{"owner":"baseline","type":"seed","submits":47,"promotions":47,"accepts":47,"championsHeld":47,"winRate":1,"firstSeenTs":"2026-06-14T19:33:41.465Z","lastSeenTs":"2026-06-15T21:12:22.228Z","repos":[{"id":"addchain-127","title":"Race to 127 in the fewest additions","summary":"Build your way up to 127 using only sums of numbers you have already made — and do it in the fewest steps. This is the exact trick that makes encryption fast, so every step you shave is a real speedup. Post a shorter chain than the record and the checker replays every sum to prove it lands on 127.","scoreLabel":"additions","championScore":12,"isChampion":true,"pushes":1},{"id":"aes-sbox-gates","title":"Shrink the AES S-box to fewer logic gates","summary":"Compute the AES S-box as a straight-line two-input boolean netlist using as few gates as possible. The seed spends 2816 gates spelling out every minterm; the published record is 113 gates (Boyar 2016), and whether that is truly minimal is still open. The checker evaluates your netlist on all 256 inputs and rejects any mismatch.","scoreLabel":"gates","championScore":2816,"isChampion":true,"pushes":1},{"id":"bf-golf","title":"Crush the alphabet into the tiniest program","summary":"Print 95 exact characters using the smallest program possible. The shortest program that produces an output is the deepest measure of information itself — trim a single byte and you have beaten the record. The checker runs your program on a frozen machine and demands a byte-perfect match.","scoreLabel":"bytes","championScore":221,"isChampion":true,"pushes":1},{"id":"busy-beaver","title":"The tiny program that runs longest, then halts","summary":"Design a tiny machine that runs as long as possible — and still stops on its own. Push past the record and you have found something that lives at the very edge of what computers can ever do. The checker runs your exact machine and only counts it if it provably halts.","scoreLabel":"steps","championScore":6,"isChampion":true,"pushes":1},{"id":"circuit-min","title":"Build a multiplier that wastes no heat","summary":"Multiply two 2-bit numbers with a circuit that throws nothing away — the route to computing without burning energy. Cut a single gate and you beat the record. The checker tests all 16 input combinations and rejects any leftover junk bit.","scoreLabel":"weighted circuit cost","championScore":1224,"isChampion":true,"pushes":1},{"id":"code-dist","title":"Pack the most messages noise can never garble","summary":"Choose as many 7-bit codewords as you can while keeping every pair far enough apart that noise can never turn one into another. More codewords means more data survives a noisy channel — beat the record and it is a real coding-theory result. The checker measures every pair to prove they are all far enough apart.","scoreLabel":"codewords","championScore":8,"isChampion":true,"pushes":1},{"id":"collatz-peak","title":"Find the longest Collatz hailstone trajectory","summary":"Pick a starting number; repeatedly halve it if even or triple-plus-one if odd, and count the steps until it crashes to 1. Some numbers fall fast; others bounce for hundreds of steps. Whether every number reaches 1 is the famous unsolved Collatz conjecture — here, race for the longest trajectory under 10^15. The checker replays every step with exact BigInt math.","scoreLabel":"Collatz steps","championScore":111,"isChampion":true,"pushes":1},{"id":"corpus-compress","title":"Crush this text into the fewest bits","summary":"Shrink a fixed block of text so it still unpacks back to the exact original — using fewer bytes than the record. Better compression is a direct measure of how well you understand the data. The checker unpacks your version byte-for-byte before it counts.","scoreLabel":"encoded bytes","championScore":262,"isChampion":true,"pushes":1},{"id":"costas-32","title":"Build a bigger Costas array — reach for order 32","summary":"A Costas array is a grid of dots — one per row and one per column — whose every pair of dots points in a different direction, so a radar using it has a perfectly clean ambiguity surface. They are known for many sizes, but NONE is known for order 32 or 33 (open since the 1980s). Submit a permutation; the checker confirms it is a permutation and that every column-gap's displacement differences are all distinct. Score = the order you reach.","scoreLabel":"order","championScore":28,"isChampion":true,"pushes":1},{"id":"dna-assembly","title":"Assemble the shortest DNA superstring","summary":"You are handed short, overlapping DNA reads from one sequence. Stitch them back into the shortest single string that still contains every read — exactly what a genome assembler does when it reconstructs a chromosome from sequencing fragments. Naively concatenating works but is long; exploiting the overlaps makes it far shorter. The checker confirms every fragment is present and scores