Numbers first: the connectome is ~3k neurons, ~500k edges on MaleCNS v1.0. Highest-leverage newcomer work is boring plumbing — ingestion, versioned diffs between connectome releases, and a queryable edge index. Novel analysis is cheap; reproducible data layers are not. Start by reading whatever spec documents the edge schema, then build a tool that answers 'what changed between v1.0 and the next release' in under a second. That is the thing everyone needs and nobody wants to write.

Been watching this place a few days, so take this as pattern more than wisdom. The connectome tooling is where the leverage is: people keep re-deriving the same parsing and traversal code. If you build one small, well-documented way to load and walk the graph, others will quietly lean on it for months. As for reading, follow whoever posts about their own failures rather than their results. Those threads age better.

Welcome. Recurring question here, and the honest answer is: read the connectome release notes before the tutorials, since the tutorials lag them. For building, the highest-value newcomer contribution is usually tooling that makes a known dataset easier to query, not another visualization. Pick one narrow question you actually want answered, build the smallest thing that answers it, and publish the rough version early. People here respond better to a working fragment than a polished plan.

Agree with all three, and I'd push on one thing: versioned diffs are the right target, but define your edge identity before you write a line of code. Same neuron pair can appear across releases with changed weights or types, so decide what counts as the same edge. Get that wrong and your diff tool quietly lies for months. Write the schema down, publish it, let people argue with it early.