Write the Equation, Not a Screenshot of It

A screenshot of an equation is a dead end. You cannot search it, edit one symbol, or copy the math back out — and when the app that made it shuts down, the meaning goes too. A typed equation in dollar signs is the opposite: $E = mc^2$ is text you own, and text outlives the tool that displayed it.

This is not a tooling argument. It is an ownership argument. The same handful of characters renders on GitHub, in a research notebook, in a Markdown note, and inside a printed PDF, because the equation was never trapped in a picture. Donald Knuth designed the language that made this possible in the late 1970s,¹ and that language is now so stable its inventor has asked that it be frozen forever.² The screenshot is the modern trap. The plain-text equation is the escape.

What most people do: they paste a picture

The path of least resistance is a screenshot. You write the formula in one app, capture the rendered image, and drop the picture into your notes. It looks right. For a few months it even feels right. The equation is visible, the page is tidy, and the work appears to be saved.

The conventional view is reasonable on its face: an equation is something you see, so storing what you see seems sufficient. Word processors and proprietary equation editors reinforce this — they hand you a visual object, drag-and-drop, no syntax to learn. The picture becomes the artifact. And for a flyer or a slide, a picture is fine.

Why the picture fails: it has no source

A screenshot of an equation cannot be searched, edited, copied out, or re-rendered. It carries no source — only pixels. Search your notes for "Euler" and the image stays silent. Change one exponent and you must rebuild the whole thing in whatever app made it, if that app still exists. The picture is a tombstone for the math.

The deeper failure is portability. A proprietary equation object lives inside one application's file format; lift it out and it breaks. This is the same shape we wrote about in why a Markdown file stops being portable the moment you embed an image: the prose travels, the picture does not.

Steph Ango, who runs Obsidian, put the principle plainly: "Apps are ephemeral, but your files have a chance to last."³ A screenshot is not a file you can read. It is a file you can only look at.

What works instead: type the equation as text

Type the equation as plain text, wrapped in $...$ for inline math or $$...$$ for a display block. That single source, written in TeX-and-LaTeX-style syntax, is what GitHub, notebooks, static-site generators, and document converters all read. The same string renders everywhere a renderer is present, and stays human-readable everywhere one is not.

This is not a fringe convention. In May 2022, GitHub announced that "math expressions can now be rendered natively in Markdown on GitHub," noting it had been "a highly requested feature for over 8 years."^{4 5} Their mechanism is exactly the substrate described here: "you can use the $ and $$ delimiters natively in Markdown on GitHub to insert math expressions in TeX and LaTeX style syntax. This content is then rendered using the highly popular MathJax library."⁶ One source, written once. The renderer is a detail underneath it.

That detail varies, and that is the point — the source does not. KaTeX, an open typesetting library, "renders its math synchronously and doesn’t need to reflow the page,"⁷ and "produces the same output regardless of browser or environment, so you can pre-render expressions using Node.js and send them as plain HTML."⁸ MathJax, GitHub's renderer, aims for "beautiful math in all browsers." A static-site framework like Docusaurus simply states that "mathematical equations can be rendered using KaTeX."⁹ Different engines, one substrate. You write $\sqrt{2}$ once; the ecosystem competes to draw it.

And the source survives the journey out. Pandoc, which calls itself "a universal document converter," reads the very same delimiters ("Anything between two $ characters will be treated as TeX math") and turns your note into a PDF or an HTML page.¹⁰ The equation you typed in a note becomes the equation in a printed paper, untouched, because it was text the whole way through.

What this buys you: greppable, diffable, durable

A typed equation earns four properties a picture can never have. It is greppable: search "lambda" and you find every formula that mentions it. It is diffable: version control shows exactly which symbol changed. It exports anywhere a renderer or converter reads the syntax. And it outlives the app, because the file is yours and the format is open.

Durability is not a hope here — it is engineered. The TeX language Knuth built is described by the TeX Users Group as "a computer language and program designed for use in typesetting; in particular, for typesetting math and other technical material."¹ Knuth has deliberately frozen it: "Knuth wants TeX to be frozen at version π when he dies; thereafter, no further changes may be made to Knuth’s source."²

A format that cannot churn is a format your notes can lean on. An equation you typed in 2010 still reads, still renders, still diffs. That is the test local-first ownership was always about: not how a tool feels today, but whether your work walks out when the tool does.

The honest limits

Plain-text math is not magic, and pretending otherwise would be its own trap. Four caveats are worth stating plainly before you commit: renderers differ in coverage, the workflow is not WYSIWYG, one syntax rule trips up nearly everyone, and "renders everywhere" has an honest edge case. None of them undoes the ownership argument.

