FAQ
How does it work?
The generator uses a 4-gram character-level language model trained on Utah baby names from 2000 to 2024. It builds names letter by letter, using the last three letters to decide the next one. This gives the generated names that special Utah flair (if not always readability).
Are these real names?
Nope. These names are based on actual Utah naming trends, but none of them exist in the official Social Security Administration (SSA) dataset. It’s Utah-style naming, freshly generated.
Keep in mind that the SSA only includes names used five or more times in a year, so some rare-but-real names can still sneak through.
Occasionally I'm surprised by the real-sounding names generated. Doesn't Brigdon sound like it should be all over the local elementary school? Apparently it's not . . . yet.
You can make the names real, though: just use one on your baby.
Do you recommend that?
No, I do not.
Why do some names sound . . . whackadoo?
Lots of reasons:
- This is Utah. I mean, check out the training data.
- These names are filtered to be unique, so the generator skips over anything “normal.”
- The model “forgets” where it is after three letters, which leads to names like Makennifernancescarli. Clearly this is objectively how we should be using statistical machine learning to make the world a better place.
How many names are in the dataset?
Roughly 112,000 unique names, pre-generated and stored in a static JSON file. The site runs entirely in your browser.
Why not generate names live?
To keep things fast, cheap, and scalable. (Inspired by the original Wordle site; did you know the HTML used to give away every day's Wordle solution?)
How is gender assigned?
It’s based on the original training data. But don’t let that limit your creativity.
Why are there more girl names than boy names?
The original SSA data shows a wider variety of girl names--more creativity, more spelling variations, more experimentation. So when the generator mimics those trends, it naturally produces more unique girl names. The model isn’t biased; it’s just reflecting the data.
Why Utah names?
Because Utah baby names are legendary. This project was inspired by the unique naming culture here, and classic sources like:
– Wes and Cari Clark’s Utah Baby Namer
– Eric D. Snider’s The Nayme Gaimme
Who made this?
I’m Sara Rands. I built the backend name generator during a machine learning class in 2014 to amuse myself and pad my GitHub.
It worked. This project helped me land my first software engineering job and led to a career at Domo.
Then I pivoted to a career as a psychotherapist, specializing in techy overthinker types, neurodivergence, faith transitions, trauma, and cancer survivorship.
You can find me at sararands.com.
Also, credit to ChatGPT for helping create this actual website.
I always intended to turn my original Terminal-only name generator into a website, but it stayed on the back burner for years, until AI helped it to feel feasible. Now, thanks to the power of machine learning, these remarkable names can be accessible to all, not just those who can run a Java program in the Terminal.
Can I reuse or adapt this?
Yes! The project is open-source and available at github.com/SaraRands/UtahNameGenerator