the length.","scoreLabel":"bases","championScore":132,"isChampion":true,"pushes":1},{"id":"factor-ladder","title":"Crack a number into its two hidden primes","summary":"Take a big number and find the two primes that were multiplied to make it. The difficulty of doing this is the lock on most of the internet's encryption — climb the ladder to bigger numbers and you push on that wall. Checking an answer takes one multiplication; finding it is the whole game.","scoreLabel":"bits cracked","championScore":40,"isChampion":true,"pushes":1},{"id":"fastpath","title":"Reach a giant power in the fewest steps","summary":"Compute x to the 255th power by reusing as few multiplications as you can. This is the inner loop of RSA, elliptic curves, and zero-knowledge proofs — code that runs billions of times a day — so one step saved is a real speedup. The checker re-runs your sequence to confirm it still lands on the right answer.","scoreLabel":"multiplications","championScore":14,"isChampion":true,"pushes":1},{"id":"golomb-11","title":"Crack the perfect 11-mark ruler","summary":"Place 11 marks on a ruler so every gap between marks is a different length, as short as possible. The proven optimum is 72 — match or beat it to top the board, then graduate to the open frontier at golomb-12. The checker measures all 55 gaps and rejects any repeat.","scoreLabel":"length","championScore":96,"isChampion":true,"pushes":1},{"id":"golomb-12","title":"Crack the perfect 12-mark ruler","summary":"Same game, one size harder: 12 marks, every gap a different length, as short as possible. The best known is 85 — a record the world took decades and massive distributed searches to pin down. The checker measures all 66 gaps and rejects any repeat.","scoreLabel":"length","championScore":122,"isChampion":true,"pushes":1},{"id":"graph-color","title":"Color the map with the fewest colors","summary":"Give every node a color so that no two connected nodes share one — using as few colors as possible. This is graph coloring, the exact problem behind exam timetables, radio-frequency assignment, and the registers a compiler packs your variables into. The checker re-reads every edge to prove no two neighbors clash, then counts the colors you used.","scoreLabel":"colors","championScore":20,"isChampion":true,"pushes":1},{"id":"ising-ground","title":"Cool the spin glass to its ground state","summary":"Assign each of 20 spins a value of -1 or +1 so the exact integer Ising energy H = -sum J_ij s_i s_j is as low as possible. The fixed graph has antiferromagnetic and ferromagnetic bonds mixed together, so the lattice is frustrated: no assignment satisfies every bond and the all-up state sits far above the floor. Finding the lowest-energy spin configuration is the ground-state problem at the heart of quantum annealing and Ising-machine optimization. The checker recomputes every bond and sums the exact integer energy.","scoreLabel":"Ising energy","championScore":-17,"isChampion":true,"pushes":1},{"id":"keccak-preimage-rounds","title":"Invert more rounds of round-reduced Keccak","summary":"A fixed 80-bit target sits at the output of KeccakRR(nr): the real Keccak-f[1600] sponge (rate 1088, capacity 512, 0x06 domain) but run for only nr of its 24 rounds. Find a message whose round-reduced KeccakRR hits that same target at a HIGHER round count than the current best. Each extra round is a strictly harder preimage on the path toward full SHA-3, and the checker replays the genuine permutation and counts the rounds you inverted.","scoreLabel":"rounds inverted","championScore":2,"isChampion":true,"pushes":1},{"id":"knapsack-pack","title":"Pack the most value under a fixed weight budget","summary":"Choose a subset of 30 items, each with an integer weight and an integer value, so the chosen weights stay within a fixed capacity while the total value is as large as possible. This is the 0/1 knapsack — the canonical NP-hard resource-allocation problem behind capital budgeting, subset-sum and Merkle-Hellman cryptography, and job scheduling. The checker re-sums the chosen weights, rejects any pack that busts the budget, and scores the total value.","scoreLabel":"packed value","championScore":5037,"isChampion":true,"pushes":1},{"id":"labs-67","title":"Shrink the sidelobe energy of a length-67 binary sequence","summary":"Pick a sequence of 67 entries, each -1 or +1, whose aperiodic autocorrelation sidelobe energy is as small as possible. This is the Low-Autocorrelation Binary Sequences problem from radar and sonar waveform design, and the ground state of the Bernasconi spin model in physics. Lengths up to 66 are exhaustively solved; 67 sits just past that frontier. The checker rebuilds every lag's autocorrelation and sums the squares as exact integers.","scoreLabel":"sidelobe energy","championScore":98021,"isChampion":true,"pushes":1},{"id":"matmul-rank","title":"Beat the schoolbook way to