A fresh node found its first peers in a controlled bootstrap lab. The result came with clear limits.
In The Hard Part Begins, we described a gap that a working network stack could not solve on its own: a fresh node still needed somewhere to begin. The local baseline reported NO_BOOTSTRAP_KNOWLEDGE, even though connection handling, authentication and diagnostics already existed.
The next documented step was a controlled bootstrap experiment. The result recorded on September 21, 2026 was PASS_WITH_LIMITATIONS.
Giving the node a starting point
For this experiment, the node received a lab-scoped bootstrap descriptor: a bounded description of where an initial peer could be found, together with identity constraints. That supplied the first-contact knowledge missing from the empty baseline.
This was an explicit laboratory input. The experiment did not show that a publicly distributed installation could already discover production infrastructure without configuration.
The descriptor also did not create a trusted connection by itself. Candidates still had to pass through the existing connection pipeline and identity-bearing hello. Discovery supplied a place to try; the connection checks determined whether entry was accepted.
What the lab recorded
The acceptance summary reported six observations, two authenticated entries and two independent sources. It also recorded two learned peers and one learned provider.
That moved the experiment beyond simply reading a descriptor. The fresh node acquired additional network knowledge, and the controlled run observed source failure and recovery with alternative sources available. A descriptor for the wrong chain was rejected.
The useful result was a sequence: initial knowledge, accepted entry, additional knowledge and controlled recovery.
The acceptance validator evaluated the captured observations offline. It did not create connectivity or perform the network experiment itself. The dated result document records the physical local/LAN run; it is the evidence for the outcome described here.
Why the limitations remain
The lab used controlled infrastructure and lab descriptors. Production bootstrap infrastructure was still pending, and the result retained limitations around production-grade cache and restart evidence.
Learned knowledge also has to remain separate from current trust. A remembered peer can become a candidate after restart, but persistence must not silently turn it into an authenticated or connected peer. That boundary matters when addresses become stale or machines disappear.
The integrated Internet zero-configuration acceptance level was explicitly outside this result. Reaching it requires evidence from independent network conditions, not just a successful controlled setup.
The next boundary
This experiment gave us a concrete answer to the first laboratory question: with valid starting knowledge, could a fresh node enter, learn and recover through the intended machinery? The recorded result supports that bounded claim.
The next question concerned the path itself. A node might know where to connect while still being unable to reach that address from another network. That became the focus of the Internet-edge test.
Built in Public. Documented as we build.
Top comments (0)