SQL Detective Mystery Unblocked

SQL Detective Mystery is a murder case where your only investigative tool is a query language. A body turned up in SQL City in mid-January 2018, and your only memory is that a police report about it was filed. Witness names, statements, car registrations and gym check-ins are all stored in tables, and you reach them by typing real SQL into boxes on the page. It was created by Northwestern’s Knight Lab, and here it runs unblocked with a full SQLite engine inside your browser, so nothing needs installing.

Working the Case

The page is a series of code boxes. Every box comes preloaded with an example you are free to change, and running it prints a results table underneath. The first boxes show you how to list every table and how to inspect the columns in one of them. That is your map.

From there it is classic detective work. Pull up the correct report using its date, crime type and city. Note the witnesses it mentions. Look them up in the person table, then read their interview transcripts. Their statements mention details like a gym membership level, part of a license plate and a check-in date, and each detail is a filter you can apply to another table.

Once you have a suspect, put the name into the answer checker at the bottom. A miss gets a gentle nudge. A hit opens the second act.

SQL Detective Mystery controls

To do this	Keyboard	Mouse	Touch
Write or change a query	Type in a code box	Click into a code box	Tap a code box to open the keyboard
Run the query	Shift + Enter	Click Run	Tap Run
Restore the box's original query	—	Click Reset	Tap Reset
Open the schema diagram	—	Click the "click here" link above the last box	Tap the "click here" link
See a suggested query on the walkthrough	—	Click Show Solution	Tap Show Solution

SQL Habits That Crack Cases Faster

1. Study the schema diagram before you query. It maps which ID columns tie the people, licence, interview and gym tables to one another. Knowing the links saves dozens of blind guesses.
2. Narrow results with WHERE. There are over a thousand crime reports, so combine the date, crime type and city in one filter.
3. Match partial text with LIKE. Sometimes all you know is part of a name or a street. A pattern with % wildcards finds them.
4. Sort for “the last” or “the first.” Sorting in descending order and keeping a single row answers questions like which house is at the end of a street.
5. Join tables instead of copying IDs by hand. A single JOIN can test a number plate, a membership and a visit date together.
6. Keep notes outside the page. Results scroll away and nothing is saved, so jot down every name and ID.

Who Should Play It

This is a perfect fit for students in a computing class, analysts who want to practise joins, or anyone who likes a whodunnit with a bit of homework. The dataset is large and realistic, with roughly ten thousand people, so you cannot solve it by scrolling.

If you like learning a technical skill through a puzzle, CSS Diner teaches selectors in a similar spirit, and Learn Git Branching does the same for version control.