Because it works better as a test of language ability. People can model others' minds but get lost in the many clauses.

That sounds plausible, but how would you determine that it's language ability specifically that's causing the issue? Do they pass the basic Sally-Anne test ("Where will Sally look for her marble?") but fail more complicated versions? Do they pass clear versions but not wordy versions (passing "Where will Sally look for her marble?" but failing "Where would Sally tell us she believes the marble is?")?

It seems to me that language failure could be because of theory of mind failure. For example, if I'm trying to trick someone trying to trick me, what I think he thinks I think he thinks is an unconsciously constructed object in my mental model. Describing it is a little tricky because there's not a single word for it, but making predictions / planning actions based on it is not difficult. If someone doesn't have that in their mental model, but has to construct it, then it seems to me the first place they'll notice difficulty is parsing the question- they look inside for something with the tag "what I think he thinks I think he thinks," find nothing, and conclude they probably didn't hear / parse it correctly. But, this is speculation by a non-psychologist, and so evidence-driven opinions are more welcome.

Here's a person who obviously has theory of mind. She fails the test. Thus the test isn't testing whether someone has theory of mind.

The test succeeds at identifying something important that most people consider obvious, but that (surprisingly) some people fail at. Whether the something that the test measures is best called "theory of mind" is debatable, but I'm not sure there is a better name that would have stuck.

Having names for things is useful for thinking about them, as long as the focus is on the thing ("how could we call the thing this test measures"), and not on the name ("how could we design a test for the presence of a theory of mind?"). There's a delicate balance to reach between avoiding sneaking in connotations ("let's call it the Soul Test!"), and not having obscure names that don't suggest anything ("let's call it Plasmeomorphic Synchronism!"). I don't see any reason to think that they struck a particularly bad balance in this case.

I have not heard this interpretation of the tests in question and I also did not have much prior knowledge of what exactly the tests hoped to prove and the subsequent conclusions, so I would greatly appreciate any links you could provide that may cause me to update my beliefs.

Specifically, I find both your explanation and the more mainstream explanation (something like http://en.wikipedia.org/wiki/Sally%E2%80%93Anne_test) plausible, so non-trivial evidence would lead me to put more certainty in a 'position'.

I don't know the history of these tests, so I will concede the point that perhaps these tests are used on non-NT's in a negative way, and that perhaps they were designed with some nefarious purpose of judging non-NTs.

However saying that because of this fact that these tests are "crap" is like saying that because people die in car accidents, that therefore cars are "crap". I think the test can be BOTH useful in demonstrating Theory of Mind that occurs in children 4 and under AND that the test can (possibly) be used in a negative way with non-NTs.

Perhaps to pass the test you need BOTH Theory of Mind, AND language skills at a 4 year-old level.

Even if it is just language skills, it still stands to reason to me, that if a child can't pass the test, that you won't be able to explain cognitive biases to them with any great success. (I am sure you could teach them the "Teacher's Passcode".) That is the reason I posted the test, and I think whether or not I am completely wrong on WHAT it tests, that the test still stands as good evidence supporting the claim I was making (can't teach them abstract material).

And I didn't even have to use this test specifically. I just wanted to show any example of the fact that children's minds don't operate on a developed enough level to yet understand abstract ideas.

Anyone even vaguely familiar with this will be very annoyed and likely to say things like "this test is crap"

You are using this as if it should EXCUSE the extremely rude grandparent comment, when instead it is an EXPLANATION. The two are not the same. You can explain bad behavior, without considering it an excuse.

Example- "I am late because I overslept" is an explanation not an excuse. "I am late because my car got a flat" is an acceptable excuse.

Though I didn't know the details, I already understood from the tone and the link, that the grandparent had some personal grudge/negative feelings towards the test in question. I assume it has been used on him/her to negative effect. I would guess, from your knowledge of this test, that it has also been used negatively on you or one of your loved ones. I am very sorry about this for the both of you. But that does not mean that it is an excuse to then leave rude and cryptic remarks, as the grandparent did.