The Premise: What This Site Does and How It Checks

Puzzles, Charts & Dilemmas is a practice ground for people who like a hard question but have had enough of headlines that dare them to be one of the few who can solve it. Each entry takes one named puzzle, paradox, chart, thought experiment or dilemma, writes it out in standard form, and then checks the answer against the rules or the original data.

The name covers three of the four sections. The fourth, thought experiments, has no word in the title but gets equal room.

What you will find here

Philosophy & Logic Puzzles come in two kinds. Some have a finite set of rules and a single checkable answer: the river crossing, knights and knaves, the Tower of Hanoi. Others are paradoxes where the answer depends on which logic or decision rule you adopt, such as Newcomb's problem and the unexpected hanging. The first kind gets a checked solution; the second gets each side's argument side by side, with no verdict where philosophers still disagree.

Chart Literacy redraws famous and misleading charts from public data and sets an honest version of the same numbers beside the original technique. Simpson's paradox, Florence Nightingale's rose diagram, Minard's map and Anscombe's quartet are all here. Each chart entry carries a “Same data, redrawn” pair and a CSV you can open yourself.

Thought Experiments & What-Ifs rebuild a famous imagined scenario premise by premise, among them the Ship of Theseus, Galileo's falling bodies and the gravity train. Each one has an Obvious / Checked box: the obvious version next to the version that survives a look at the sources. What-ifs appear only when they are anchored to a named thought experiment or event.

Applied Philosophy & Ethical Dilemmas set out each side of a named dilemma, among them the prisoner's dilemma, the experience machine and the veil of ignorance. They discuss arguments only and give no advice on real decisions.

The site is written for curious readers who have heard of a named puzzle or paradox and want to see it worked through. It is also written for teachers and data-literacy instructors who need a worked example or a variant table for a class, and for writers checking how a chart misleads before they cite it.

This site is not the place for a test answer key or a score that ranks how clever you are. It does not give advice on personal medical, legal or financial decisions. It does not teach spreadsheet or business-intelligence tools, and it does not fact-check the chart in today's news.

Where the site stands

When an argument will not end, check first whether the two sides are answering the same question. The first step in clear thinking is to rewrite the question until both sides would sign it. That is the position this site takes, not a finding about how people reason, and it is why the motto is Untie the question first.

A few working positions follow from it, and each can be argued with:

• Write down the rules before computing a probability. The answer to a probability puzzle often depends on rules it leaves unstated: who knows what, and how each choice is made. Where the rules are missing, the entry says so and shows the answer under each reading.
• Check the denominator before the axis. A count without a denominator can mislead without a single false number, so it sits on the same checklist as a truncated axis. This is the order of checking used here, not a claim about which trick is more common.
• A moral dilemma is a comparison exercise, not a decision guide. Change one condition, see whether the judgment changes, and say what else changed along with it.
• Moral intuitions are data, not verdicts. A theory can overrule an intuition, but it has to pay for that and say what it paid.

These positions concern the structure of arguments and charts. The site takes no side on political parties, on whether any religion is true, or on the moral worth of any group of people. In the dilemma entries, the site's own view appears once, labeled “Our read:”, just before the closing objection, and you can skip it.

A tangled question is a knotted rope, and the aim is to loosen it rather than cut it: you should leave each entry with one move you can reuse on a question this site never covers.

The yellow on these pages is a highlighter dragged across a legal pad, an image chosen for the site because it marks only the cell a reader should look at.

How answers are checked

The method depends on the kind of question.

Puzzles with a single answer. The rules are written as conditions that can be enumerated, and the answer is checked by exhaustive search or with a solver that lists every solution. The entry reports whether the solution exists, whether it is unique, and every solution if there is more than one. Probability questions get an exact calculation as a fraction and a simulation of at least 1,000,000 rounds, and the two must agree to within half a percentage point. Each puzzle names where it was first published, or says that this could not be established. The tempting wrong answer is shown and explained first, and the checked solution sits in a fold below it.

Disputed paradoxes. No solver decides a philosophical question. Instead, each position is written as an argument in standard form and checked for validity: if its premises were true, would its conclusion follow? Each premise is traced to its source, and each position is tied to a specific passage or a reference-work entry. When philosophers still disagree, the entry says so. Competing positions are never labeled wrong answers.

Charts. Every chart is redrawn from a public data file you can download beside it, and the data values quoted in the entry come from that same file. Sources are named at a level you can check, such as a dataset page or a report table. Where only a bibliographic record could be read, the entry says that plainly instead of implying the full paper was read. No chart is traced from someone else's image.

Thought experiments. Historical premises need two independent sources; where only one could be read, the entry names it beside the claim. Scientific premises rest on textbook-level references or refereed journal articles, and a preprint is labeled as one.

Dilemmas. Each variant table marks the source of every judgment: a published experiment (giving its source, sample size and year), the reader's own answer, or the site's own hypothesis, labeled [hypothesis]. A changed judgment is reported as something the swap may affect, not as proof of what drives the judgment.

When sources conflict. Original papers and primary texts come first, then official databases, then the Stanford Encyclopedia of Philosophy and the Internet Encyclopedia of Philosophy, then textbooks, then everything else. A conflict is stated in the text rather than settled quietly.

What you can download. Puzzle checks and charts come with their outputs as CSV or JSON files, so you can run your own comparison against the same answer. There are no code files to download.

Illustrations. Drawn scenes are labeled Illustration and carry the line “Generated schematic illustration — not a photograph or historical record.” They carry no evidence and never stand in for a real person, document or chart.

Found a mistake?

Write to contact@puzzleschartsanddilemmas.com; the error link at the foot of every page opens the same address. Please name the page and, where you can, the source that shows the error.

Corrections are public. Each one is noted at the foot of the entry it changes and listed on the corrections page. If a puzzle answer turns out to be wrong, the wrong answer stays on the page with a line through it and the corrected answer follows. Nothing is quietly rewritten.

Reusing and citing

Figures drawn for this site, its data tables and its check outputs are released under CC BY 4.0. Third-party images keep their own licenses, credited under each image. The details are on the reuse and licensing page. Every entry ends with a “Cite this page” block giving its title, its publication date and its last substantive revision.

Three terms used here

• The decoy answer: the first answer to a puzzle that looks right and is shown, by checking, to be wrong.
• The swap test: change one disputed condition in a dilemma, list what else changed with it, and see whether the judgment moves.
• The lone count: any number given with no denominator and nothing to compare it with.