Pick a World: A Tour of the Five Field Books

Clearstone’s books say about €8.3 million is overdue.

That’s a lot of unpaid invoices for a company with five departments. It also isn’t true. Clearstone is a fictional company, built for teaching and deliberately messy, and its invoices carry a status column that appears to have been filled in by someone with somewhere better to be. Ask the due dates instead of the label, as of 1 January 2027, the date the book works to, and the overdue pile shrinks to about €4.0 million.

That leaves the awkward part. A finance director who saw €8.3 million last week will want to know where the difference went. (How the ledger got into this state, and where its late payers are heading, is a story our Blog tells in full.) The answer is a table with three rows, and we’ll come to it, because it turns up in three different places on this site.

Five books, five worlds

That ledger belongs to Follow the Money, one of five field books in the LucidPlexus Library. They’re built the same way: a world with its own data, a question somebody needs answered, and faults in the data that you’re meant to find. None of them teaches a language in the abstract. Each one teaches it by using it on one world’s problem, from the first page to the last.

• Follow the Money. SQL from zero, on Clearstone’s budget, invoices, cash movements and departments. The question: is the company in good shape? You learn to ask the books for exactly the rows you want, then find what they get wrong, fix it in a way you can explain, and turn the answer into a report, a dashboard or a single number. The book runs 251 queries and prints every one with its result.
• Can I Trust This Data? The odd one out: no mascot company, just three agricultural spreadsheets (exports, prices and farm technology) and a short list of the countries they should mention. It covers the work between receiving data and acting on it: profiling, duplicates, gaps, wrong values that look right, cleaning in layers without touching the original, and a write-up someone else can rerun. It’s written in SQL, with each check also shown in pandas, spreadsheets and tools like dbt.
• Explain Your Query. The SQL interview, answered three ways: precisely, out loud, and on six tables from HealSync, a fictional hospital network. Its data has rows recorded twice, sessions whose doctor isn’t on file and one table with no key at all, which is roughly what interviews hand you. Wherever an answer changes between products, it says how for PostgreSQL, SQL Server, MySQL, Snowflake and BigQuery.
• Green on Paper. Python from the first line, on twelve tables about the Verdant Union, an invented union of five countries. Its Green Progress Report makes six claims; every one is arithmetically true, and every one misleads. You learn pandas and matplotlib by finding out how, with 25 faults planted in the data along the way.
• Worth Backing? SQL and Python together, on four years of records from Lanternfield, a fictional accelerator and seed fund. Eighty applications, $2.5 million, room for at most ten cheques. Which ten? It ends with a scorecard, ten picks and a one-page investment memo, and the data holds 30 planted faults for anyone who assumed startup spreadsheets would be tidy.

Look past the subjects and all five open the same way. Nobody hands you clean data and a puzzle with a known answer. Somebody hands you tables and a question, and some of what the tables say is wrong.

One book, five addresses

Here’s what makes them more than books. Each field book lives at five addresses on the site, and the five are built to line up.

• The data sits in the Community datasets library, table by table, each with a tick and a label saying what one row is. Anyone can browse it; downloading takes a free account. The Library’s Practice Dataset Pack also bundles all 32 tables in one zip.
• The book is in the Library. The quickest way to run its examples is the Workbench, a notebook in your browser with the tables already loaded. Four of the five list a companion download with the purchase: a two-page cheat sheet, a workbook, and the code for every chapter with a script that checks each result against the book.
• The Academy course is the book’s Academy edition: seventeen lessons, one per chapter, with the book’s figures and every query runnable on the page. Each course ends with a career project that a reviewer marks against a rubric. The three SQL courses make up the Data Analyst with SQL path; Green on Paper and Worth Backing? make up Data Analyst with Python.
• The guided project in the Community has seventeen tasks, one per chapter, all worked in one notebook. Each task follows one chapter of the book, and every task has an automatic check that reads what you built.
• The track wraps that guided project, under the same name. It’s open to every Community member, a human reviews the project, and finishing earns a certificate anyone can verify.

Time to settle up with Clearstone, and the money that wandered off.

Where the difference went

An analyst doesn’t just announce the smaller number and hope. They build a bridge: start from what the label calls overdue, take away the invoices that aren’t due yet, and add the ones the label missed, so that every euro of the difference has a named reason.

The arithmetic isn’t the point here, and the Blog has already taken that ledger apart. The point is where this bridge lives. In the book, it’s chapter nine, drawn as Figure 9-3. In the Academy course, lesson nine, When the Data Is Lying to You, draws the same figure, then a Try It box asks you to build it on the page. In the Community guided project, task nine is called Bridge the two overdue figures, and its automatic check won’t pass until your bridge closes to the euro.

Chapter nine, lesson nine, task nine: one bridge.

That’s the design, and it’s deliberate. You can read the bridge, then build it with the lesson beside you, then build it on your own with a check watching, without ever switching company, table or number. The book explains, the course rehearses, the project tests and the track keeps the record. Start wherever suits the way you learn; all five lead to the same bridge.

Where to start

If you’ve never written a query, Follow the Money and Green on Paper both start from nothing, one in SQL and one in Python. Worth Backing? also starts from the first line, only faster, teaching SQL and Python side by side; its course suggests Green on Paper as the gentler first step if Python is new to you. Can I Trust This Data? and Explain Your Query expect you to write a basic SELECT already.

Back at your desk

None of those places is real: Clearstone, HealSync, the Verdant Union and Lanternfield are all made up. Their faults have close relatives everywhere: a status nobody updated, a row loaded twice, a country under more than one name, a number typed in the wrong unit.

So which of these five worlds looks most like the one waiting in your inbox on Monday?