Preview. The search is not live yet; the lights on the globe are simulated.
The goal
One laptop is nothing. Twenty thousand tabs for a month is a step nobody has taken for 114. Two million tabs for a day would settle it. Every person counts, literally.
The browsers together become one machine, bigger than any university cluster, that costs nothing to run. The globe above is that machine, drawn live.
114 is the first target. Then 390, 627, 633, 732, 921 and 975, the last six below 1,000. Then the same engine pointed at other open problems where raw computing power decides.
This only works if it spreads. Every share is computing power. Send it to one person who would leave a tab open.
The search
Trying every x, y and z would take longer than the universe has existed. The trick, found by Andrew Booker in 2019: if x³ + y³ + z³ = 114, then x + y must divide 114 − z³ exactly. So the search walks through every possible x + y, and for each one only a thin list of z can work. A test with small primes throws out 13,000 of every 13,001 candidates in a millionth of a second. The rest get the exact arithmetic.
The map
Every square is a whole number. Grey ones are impossible, they leave remainder 4 or 5 when divided by 9 and no three cubes can ever add up to them. Dark ones are done. The seven lit squares are the ones nobody has cracked. 114 comes first.
What a solution looks like
42, found in September 2019 by 500,000 volunteer computers after 65 years. Three numbers of seventeen digits. The answer for 114 will look like this, and anyone with a calculator can check it.
How it works
Laptop or desktop with a modern browser. Phones can watch the globe and the numbers; the work needs a real machine.
Your browser is handed one slice of the search, a range of values of x + y to check. Nothing is installed. Nothing runs once the tab is closed.
Every slice is checked by two different computers before it counts. Finished slices push the frontier forward, and your name goes on the record if you want it there.
The odds
Each step of the search costs ten times the last one and carries roughly the same chance of holding 114. The only thing that moves the odds is how many people keep a tab open.
The frontier of the search for 114, on a logarithmic scale of x + y. Solid: done. Dashed: the next steps and the chance each one holds the answer.
| Step | Computer time | Chance 114 is in it | People with a tab open for a month |
|---|---|---|---|
| Double the search | 380 core-years | 4% | 2,100 |
| Ten times | 3,500 core-years | 13% | 19,500 |
| A hundred times | 40,000 core-years | 24% | 220,000 |
Measured on 20 September 2026 with the published code of Booker and Sutherland: one CPU core checks 88 million candidates per second; a gaming graphics card, in the part of the search it can do, 50 billion. A core-year is one core working for a year; "people" assumes a laptop lending four cores. Whether 114 turns up is luck. Whether the frontier moves is not: every finished step is a new record, published here with the counts to check it.
Token
The search has a coin. It is launching on pump.fun, on Solana, and this is the only place its address will be published. Anything else using this name is not it.
Where the money goes. Trading fees on $114 go to the creator. They pay for this site, the servers that hand out the slices, and the time to run it all.
You do not need the coin to compute. Anyone can press Start. The coin is how you back the hunt.
Nothing here is financial advice. The coin can go to zero. The mathematics does not care.
Sources
Not yet
This is a preview to hear what you think before we open it. When it goes live, this button hands your browser its first slice of the search.