This only works one way: A good score according to a formula does in no way guarantee a text is easy to read. There are many factors involved, and they can't always be quantified. But a high score usually will guarantee it isn't, so it's one step on the way.

And what about combining two tools?

In T146899#2674876, @Trizek-WMF wrote:

And what about combining two tools?

What do you mean?

If you are afraid to have a biais if focusing on good score instead of understandability, you should use two tools instead of one and see which scores are the best. One tool may weight the other?

I'm not sure that'd take away the problem. The old truth about measuring things is that you get exactly what you measure (a low score on the Flesch–Kincaid reading ease scale) and not what you're trying to get at but can't quantify (ease of reading). I don't think we can get away from that even if we use more scores, and I think Flesch–Kincaid is pretty accurate for what it's trying to do. I'm just saying we should remember we can't look blindly at the scores, but that they're one step on the way. (:

For what it's worth, people who think it's a fun hobby to write an encyclopedia are typically going to have have above average reading comprehension :)

Probably true for most editors – when I translate Tech News into my native Swedish, I'm not as concerned about style as I am when I write it in English, because we have far fewer non-native readers. Still, I think we should try to make sure our information is easy to read. The main problem is that we want the news to be accessible to at least en-2 speakers who might not get it translated into their language. English could be their fourth language, they could come from a fairly monolingual culture etc.

Is this task stalled or supposed to move to Community-Relations-Support (Jan-Mar 2017)

I think you are doing fine. Not sure whether we really need to measure that consistently, but we can revisit if we find out readability is a real issue?