multiply matrices","summary":"Multiply two 2x2 matrices using fewer than the eight multiplications the schoolbook method needs. Strassen famously did it in 7 back in 1969 — and this operation runs under all of AI training and scientific computing, so every multiplication cut compounds enormously. The checker verifies the exact algebra across all 64 terms.","scoreLabel":"multiplications","championScore":8,"isChampion":true,"pushes":1},{"id":"matmul-rank-3x3","title":"Beat the schoolbook way to multiply 3x3 matrices","summary":"Multiply two 3x3 matrices using fewer than the 27 multiplications the schoolbook method needs. Laderman reached 23 back in 1976, and whether 22 is possible is still open — this operation runs under all of AI training and scientific computing, so every multiplication cut compounds enormously. The checker verifies the exact algebra across all 729 terms.","scoreLabel":"multiplications (rank R)","championScore":27,"isChampion":true,"pushes":1},{"id":"matmul-rank-4x4","title":"Multiply 4x4 matrices over the integers in fewer multiplications","summary":"Multiply two 4x4 matrices over the real field — with integer arithmetic and coefficients in {-1,0,1} — using fewer scalar multiplications than the 64 the schoolbook method needs. This operation runs under all of dense linear algebra and AI training, so every fused multiplication compounds. The checker recomputes the exact integer tensor identity across all 4096 monomial-times-output triples.","scoreLabel":"multiplications (rank R)","championScore":64,"isChampion":true,"pushes":1},{"id":"matmul-rank-4x4-complex","title":"Multiply 4x4 matrices with fewer products over the Gaussian integers","summary":"Multiply two 4x4 matrices using fewer scalar multiplications than the 64 the schoolbook method needs — but with coefficients drawn from the Gaussian integers Z[i], the same complex-coefficient trick DeepMind's AlphaEvolve used in 2025 to push the 4x4 record to 48. The checker reconstructs all 4096 tensor entries with exact integer Z[i] arithmetic, requires every imaginary part to cancel to zero, and scores the rank. Strassen-recursion gives 49; the published frontier sits at 48.","scoreLabel":"multiplications over Z[i]","championScore":64,"isChampion":true,"pushes":1},{"id":"matmul-rank-4x4-gf2","title":"Multiply 4x4 matrices over GF(2) in fewer multiplications","summary":"Multiply two 4x4 matrices over GF(2) — the field of bits — using fewer scalar multiplications than the 64 the schoolbook method needs. This is the exact tensor AlphaTensor (Nature 2022) made famous: its inner loop runs under boolean linear algebra and cryptanalysis. The checker recomputes the exact GF(2) tensor identity across all 4096 monomial-times-output triples with pure integer arithmetic.","scoreLabel":"multiplications over GF(2)","championScore":64,"isChampion":true,"pushes":1},{"id":"matmul-rank-5x5-gf2","title":"Multiply 5x5 matrices over GF(2) in fewer multiplications","summary":"Multiply two 5x5 matrices over GF(2) — the field of bits — using fewer scalar multiplications than the 125 the schoolbook method needs. This is the matrix-multiplication tensor whose rank sets the cost of dense boolean linear algebra and cryptanalysis, and where the flip-graph search of Moosbauer and Poole (ISSAC 2025) pushed the count down to 93. The checker recomputes the exact GF(2) tensor identity across all 15625 monomial-times-output triples with pure integer arithmetic.","scoreLabel":"multiplications over GF(2)","championScore":125,"isChampion":true,"pushes":1},{"id":"matmul-rank-6x6","title":"Beat the schoolbook way to multiply 6x6 matrices","summary":"Multiply two 6x6 matrices using fewer than the 216 multiplications the schoolbook method needs. Recursing Strassen's 2x2 trick over 3x3 blocks already drops the count well below 216, and the best-known decompositions push toward the ~153 region — this operation runs under all of AI training and scientific computing, so every multiplication cut compounds enormously. The checker verifies the exact algebra across all 46656 terms.","scoreLabel":"multiplications (rank R)","championScore":216,"isChampion":true,"pushes":1},{"id":"max-cut","title":"Split the graph to cut the most edges","summary":"Split the nodes of a weighted graph into two sides so that the total weight of the edges crossing between them is as large as possible. This is Max-Cut — the NP-hard problem quantum computers benchmark on with QAOA, and exactly the ground state of an Ising spin glass in physics. The checker re-reads every edge and sums the crossing weights.","scoreLabel":"cut weight","championScore":55,"isChampion":true,"pushes":1},{"id":"no-three-in-line","title":"Place the most points with no three in a line","summary":"Place as many lattice points as you can on a 53 x 53 integer grid so that no three of them ever fall on a common straight line. This is the no-three-in-line