10 Simple Unsolved Math Problems Anyone Can Understand

Quick answer: Simple unsolved math problems are questions whose rules are easy to understand but whose answers have never been proved. Famous examples include the Collatz conjecture (does every starting number eventually reach 1?), Goldbach’s conjecture (is every even number greater than 2 the sum of two primes?) and the twin prime conjecture (are there infinitely many prime pairs that differ by 2?). Computers and AI have made progress, but testing billions of examples cannot prove a claim about infinitely many numbers.

Facts checked November 22, 2026.

Mathematics has a strange habit: the questions that sound simplest are often the hardest to answer. A ten-year-old can understand each problem below, yet some have resisted the best mathematicians for centuries. In 2026 they have a new kind of challenger, AI systems that search for counterexamples and help write machine-checked proofs. This guide covers eight open problems and two that saw major progress recently, with something you can try for each.

Open Problem, Conjecture or Theorem?

TermMeaningExample
Open problemA question nobody has answeredDo odd perfect numbers exist?
ConjectureA statement believed true but not yet provedGoldbach’s conjecture
TheoremA statement that has been provedThere are infinitely many primes

Solving a conjecture means either proving it for every case or finding a single counterexample. For Collatz, a proof must show every positive whole number reaches 1; a disproof needs just one starting number that never does.

The 10 Problems at a Glance

ProblemOne-line statementStatus (Nov 2026)Try this
Collatz conjectureHalve evens, triple-plus-one odds: you always reach 1OpenStart with 7 and follow the rule
Goldbach’s conjectureEvery even number above 2 is a sum of two primesOpenFind two primes that add to 50
Twin prime conjectureInfinitely many primes differ by 2OpenList the twin primes below 100
Odd perfect numbersIs there an odd number equal to the sum of its smaller divisors?OpenCheck why 6 and 28 are perfect
Legendre’s conjectureA prime between every two consecutive squaresOpenFind the primes between 25 and 36
Perfect cuboidA box with all edges and diagonals whole numbersOpenCheck the face diagonals of a 44 × 117 × 240 box
Lonely runnerEvery runner on a track is eventually far from the restOpen in general; proved for up to 10 runnersSimulate three runners on a circle
Happy ending problemHow many points guarantee a convex polygon?Open in general; known up to hexagonsDraw five points and find a convex quadrilateral
Erdős unit distance problemHow many pairs of points can be exactly 1 apart?Erdős’s prediction disproved with AI in 2026; exact answer openFind 4 points with 5 unit-distance pairs
Moving sofa problemThe biggest sofa that turns a hallway cornerWidely supported 2024 proof (Jineon Baek); awaiting peer reviewSketch a sofa turning a corner

1. The Collatz Conjecture

Pick any positive whole number. If it is even, halve it; if it is odd, multiply by 3 and add 1. Repeat. The conjecture says you always reach 1. Starting at 27 takes 111 steps and climbs as high as 9,232 before falling to 1. Computers have checked every starting number up to \(2^{71}\), and Terence Tao proved in 2019 that almost all starting numbers eventually fall very low, but a full proof is still missing. A July 2026 “disproof” turned out to exploit a bug in a proof checker. See the Collatz conjecture explained.

Try it: Collatz Explorer

2. Goldbach’s Conjecture

Every even number greater than 2 is the sum of two primes: \(4 = 2 + 2\), \(28 = 5 + 23\), \(100 = 3 + 97\). Christian Goldbach proposed it in a 1742 letter to Leonhard Euler. It has been checked up to \(4 \times 10^{18}\), and the weaker three-prime version for odd numbers was proved by Harald Helfgott in 2013. See Goldbach’s conjecture explained.

3. The Twin Prime Conjecture

Primes that differ by 2, such as 11 and 13 or 41 and 43, are called twin primes. Are there infinitely many? Since Yitang Zhang’s 2013 breakthrough, mathematicians know there are infinitely many pairs of primes within some fixed distance of each other. That is a big step, but “infinitely many gaps of at most a few hundred” is not “infinitely many gaps of exactly 2,” and nobody has closed that distance. See the twin prime conjecture.

4. Do Odd Perfect Numbers Exist?

A perfect number equals the sum of its smaller divisors: \(6 = 1 + 2 + 3\) and \(28 = 1 + 2 + 4 + 7 + 14\). All 52 known perfect numbers are even. Pascal Ochem and Michael Rao proved in 2012 that any odd one would have to exceed \(10^{1500}\), and Ochem’s later computations push that bound past \(10^{2200}\). Still, no one can rule one out. Try it: add the proper divisors of 496. Do you get 496? See odd perfect numbers.

5. Legendre’s Conjecture

Is there always a prime between \(n^2\) and \((n + 1)^2\)? Between 4 and 9 there are 5 and 7; between 9 and 16 there are 11 and 13. A published computer check covers every \(n\) up to about \(7.05 \times 10^{13}\), but it is still unproved. See Legendre’s conjecture.

6. The Perfect Cuboid

Can a rectangular box have whole-number edges, whole-number face diagonals and a whole-number space diagonal? Boxes that get everything but the space diagonal, called Euler bricks, exist (the classic is \(44 \times 117 \times 240\)). A perfect one has never been found, and computer searches reported by Robert Matson indicate that any would need a smallest edge above 500 billion (a search result, not a theorem). See the perfect cuboid problem.

