Smart Quotes, Real Dashes, and the Typography Your Notes Lose in Plain Text

Correct typography uses curly quotes and real en and em dashes, not the straight marks and double hyphens a typewriter left behind. For a note, though, it is a tradeoff: Unicode punctuation is correct, plain ASCII is portable and searchable. The best of both is to write in ASCII and let your renderer make the marks smart.

The characters at issue are old. The straight quote and the lone hyphen are typewriter compromises, one key doing the work of several marks a printer once set by hand. Unicode restored the distinctions: a code point for each curly quote, and separate points for the hyphen, en dash, and em dash ¹. Matthew Butterick, whose Practical Typography is the working reference for this, states the rule without hedging: "Use curly quotation marks, not straight ones" ². The question for a note is whether correctness is worth the portability you trade for it.

Should your notes use smart quotes and em dashes?

Yes if you want correct typography, no if you need maximum portability — and you rarely have to choose. Curly quotes and real dashes are the typographic standard; plain ASCII is the searchable, universally safe fallback. The durable answer for a note is to type ASCII and let a renderer convert straight quotes and hyphens into smart marks.

This is a tradeoff, not a verdict. The correct-typography case is strong and old: in traditional printing, all quotation marks were curly, and the standards bodies agree the curly apostrophe is the right one ^{3 4}. The portability case is just as real: ASCII travels everywhere, survives any pipeline, and answers a plain search. Treating punctuation as an either-or misses that a Markdown source and its rendered page can hold different characters on purpose.

Correct on the page, plain in the file. You can have both.

The four curly quotes and the two straight ones

There are four curly quote characters and only two straight ones. Unicode gives each curly mark its own code point: U+2018 and U+2019 for single quotes, U+201C and U+201D for double ¹. The straight marks are typewriter compromises — one vertical stroke standing in for both an opening and a closing quote.

Butterick counts them the same way: "There are four curly quote characters: the opening single quote, the closing single quote, the opening double quote, and the closing double quote," against the two generic vertical marks a keyboard offers ⁵. The distinction matters because the marks are directional. An opening quote leans one way, a closing quote the other, and the typeface draws them to match its other letters. This is punctuation composed inside a typeface, which is one more reason the font you read in shapes the page.

Two straight strokes, four correct marks. The keyboard is the compromise.

Hyphen, en dash, em dash: three marks, three jobs

The hyphen, en dash, and em dash are three different characters with three different jobs. Unicode names them U+002D hyphen-minus, U+2013 en dash, and U+2014 em dash ¹. The hyphen joins words, the en dash spans ranges and pairs, and the em dash breaks a sentence. Typing two or three hyphens is the typewriter workaround for marks a keyboard omits.

Butterick names the widths so you can see the difference: "The em dash ( — ) is typically about as wide as a capital H. The en dash ( – ) is about half as wide" ⁶. He is blunt about the hyphen habit, too: "En and em dashes are often approximated by typing two or three hyphens in a row ( -- or --- ). Don't do that—it's another typewriter habit. Use real dashes" ⁶. The mark you reach for should match the job, not the key.

Three characters, three purposes. The keyboard hides two of them.

Why grep can't find your apostrophe

The correct apostrophe is not the ASCII apostrophe. Type a contraction with smart quotes on and you get U+2019, the right single quotation mark — the same glyph Unicode and Chicago both name the preferred apostrophe ^{3 4}. Search that note for the straight ASCII apostrophe and the line will not match. This is the real cost of Unicode punctuation.

The gap is easy to demonstrate. Write the same word twice, once with the ASCII apostrophe U+0027 and once with the curly U+2019, then search for the plain version:

printf "It isn't there.\n"            > note.txt   # ASCII apostrophe U+0027
printf "It isn\xe2\x80\x99t there.\n" >> note.txt   # curly apostrophe U+2019
grep -c "isn't" note.txt            # -> 1  (matches only the ASCII line)

The count comes back as 1, not 2. The curly line is invisible to a search that types the straight apostrophe, because at the byte level they are different characters that only look alike. That is a portability tax paid at the byte layer, which is what "plain text" actually turns out to mean.

A note you cannot search is a note you have half-lost.

