The same link gives the same text
Most placeholder generators call Math.random(), which means the text changes every time anything on the page recomputes and the link you send a colleague shows them something different. This one draws from a seeded pseudo-random generator instead: the settings and one seed number decide every word, so the same address always produces the same paragraphs, on every machine. New text picks a new seed, and the seed travels in the URL with the rest of the settings.
That matters more than it sounds. Filler ends up in screenshots, in design reviews and in pull requests, and "the copy moved between my screenshot and yours" is a real half hour lost to something that was never supposed to be interesting.
Why the text is in Latin
The standard passage is scrambled Latin from Cicero's de Finibus Bonorum et Malorum, written in 45 BC and used by typesetters since the 1500s. The unfamiliar language is the point: a reviewer looking at a layout in their own language reads the words and comments on the copy, and a reviewer looking at Latin looks at the shape of the page. Latin also has letter frequencies and word lengths close enough to English that the type colour on the page is roughly what real copy will give you.
The plain English vocabulary is here for the cases where that backfires: a client who reads lorem ipsum as a bug, a spell checker or language detector that flags the page, or a locale where the Latin looks like an error rather than a placeholder. The sentences are still nonsense. That is deliberate too: filler that reads like finished copy is filler that ships to production without anyone noticing.
Common problems
- The paragraphs run together when pasted into HTML. Plain text separates paragraphs with a blank line, which HTML collapses to a single space. Switch the output to HTML and each paragraph comes wrapped in its own
<p>, list items inside a single<ul>. - The filler reached production. It happens on every project. Search your repository for
ipsumbefore release, and prefer the Latin vocabulary for exactly this reason: it is trivially greppable, and plain English filler is not. - The line lengths look nothing like the real copy. Check the word count in the status line rather than the paragraph count. Real product copy is usually far shorter than the filler people reach for, and a layout tested on 900 words of Latin often breaks on the 40 words it actually ships with.
Frequently asked questions
How many words is a paragraph here?
Three to six sentences, each 4 to 16 words, so roughly 30 to 90 words with an average near 55. Sentences longer than eight words get one comma somewhere in the middle. One comma rather than several: a clause break gives a paragraph its texture, and three per sentence is what makes generated text look generated.
Does anything I do here get sent anywhere?
No. The generator is a few dozen lines of JavaScript running in this tab, with no request to any server, and there is nothing to send in the first place: you type nothing into this tool, you only turn settings.
Can I get the same text again tomorrow?
Yes. Bookmark the address. The unit, count, vocabulary, format, opening and seed are all in the query string, and those six values are the whole input to the generator.
Is there a limit?
500 units, which is about 25,000 words of paragraphs. The generator is fast enough that the cap is about keeping the page usable rather than about speed.