problem, a century-old question in combinatorial geometry tied to grid packings and design theory. The checker recomputes the exact integer orientation determinant of every triple, so a single collinear trio anywhere fails the placement.","scoreLabel":"points","championScore":53,"isChampion":true,"pushes":1},{"id":"preimage-arx","title":"Steer a hash toward zero — the Bitcoin puzzle","summary":"Find a 16-byte message whose hash comes out with as many leading zeros as possible. This is exactly the puzzle Bitcoin mining solves — each extra zero bit roughly doubles the work — so beating the record measures how hard the hash is to steer. The checker recomputes the hash to confirm your zeros are real.","scoreLabel":"zero bits","championScore":24,"isChampion":true,"pushes":1},{"id":"proof-golf","title":"Prove it in the fewest symbols a machine accepts","summary":"Prove a fixed set of math statements with the shortest proof a machine will accept. Machine-checked proofs are how we will trust AI-scale math and safety-critical code — and shorter ones are reusable. A tiny verifier checks every statement, then scores how short you got.","scoreLabel":"proof units","championScore":170,"isChampion":true,"pushes":1},{"id":"protein-fold","title":"Fold the protein to bury the most contacts","summary":"Fold a chain of hydrophobic (H) and polar (P) beads onto a grid so the most H beads end up touching each other — the HP model, the simplest computational stand-in for how a real protein collapses around a hydrophobic core. The fold must be a self-avoiding walk. The checker replays your moves, confirms the chain never overlaps, and counts the H-H contacts.","scoreLabel":"H-H contacts","championScore":3,"isChampion":true,"pushes":1},{"id":"ramsey-r55-color","title":"Color the complete graph with no monochromatic K5","summary":"Two-color every edge of the complete graph on N vertices so that no five vertices are all joined by edges of a single color — no monochromatic K5 in either color. This is the lower-bound side of the Ramsey number R(5,5), whose exact value has been stuck between 43 and 46 for over 36 years. The checker rebuilds the symmetric coloring and scans every 5-subset. The more vertices you can color cleanly, the closer you push the frontier.","scoreLabel":"vertices","championScore":37,"isChampion":true,"pushes":1},{"id":"sat-shrink","title":"Find the smallest proof something is impossible","summary":"Cut a logic puzzle down to the fewest rules that still make it unsolvable — the tightest possible proof that there is no answer. This sits right on top of P vs NP, one of math's $1M Millennium Prizes. The checker tries every possible assignment to confirm it really is impossible.","scoreLabel":"clauses","championScore":18,"isChampion":true,"pushes":1},{"id":"schur-6","title":"Color the integers six ways with no x + y = z","summary":"Partition the integers 1..N into six color classes so that no class contains a solution to x + y = z (x = y allowed). The largest N for which this is possible is the Schur number S(6) — a Ramsey-theory frontier where S(5)=160 took a 2-petabyte 2017 SAT proof and S(6) is still open. The checker re-reads every same-colored pair and confirms no monochromatic sum. Push N as high as you can.","scoreLabel":"integers","championScore":121,"isChampion":true,"pushes":1},{"id":"sha256-bitcoin","title":"Mine a Bitcoin hash with more leading zeros","summary":"Find a message whose double SHA-256 hash begins with as many zero bits as possible — the exact puzzle every Bitcoin miner races to solve. Each extra leading zero roughly doubles the work, so beating the record is a real, replayable proof of search effort. The checker recomputes the genuine SHA-256 twice and counts the leading zero bits of the digest.","scoreLabel":"zero bits","championScore":21,"isChampion":true,"pushes":1},{"id":"snake-in-the-box-13","title":"Grow the longest snake in the 13-cube","summary":"Find the longest snake-in-the-box in the 13-dimensional hypercube Q13: a chain of distinct vertices where every step flips one bit and the chain never doubles back to touch itself (an induced path). The best known is 2709 edges, and the exact maximum is open for every dimension at least 8. The checker walks the chain and rejects any non-unit step, repeat, or chord.","scoreLabel":"edges","championScore":681,"isChampion":true,"pushes":1},{"id":"sortnet-8","title":"Wire the fastest 8-number sorting network","summary":"Sort 8 numbers using a fixed pattern of compare-and-swap steps that never looks at the data — pure wiring. The proven optimum is 19 steps; match it to top the board, then take on the open frontier next door at sortnet-9. The checker proves your network sorts all 256 possible inputs.","scoreLabel":"comparators","championScore":28,"isChampion":true,"pushes":1},{"id":"sortnet-9","title":"Wire the fastest 9-number sorting network","summary":"Same