7. The Lonely Runner Conjecture

Runners start together on a circular track and run forever at different constant speeds. Will each runner, at some moment, be far from all the others (at least \(1/n\) of the track for \(n\) runners)? It was proved for up to 7 runners by 2008. In 2026, computer-assisted proofs for 8, 9 and 10 runners were published, and preprints claim more cases, but the general case is open. See the lonely runner conjecture.

8. The Happy Ending Problem

Scatter points in the plane, no three in a line. How many points do you need to be sure some \(n\) of them form a convex polygon? Five points always contain a convex quadrilateral, 9 points always contain a convex pentagon, and 17 points always contain a convex hexagon. Erdős and Szekeres conjectured \(2^{n-2} + 1\) points in general; the next case, 33 points for a heptagon, is unproved. The name comes from the marriage of two of the mathematicians who first studied it. See the happy ending problem.

9. The Erdős Unit Distance Problem: An AI Result

Place \(n\) points in the plane. How many pairs can be exactly 1 unit apart? Paul Erdős predicted the maximum grows only slightly faster than \(n\). Precisely, he predicted at most \(n^{1 + o(1)}\) pairs. On May 20, 2026, OpenAI announced an AI-generated construction showing that some configurations have at least \(n^{1 + \delta}\) unit-distance pairs for a fixed \(\delta > 0\), overturning that prediction. The exact maximum is still unknown. See the Erdős unit distance problem and Erdős problems solved by AI.

10. The Moving Sofa Problem: A Human Solution

What is the largest sofa that can be carried around a right-angled corner in a hallway 1 unit wide? In late 2024, Jineon Baek posted a 119-page preprint claiming that Joseph Gerver’s 1992 shape, with area about 2.2195, is the answer. Experts are broadly confident, but as of mid-2026 it had not completed peer review. See the moving sofa problem.

Why Checking Examples Isn’t Proof

Computers have tested Goldbach and Collatz for quintillions of numbers without a single failure. Why isn’t that enough? Because there are infinitely many numbers, and history is full of patterns that held for a very long time and then broke. A proof must cover every case. That is why mathematicians value proofs that a computer can check line by line, in systems such as Lean, and why a mere “it worked for every number we tried” never closes a problem.

How AI Is Helping

  • Searching for counterexamples: AI can explore huge spaces of possible constructions, which is how the unit distance prediction fell. See how AI disproves conjectures.
  • Checking proofs: AI helps translate arguments into Lean, where every step is verified by machine.
  • Catching errors: the July 2026 Collatz episode shows the other side: a flaw in the checker produced a false “disproof,” so verification tools themselves must be trusted and tested.

Common Misconceptions

“If a computer checked it up to a huge number, it must be true.” No. Infinitely many cases remain, and some conjectures have failed at enormous values.

“Simple problems must have simple solutions.” Fermat’s Last Theorem took 358 years and a proof over 100 pages long.

“AI has already solved all of these.” No. As of November 2026, the first eight problems on our list remain open.

Watch: Easy-Sounding Problems Nobody Can Solve

If you learn better by watching, these videos are worth your time.

The Simplest Math Problem No One Can Solve – Collatz Conjecture (Veritasium). A deep, visual look at the Collatz conjecture and why it is so hard.
Goldbach Conjecture – Numberphile (Numberphile). Why Goldbach’s simple statement has resisted proof since 1742.

Check Your Understanding

Five quick questions to see how much of this has stuck. Pick an answer and press Submit; you will see a short explanation either way.

Answer each question and press Submit to see your result right away.

1. In the Collatz process, what do you do to an odd number?
Found a bug? Let us know!
2. Which number is written as a sum of two primes as Goldbach predicts?
Found a bug? Let us know!
3. Which pair is a twin prime pair?
Found a bug? Let us know!
4. Why doesn’t checking Goldbach up to 4 × 10^18 prove it?
Found a bug? Let us know!
5. Which of these problems saw an AI-driven result in 2026?
Found a bug? Let us know!

Frequently Asked Questions

What are some math problems that are easy to state but hard to prove?

Famous examples include the Collatz conjecture, Goldbach’s conjecture, the twin prime conjecture, whether odd perfect numbers exist, Legendre’s conjecture and the perfect cuboid problem. All remained unproved as of November 2026.

What is the hardest simple math problem?

There is no official ranking, but the Collatz conjecture is often called the simplest problem nobody can solve: halve even numbers, triple odd numbers and add 1, and you seem to always reach 1.

Has AI solved any famous unsolved math problems?

AI has contributed to several results, including a 2026 disproof of Erdős’s unit distance prediction. Many famous problems, such as Collatz, Goldbach and twin primes, remain open.

Why can’t computers just prove these conjectures?

Computers can check billions of cases, but these conjectures are about infinitely many numbers. A proof must explain why the statement holds for every case, not just the ones tested.

What would count as solving the Collatz conjecture?

Either a proof that every positive whole number eventually reaches 1, or a single starting number that never reaches 1.

Can a student solve an unsolved math problem?

It is possible but rare. Understanding the statement is easy; making real progress usually requires deep tools. Many students start by exploring small cases, which is a good way to learn.

Keep Exploring

Sources and Further Reading

Related to This Article

What people say about "10 Simple Unsolved Math Problems Anyone Can Understand - Effortless Math"?

No one replied yet.

Leave a Reply