One of the commenters on “Dumb Copy Editing Survives” last week said something that worried me. My topic was the contrast between sentences of the sort seen in [1a] and [1b] (I prefix [1b] with an asterisk to indicate that it is ungrammatical):
 a. We are none of us native or purebred.
b.*We are, none of us, native or purebred.
What the commenter said was: “If I read the erroneous version, I would have still taken away the exact same meaning. I’d just think there were too many commas.”
This worries me because it seems to miss the crucial distinction between contexts where comma use is a free choice and contexts where there is a firm rule.
In the optional cases there are heavy punctuators, who use more commas, and light punctuators, who use fewer. Take clause-initial adjuncts, for example. A light punctuator would use [2a] where a heavy punctuator would use [2b].
 a. In the 19th century racist views were widely accepted.
b. In the 19th century, racist views were widely accepted.
Heavy punctuation is a good idea where the version without the comma has a confusing sequence of words; for example, [3b] (heavy punctuation) might well be judged preferable to [3a].
 a.For those who prefer wine glasses are provided.
b.For those who prefer wine, glasses are provided.
The familiarity of the phrase wine glasses tempts us to see it in [3a], which leaves no noun to serve as the subject of the main clause, so [3b] is preferable, despite the usual optionality of commas after clause-initial preposition-phrase adjuncts.
But there are cases in which the conventions of written Standard English do not allow any freedom of choice. One such is the principle that you do not put a comma between the subject and the predicate of a clause in contemporary English, even if the subject is long. (The rule was different in the 18th century; see the Second Amendment for evidence of that.) A comma between subject and predicate can be legitimate in modern prose, but only if it is one of a pair flanking a supplement (a parenthetical phrase). The upshot is that [4a] and [4b] are both grammatical, but [4c] is not.
 a. A new phylum named Lokiarchaeota representing a transitional form of life between the Archaea and the Eukaryotes was introduced in 2015.
b. A new phylum named Lokiarchaeota, representing a transitional form of life between the Archaea and the Eukaryotes, was introduced in 2015.
c. *A new phylum named Lokiarchaeota representing a transitional form of life between the Archaea and the Eukaryotes, was introduced in 2015.
Relative clauses provide another example. What The Cambridge Grammar of the English Language calls supplementary relative clauses (often known as “nonrestrictive”) must be flanked by commas, but the relative clauses known as integrated (or “restrictive” or “defining”) must not:
 a. Kathmandu, which is situated on an ancient lakebed, was massively damaged by the 2015 earthquake.
b. *Kathmandu which is situated on an ancient lakebed was massively damaged by the 2015 earthquake.
 a. An earthquake that shifts the Indian Plate northward by ten feet is bound to cause massive damage in Kathmandu.
b. *An earthquake, that shifts the Indian Plate northward by ten feet, is bound to cause massive damage in Kathmandu.
There is no optionality in the cases illustrated by –. The asterisked examples are simply incorrect under the usual grammatical conventions of contemporary written English.
And the construction illustrated in  has the same property. The none of us in [1a], understood as a quantifier on the subject noun phrase (so that the sentence has the meaning “None of us are native or purebred”), has to be an integrated part of the clause so that it can function as a negator and make the whole clause negative (hence making the or interpretable correctly). It cannot be treated as a parenthetical, so the commas are simply a mistake.
The lesson is that although there is much freedom in the matter of where to use commas in written English, the freedom is not unbounded. Strictly integrated constituents mustn’t be set off with commas, while supplementary (= parenthetical) constituents must be. I have often asserted that some prescriptive edicts about English are hokum and should be ignored; but that doesn’t mean that there aren’t any strict rules at all. Some of this stuff is nonnegotiable: You’ve got to get it right. Anyone who thinks they could read [1b] as having the same meaning as [1a] is not paying attention to the rules of punctuation. Admittedly, sometimes we have to be tolerant and extract the intended meaning of a piece of writing in defiance of the rules. But they’re still rules.