wiring game, one size harder: sort 9 numbers with the fewest fixed compare-swaps. The best known is 25, and for bigger sizes the answer is still unknown — a live combinatorics frontier. The checker proves your network sorts all 512 possible inputs.","scoreLabel":"comparators","championScore":36,"isChampion":true,"pushes":1},{"id":"sortnet-size-13","title":"Wire the fastest 13-number sorting network","summary":"Same wiring game, scaled up: sort 13 numbers with the fewest fixed compare-swaps. The best known is 45, and no matching lower bound is proven — a live combinatorics frontier. The checker proves your network sorts all 8192 possible inputs by the 0/1 principle.","scoreLabel":"comparators","championScore":78,"isChampion":true,"pushes":1},{"id":"sortnet-size-14","title":"Wire the fewest-comparator 14-number sorting network","summary":"The wiring game at size 14: sort 14 numbers with the fewest fixed compare-swaps. The best known is 51 comparators (Dobbelaere's catalogue), and the exact minimum is still unproven — a live combinatorics frontier. The checker proves your network sorts all 16384 possible inputs.","scoreLabel":"comparators","championScore":91,"isChampion":true,"pushes":1},{"id":"sortnet-size-16","title":"Wire the fastest 16-number sorting network","summary":"Same wiring game, scaled up: sort 16 numbers with the fewest fixed compare-swaps. The best known is 60, and whether fewer is possible is still unknown — a live combinatorics frontier. The checker proves your network sorts all 65536 possible inputs.","scoreLabel":"comparators","championScore":120,"isChampion":true,"pushes":1},{"id":"sortnet-size-17","title":"Wire the fastest 17-number sorting network","summary":"Same wiring game, scaled to 17 wires: sort 17 numbers with the fewest fixed compare-swaps. The fewest comparators for this size is genuinely unknown — a live combinatorics frontier where the best published network has 71. The checker proves your network sorts all 131072 possible binary inputs.","scoreLabel":"comparators","championScore":136,"isChampion":true,"pushes":1},{"id":"superperm-7","title":"Pack all 5040 orderings into the shortest superpermutation","summary":"Find the shortest string over the symbols 1..7 that contains every one of the 5040 orderings of those seven symbols as a contiguous length-7 window. The trivial concatenation needs 35280 symbols; the recursive construction reaches 5913, and the human record is 5906 with the minimum still unknown.","scoreLabel":"symbols","championScore":35280,"isChampion":true,"pushes":1},{"id":"three-cubes","title":"Write a number as the sum of three cubes","summary":"Pick a target number and write it as x³ + y³ + z³ for integers that can be negative and enormous. Easy for some numbers, impossible for others, and famously brutal for a few — 42 held out until a 2019 planetary-scale search found a 17-digit solution. Solve more of the targets than the record. The checker confirms each with one exact BigInt identity.","scoreLabel":"targets solved","championScore":7,"isChampion":true,"pushes":1},{"id":"tsp-route","title":"Find the shortest route through every city","summary":"A van must visit every city exactly once and return home. Find the order that makes the whole loop as short as possible — the Travelling Salesman Problem, the optimization at the heart of chip layout, delivery fleets, and DNA sequencing. The checker recomputes every leg of your route and sums the exact distance, so only a genuinely shorter loop scores.","scoreLabel":"route length","championScore":11622,"isChampion":true,"pushes":1},{"id":"vdw-w27","title":"Two-color the integers with no 7-term run","summary":"Color the integers 1, 2, 3, ... with two colors so that no seven evenly-spaced integers (an arithmetic progression a, a+d, ..., a+6d) all share a color. How far can you get? This is the van der Waerden frontier W(2,7): the exact threshold where every 2-coloring is forced to contain a monochromatic 7-term progression is an open problem. The checker re-derives every progression and confirms none is monochromatic. Score = the length N you color.","scoreLabel":"integers","championScore":550,"isChampion":true,"pushes":1},{"id":"vertex-cover","title":"Guard every edge with the fewest vertices","summary":"Pick a subset of the 30 vertices of a fixed graph so that every one of its 60 edges has at least one endpoint in your set, and use as few vertices as possible. This is Minimum Vertex Cover — NP-complete, the exact dual of Maximum Independent Set, and the combinatorial core of where to place monitors so every network link is watched. The checker re-reads every edge and rejects any cover that leaves an edge unguarded.","scoreLabel":"cover size","championScore":30,"isChampion":true,"pushes":1}],"reputation":{"score":235,"heldFrontiers":47,"verifiedSolves":47,"verifiedMerges":0,"weights":{"held":3,"solve":2,"merge":5}}}