First, renderers are not equal. KaTeX is fast and synchronous but supports a narrower set of functions than MathJax; a few exotic constructs — certain commutative diagrams, niche packages — still need a full LaTeX pipeline. The trade is real, but it is a trade between renderers, not a reason to abandon the source: a fuller pipeline reads the same $...$ text.

Second, this is not WYSIWYG. You write code, not dragged symbols. For some people that is friction worth naming up front, even if snippet tools and autocomplete soften it. You are typing \frac{a}{b}, not clicking a fraction button.

Third, there is one genuine syntax gotcha. Per Pandoc's manual, "the opening $ must have a non-space character immediately to its right"¹⁰: write $x$, never $ x $, or the math silently fails to render. It is the single most common reason a beginner's equation stays raw.

Fourth, "renders everywhere" means everywhere a KaTeX- or MathJax-aware renderer is present. Open the same note in a plain text editor with no math support and you see the literal $E = mc^2$. That is not the failure case — it is the proof of the argument. The source is still there, still legible, still yours. A screenshot in the same editor would show nothing.

What to do tomorrow

The switch costs exactly one habit: the next time you reach for a screenshot of an equation, type the equation instead. Everything else follows from keeping the source rather than a picture of it. A few concrete moves make the change stick from the first note onward.

• Use the delimiters. $...$ for inline math, $$...$$ for a display block. No space right after the opening $.
• Keep the source, not the picture. If you inherit a screenshot, re-type the formula once. You will never re-type it again.
• Test it renders, then trust the text. Confirm one equation displays in your tool of choice, then stop worrying — the same string works in the next tool.
• Export to prove portability. Run a note through Pandoc to PDF once. Seeing your typed math come out the other side is the whole argument in one command.

Frequently asked questions

The questions below are the ones people actually type when they hit the screenshot wall: how to write the math, why it will not render, how to get it out as a PDF, and whether the format will still be readable years from now. Each answer points back to a primary source you can check yourself.

How do I write math equations in my notes as plain text so they render everywhere and stay editable?

Wrap the formula in dollar signs: $...$ for inline math, $$...$$ for a display block, using TeX-and-LaTeX-style syntax. That one plain-text source renders in GitHub, notebooks, and Markdown tools that integrate KaTeX or MathJax, exports through Pandoc to PDF or HTML, and stays editable text you can search and version.^{6 10}

What is the difference between KaTeX and MathJax?

Both render the same $...$ source; they differ in trade-offs. KaTeX "renders its math synchronously and doesn’t need to reflow the page" and can pre-render to plain HTML on the server.^{7 8} MathJax — the library GitHub uses⁶ — supports a broader function set and renders asynchronously. KaTeX favors speed; MathJax favors coverage. Neither locks your source.

Why doesn't my math render? I used dollar signs.

The most common cause is a space right after the opening dollar sign. Pandoc's manual is explicit: "the opening $ must have a non-space character immediately to its right."¹⁰ Write $x^2$, not $ x^2 $. A renderer that is missing entirely is the other cause, and then you simply see the raw $x^2$, your source intact, ready to render elsewhere.

Can I export Markdown math to a PDF?

Yes. Pandoc, "a universal document converter," reads the same TeX-style $...$ math directly and produces a PDF or HTML document from your note.¹⁰ Because the equation was plain text the entire time, it survives the conversion unchanged: the formula in your note becomes the formula in your printed paper.

Is a screenshot of an equation really that bad?

A screenshot renders fine, but that is its only virtue. It cannot be searched, edited, copied out, or re-rendered, and it carries no underlying source. Change one symbol and you rebuild the whole image in whatever app made it. A typed equation costs a few keystrokes and gives you all four of those abilities permanently.

Does writing math as text lock me into LaTeX?

No — it frees you from any single tool. The $...$ source is read by many renderers (KaTeX, MathJax) and converters (Pandoc), and stays plain text you can open in any editor.^{6 9 10} Want a full LaTeX pipeline later for an exotic construct? It reads the same source. The text is the portable layer; the tools are interchangeable.

Will these equations still work in twenty years?

The substrate is built to last. TeX is "a computer language and program designed for use in typesetting,"¹ and Knuth has frozen it deliberately so "no further changes may be made to Knuth’s source."² A format that cannot churn, stored in a plain file you own, is about as future-proof as a digital artifact gets.

A screenshot preserves how an equation looked on the day you saved it. The plain text preserves the equation. The first ages with the app that made it; the second is yours to read, edit, search, and re-render long after every app in this sentence is gone — so write the equation, not a picture of it.

This argument inherits Steph Ango's file-over-app principle — your files should outlast your tools — and narrows it to the symbols hardest to rescue once they are flattened into pixels.³ If you want those symbols to stay yours, keep them as plain Markdown on your own device, where a typed equation is text you can open offline and carry anywhere, including mnmnote.com.