I don't think so. You can still teach to a good test badly. For example: reading a variety of books and letting people do more on their own would be would be better overall than reading just things on the level of the test, asking just the test style questions and spending time learning the answering style to get 100%.

It's kind of like those articles about young kids passing IT certifications. Those tests are reasonably good, but those kids just memorised lots of material and never actually worked as, for example, a networking engineer.

Or like in IELTS style language test you can easily score higher (even than a native English speaker) if you learn the format enough.

… so if the test is good …

Reading is such a basic skill it seems like a pretty reasonable thing to come up with a decent test.

"When a measure becomes a target, it ceases to be a good measure, except when the measure is something basic, like teaching reading"?

Has the last twenty years of teaching to the test proven that out?

'Teaching to the test' has not caused the drop in literacy. Rather, it's teaching reading using bizarre methods, like whole word recognition instead of phonics. This has nothing to do with 'teaching to the test', since phonics would teach more to the test than whole word anything. A phonetically competent adult would be able to make out almost any English word. A whole word one would not.

"When a measure becomes a target, it ceases to be a good measure, unless the teacher is poor"?

This 'truth' is not always true. For binary questions, teaching to measure makes sense

Unfortunately, "is someone literate" is not a binary question. There's a whole range of possible answers there.

Is there a way to teach reading, which increases reading test scores, that is bad? I’m having a hard time imagining what that would look like. Just because someone comes up with a pithy saying that some people call a “law” doesn’t make it a universal truth.

you trying to establish some sort of measure for Goodhart's law?