Qualitative and quantitative social network analysis, taught entirely by hand. No software, no coding, no maths beyond counting, and nothing to install. Two days, six sessions, one argument: the same picture can mean completely different things.
Sign up Student brochure (PDF)
Open to all Krea students. No department restriction, no year restriction, no prerequisites. Places are limited by the number of working tables, not by selection. Faculty and staff who want a seat should write directly.
The form asks three things: your name, whether you are coming with someone you already know, and whether there is anything we should arrange for you in advance. The third question is optional and goes only to the convenor.
Social network analysis is usually taught as a software course. Participants learn to produce a picture before they can say what the picture asserts, and the picture is persuasive out of all proportion to the evidence behind it. This bootcamp inverts that. Every quantity is counted by hand on networks small enough to hold in view, and the analytic work is the argument about what should count as a node and what should count as a tie.
Nodes, ties and the adjacency matrix, built by hand. Sixteen Florentine families worked two different ways. Then your own group maps a small social world it shares, and commits to a sealed prediction about it.
The convenor turns every group's paper sheet into a laid-out diagram with degree and betweenness. That is the only time a computer touches anything all weekend.
Predictions opened against the machine-drawn version. Then the Indian Ocean as a knowledge network, Randall Collins on intellectual lineages, horizons of understanding, and the shift to symptom networks.
No laptop. There is no software in this bootcamp at all. No coding, on either day. No maths beyond counting. No reading beforehand — the reading list comes afterwards and is optional. No prior anything: we assume you have never heard the word “node”.
Every metric is counted by hand before any machine computes it. And no more than five new terms per session, all of them on a printed glossary card handed out in the first ten minutes. Nothing is taught that is not on that card.
| Time | Block | Content |
|---|---|---|
| 09:00 | Arrive | Sign in, glossary cards, groups seated. Tea available. |
| 09:30 | Session 1 | What is a network? Nodes, ties, directed and undirected, the adjacency matrix as a table of ticks. Why relational data cannot be anonymised the usual way. |
| 11:00 | Tea | |
| 11:15 | Session 2 | Sixteen Florentine families, worked two ways on chart paper. Degree by counting. The loudest session of the weekend. |
| 13:00 | Lunch | |
| 14:00 | Session 3 | Your own network. Each group maps a small world it shares, pseudonymised at source, then counts and seals a written prediction. |
| 16:00 | Wrap-up | Glossary review; sheets collected; pseudonym keys destroyed in front of the room. |
| Time | Block | Content |
|---|---|---|
| 09:30 | Session 4 | Every group's network machine-drawn. Sealed predictions opened first. Bridges, brokers, structural holes and cut vertices, defined once they have seen instances of them. |
| 11:00 | Tea | |
| 11:15 | Session 5 | Fourteen Indian Ocean ports. Remove one and the system breaks — which one? Then Randall Collins on teacher–student chains and the law of small numbers. |
| 13:00 | Lunch | |
| 14:00 | Session 6 | Horizons of understanding, network closure, and the shift to a third kind of node. Guided demonstration on published symptom networks, with bootstrapped confidence intervals beside the pretty graph. |
| 16:00 | Close | The one-page memo, written in the room. Volunteers present. What we did not teach, named as a door rather than a gap. |
Three datasets, chosen so the same counting skill lands on radically different kinds of object. Everything below is worked on paper before anyone sees a computed version.
We assign the groups. Three or four people, the same group both days, seated at registration. Tell us on the form if you are coming with someone you know.
You never have to be one of the dots. On Saturday afternoon your group maps a social world. If you would rather not be in it, say so and you still do every other part; this is announced to the whole room so declining is not conspicuous.
The questions are deliberately dull. Only who has studied or worked with whom. Nothing about liking, conflict, money or health. People outside the room appear as A1, A2, A3; real names never leave your table, and the key sheet is destroyed before anything is handed in. Submitted data is deleted within seven days.
Session 6 is announced before it starts. It uses published research on mental health, including personality disorders, as an example of the third kind of edge. Nobody rates themselves or anyone else on anything, and you are free to step out for that part.
Assistant Professor, SIAS; Principal Investigator, Computational Cognition Laboratory. Works on network topology in misinformation spread and on complexity accounts of cognition. Teaches all six sessions and runs the overnight processing.
Floor support across all hand-work sessions, keeper of the glossary board, and timekeeper. Author of the lab's computational work on temporal narrative drift in echo chambers, which is where the bootcamp's interest in network topology as a cause rather than a description comes from.
Questions about what a session involves, whether you will be made to speak, whether you can bring a problem from your own discipline, or anything you would like arranged in advance: write to the convenor. Asking before you decide is completely normal and gets a straight answer.