haven't read (though I took 2/3 of the title of this post from its title): Tuggy, David. 1996. "The thing is is that people talk that way. The question is is why?." In Eugene H. Casad (ed.), Cognitive linguistics in the redwoods: The expansion of a new paradigm in linguistics , 713-52. Cognitive Linguistics Research, 6. Berlin: Mouton de Gruyter. Its content is summarized here. Apparently Tuggy points out that this construction gets reinforcement not only from the what-clauses, but also from real production errors, namely disfluent repetitions of is, a very common occurrence.
It's worth noting that this construction, though stigmatized, is widely used by highly educated people. I have a valued colleague who can be counted on to use it several times per lecture, and here's a quote from Bob Moffet, who is a health analyst at the Heritage Foundation and was deputy assistant secretary in the Department of Health & Human Services during the Reagan administration:
(link) But the important thing is, is that it would give individuals and families the right to pick and choose the plans they want at the prices they wish to pay and control their own health care.
I can't resist adding one more quote before closing -- the one that I took the last third of the title from:
Posted by Mark Liberman at June 27, 2004 10:31 PM(link) You got it; that's the question. The answer is is (drumroll please) "Delegate to the modules".