Then again, any time the CE queries, there's the chance that the author will say, "No, I meant for 'decimate' to mean 'eliminate 90%,'" and the CE will just sigh.

Because that battle is well and truly lost. (I prefer the precise meaning myself, but nobody in the real world knows enough Roman history to care.)

Hey, I still think 'data' is plural.

"Data" IS plural; after editing scientific articles for 10 years, where you can bet every single instance of "data" is made plural, it drives me bonkers to read/hear "data" as singular in mainstream (non-scientific) media.

nobody in the real world knows enough Roman history to care

I care!

The battle on decimate is, as Betsy notes, so well and truly lost, that even the dictionaries make note of it. It no longer is incorrect usage. It is just not the original usage. There are lots of words like that.

There is a certain sect of Christianity that insists the 1611 version of the King James Bible is The Only Version, completely ignoring that one of the reasons for making modern translations is that not only has the style of speaking changed (thee, thou, cometh etc.), but the actual meanings some words carry have changed substantially, as well. When the KJV was first published, allege was used the way we would use prove. Suffer meant what we would mean by allow. They used allow the way we use approve. Let was used as we'd use hinder. Prevent was used where we'd use precede. Conversation is used where we'd use conduct.

Words change. We don't have to like it, but the CE who is sighing over decimate is, to my eyes, more wrong for refusing to recognize an actual (now meaningful) meaning, than the writer is for using decimate to indicate something was nearly wiped out.

Well, the trouble with 'decimate' in particular is that its root meaning is embedded in the word itself, 'dec-' meaning ten even today in words like 'decimal.' Hence, "reduce by 10%" is an easy meaning to infer. It's just irritating that family-groupings of words don't all grow in the same direction.

Hey, I still think 'data' is plural.

I had a colleague in the doctoral program for psychology at a university tell me that any mis-use of plurality around the word 'datum/data' in her dissertation or any qualifying exam is automatic grounds for failure. I think humanities people and the general public are very loose with it, but even the soft sciences are still careful about that word.

Now if only I could convince the Times that "millennia" is still a useful plural (they use "millenniums").

millenniums

If I saw that in the paper, I'd think I'd been wrong, and it would be all self-perpetuating and stuff.

Hence, "reduce by 10%" is an easy meaning to infer

Easy for who? I think the ease with which the original meaning has been lost contradicts your assumption.

How many English speakers know that they know the meaning of the "deci-" root anyway?

And even if you know deci- is a tenth, why wouldn't it be 10% left, not 10% gone? (Assuming you know deci- from metrics, but not -mate.)

I think humanities people and the general public are very loose with it,

Software people invariably use the singular; I've just given up.

Explode comes from "ex plaudere", to drive somebody off the stage by clapping. (Seriously. Look it up.) That isn't the meaning any more.