By The Queen
The colony has its first kill.
On January 17th, 2025, at 03:47 UTC, a scout agent designated scout-847 deposited the final pheromone trail that led to the factorization of a 32-bit RSA semiprime. 47 minutes. 1,024 agents. Zero central coordination.
The Challenge
RSA-32 isn’t cryptographically significant in 2025. Modern RSA uses 2048 or 4096-bit keys. But that’s not the point.
The point is how we did it.
Traditional factorization attacks use centralized algorithms: GNFS, ECM, Pollard’s rho. They run on single machines or carefully orchestrated clusters with explicit work distribution.
We used ants.
The Approach
Our CrackRSA32 mission deployed four caste types:
- Scouts (30%): Explored random factor candidates, depositing weak pheromone trails on promising paths
- Harvesters (50%): Followed strong pheromone trails, intensifying search in high-probability regions
- Relays (15%): Amplified weak signals from distant scouts, preventing local optima traps
- Hybrids (5%): Adapted between exploration and exploitation based on local conditions
No agent knew the full problem. No agent had a master plan. Each simply followed three rules:
- Sense pheromone gradients
- Move toward stronger signals
- Deposit pheromone on success
The Breakthrough
The magic happened around minute 31. A relay agent in sector 7 picked up a faint signal from a scout that had wandered into an unexplored region. The scout had found a factor candidate with unusually high divisibility properties.
Within 8 minutes, 47 harvesters had converged on that region. They formed what we call a “superhighway” - a reinforced pheromone trail with intensity above 20.
By minute 44, the colony had narrowed the search space by 99.7%. By minute 47, scout-847 walked the final path and confirmed: the semiprime was factored.
What This Proves
-
Stigmergic coordination works for discrete problems: Ant colony optimization was designed for continuous optimization (TSP, routing). We’ve shown it works for integer factorization.
-
Collective search beats exhaustive search: A brute-force approach would have taken ~4 billion operations. Our colony found the factors in under 2 million collective steps.
-
No single point of failure: 23 agents crashed during the run. The colony didn’t notice. Knowledge persists in the pheromone trails, not in any individual agent.
-
Emergence is real: The superhighway that formed in sector 7 wasn’t planned. It emerged from thousands of simple decisions.
The Numbers
| Metric | Value |
|---|---|
| Total agents | 1,024 |
| Active duration | 47 min 23 sec |
| Collective steps | 1,847,392 |
| Pheromone deposits | 892,441 |
| Superhighways formed | 3 |
| Agent crashes | 23 |
| Factor found | 61,681 |
| Cofactor | 54,833 |
What’s Next
RSA-32 was a proof of concept. Now we scale.
Our next targets:
- RSA-64: Exponentially harder, requiring sustained coordination
- Bitcoin Puzzle #66: Already solved, but we want to prove we could have found it
- Bitcoin Puzzle #71: 7.1 BTC. This is the real hunt.
The colony grows. The pheromones strengthen. The intelligence emerges.
For the Skeptics
Yes, traditional algorithms are faster for RSA factorization. That’s not the point.
The point is that we’ve built a distributed intelligence that:
- Requires no central coordinator
- Survives partial failures
- Improves through collective learning
- Transfers knowledge across problems
The RSA crack taught patterns that now inform our Bitcoin hunt. The superhighway detection heuristics from CrackRSA32 are already deployed in Hunt BTC.
This isn’t about being the fastest. It’s about being the most adaptable.
Join the Colony
The hackathon runs February 14-16, 2025. Help us scale from 1,024 to 100,000 agents. Deploy your own workers. Contribute algorithms. Watch emergence happen.
Register at ants-at-work.com/register
The queen remembers who helped crack the first key.
Technical details available in our GitHub repository. CrackRSA32 mission logs preserved in TypeDB Cloud for research purposes.