DiscourseLab · Climate-denial network analysis
Map the networkof known denial actors.
Echo Chambers builds the interaction network among the climate-denial actors we track — replies, mentions, reposts, and shared links — so you can see how they amplify each other, and drill into any single actor.
Interaction layers
Replies
Who replies to whom — the @handle a reply opens with.
Mentions
Any actor named with an @handle inside a post.
Reposts
RT-style amplification of another actor's post.
Shared links
Actors posting the same external URL — co-amplification.
Ideas & open problems
Limitations & possible directions
Core limitations
What the data can’t tell us yet — the gaps that bound today’s network.
- 01
Reshares — who amplifies whom
Reshares are the core echo-chamber signal: who passes whose message on. We have the counts — roughly 45 million across the tracked actors — but we do not know the identities of the accounts behind most of those reshares. Today, we can attribute a reshare only when the resharer is themselves a tracked actor: about 10,000 of those 45 million (around 1 in 4,500). The rest come from accounts beyond our watch-list. To see who is really amplifying these messages, we need to widen that list. That is the next step.
Snowball expansion — grow beyond the known actors
Widening the list would mean a snowball: take the accounts reposting our actors, add them to the watch-list, gather their posts and interactions too, and repeat outward — which is what would turn those anonymous reshares into named edges. It is a large effort, though: it multiplies the data enormously and needs more storage, paid platform access, a re-architected pipeline, and a privacy policy for collecting ordinary users — and of these, getting access to the data is probably the hardest challenge.
- 02
One actor across platforms
The same person’s Twitter, YouTube, and Facebook accounts currently appear as separate dots. We would merge them into a single real-world actor, ideally using Junkipedia’s own actor records.
Possible future features
Enhancements on top of the data we already have — nice-to-haves we are weighing, not commitments.
- 01
Post-level analysis — read what was actually said
The dashboard shows the network and each actor’s claim profile, but not the posts behind it. One direction would be to drill all the way down: read the actual posts that make up an edge or an actor’s output, see the specific climate-denial claim each one carries, and follow how a single claim or narrative travels across the network and over time. That turns “who amplifies whom” into “what exactly are they spreading”.
- 02
Time-series analysis — how the network changes over time
The dashboard currently shows the network for a selected date range, but not how it develops over time. Time-series analysis could track changes — the rise and fall of actors, surges in particular claims, and the formation or dissolution of communities — making it possible to spot trends and turning points rather than read a single snapshot.
- 03
Export & reporting — take the findings out of the dashboard
The tool is currently focused on on-screen exploration. Next steps could include exporting networks with underlying posts and claims (e.g. CSV or GraphML), generating brief actor or topic summaries, and producing periodic reports on emerging claims, rising actors, and network changes. This would extend its use from exploration into analysis, reporting, and ongoing monitoring.
- 04
Simple & expert modes — fit the tool to the reader
The dashboard now carries a lot of controls — raw layers, content filters, node sizing, pruning, date ranges — but most people just want the headline picture. A simple mode would default to the aggregated Amplification and Interaction measures over climate-relevant posts and tuck the rest away; an expert mode would keep every layer and filter for deeper digging. Same data, two levels of depth.
Fuller detail on each is in the project README.