Author in ASCII, render smart

The way out of the tradeoff is to separate what you type from what you see. Write straight quotes and double hyphens, then let your Markdown renderer convert them: Pandoc's smart extension turns straight quotes curly, -- into an en dash, and --- into an em dash ⁷. Your source stays plain and searchable; the page reads as correct typography.

The mechanism is well documented. Pandoc's manual describes the extension exactly: "Interpret straight quotes as curly quotes, --- as em-dashes, -- as en-dashes, and ... as ellipses" ⁷. What you keep on disk and what a reader sees are two different strings of characters:

What you type (ASCII):   "It isn't done" -- he said, 3--5 p.m.
The correct marks:       “It isn’t done” — he said, 3–5 p.m.

Markus Kuhn, at Cambridge's Computer Laboratory, framed the same split years ago: "If you can use only ASCII's typewriter characters, then use the apostrophe character (0x27) as both the left and right quotation mark... If you can use Unicode characters, nice directional quotation marks are available in the form of characters U+2018, U+2019, U+201C, and U+201D" ⁸. Write for the machine; render for the eye.

When plain ASCII is the right call

Sometimes ASCII is not a compromise but the correct choice. A file read by machines, shared across unknown systems, or grepped often is safer in pure ASCII, because Unicode punctuation garbles when a file is served without a declared character set. Correct typography is a property of the rendered page, not always of the bytes you should store.

This is why a machine-readable file served as plain text is often kept deliberately ASCII: an llms.txt published for AI crawlers, delivered without a stated charset, will turn a curly quote or em dash into mojibake, so the safe version uses none. The honest limit on the craft side is worth stating too. Butterick's claim that curly quotes are "more legible on the page and match the other characters better" is a typographer's judgment, the settled consensus of the trade, not a measured legibility study ⁵. It is the right default for reading; it is not a law of physics that binds a plain-text file. When the reader is a program, how a note looks matters less than whether every byte survives the trip.

Correct is contextual. The right mark depends on who reads the file.

Frequently Asked Questions

The short version: correct typography is curly quotes and real en and em dashes, the apostrophe is U+2019, and ASCII is the portable fallback. These answers cover the punctuation questions people actually search — straight versus curly, hyphen versus dash, and why a curly apostrophe hides from search.

Should I use smart quotes?

For anything people read, yes — curly quotes are the typographic standard, and Matthew Butterick's rule is blunt: use curly quotation marks, not straight ones. For a note you also grep or share as plain text, the safe move is to type straight quotes and let your renderer make them curly, keeping your source searchable.

What's the difference between an en dash and an em dash?

Width and job. The em dash (U+2014) is about as wide as a capital H and breaks a sentence, the way an interruption does. The en dash (U+2013) is roughly half that width and spans ranges and pairs, as in a 3 to 5 p.m. window. The hyphen is shorter still and only joins compound words.

How many hyphens make an em dash?

None, strictly — an em dash is its own character, U+2014. The habit of typing two or three hyphens comes from typewriters. Under a smart renderer the convention is fixed: two hyphens (--) become an en dash and three (---) become an em dash. Butterick's advice is to skip the workaround and use real dashes.

Why does grep not find my apostrophe?

Because your apostrophe is probably not the ASCII one. With smart quotes on, a contraction like isn't uses U+2019, the curly right single quote, while your search types the straight U+0027. They look nearly identical and are different characters, so the plain search skips the line. Search for the curly character, or store ASCII.

What is the difference between an apostrophe and a single quote?

In practice, none — they are the same glyph. Chicago states an apostrophe is identical in appearance to a right single quotation mark, Unicode U+2019, and both Unicode and Chicago name U+2019 the preferred apostrophe. The opening single quote, U+2018, points the other way and is not an apostrophe, which is where smart-quote software sometimes guesses wrong.

How do I type curly quotes and em dashes?

You usually do not have to. Turn on smart quotes in your editor and it substitutes them as you type — a feature Butterick notes has been built into word processors for nearly 30 years. To type them directly, use your platform's shortcuts, or write ASCII and convert at render time with a tool like Pandoc's smart extension.

Straight quotes and doubled hyphens are a habit the typewriter gave us and no keyboard has fully taken back. Correct typography is a choice you make on purpose, one character at a time — and the choice is yours, per note, to store as plain bytes and render as clean marks.

MNMNOTE keeps your notes in open Markdown on your own device, where the characters you type stay yours to keep, search, and render however you like: mnmnote.com.