Political dynamics increasingly play out in observable digital spaces: a politician posts about migration and a polarised issue public mobilises within hours; a far-right account amplifies a single story and it trends across ten countries; trust in institutions erodes — and the signal appears in search behaviour and platform participation long before it shows up in surveys. An observatory for social-media-based political communication could detect such shifts earlier, at finer temporal resolution, and across more political systems than traditional polling or media monitoring.

The challenge is that social media traces — posts, replies, stances, engagement patterns — were not designed as scientific measures of political opinion or democratic health. The real task is not just collecting data, but constructing valid measures of concepts like congruence, radicalisation, issue ownership, and democratic erosion from noisy, platform-shaped signals.

TRACE would be a scientific instrument that combines social media content and interaction data, party manifesto corpora, web search and participation platform traces, cross-national survey benchmarks, and experimental or observational data on recommendation algorithm designs and their effects on political attitude formation — to detect and interpret shifts in political attitudes and elite-mass dynamics across cultural issues, platforms, and democratic systems.

Building TRACE raises fundamental challenges of measurement, integration, and validation: how concepts like political congruence, issue salience, or radicalisation can be operationalised from platform-native data; how signals can be aligned across languages, countries, and platform architectures; and how automated classifications can be grounded in human annotation, survey validation, and domain expertise rather than only in model outputs. A second set of challenges concerns observability and governance: social media environments are structurally opaque, algorithmically shaped, and unevenly accessible — meaning that what the observatory can see is itself a research problem, and that the design choices of platform infrastructures are not merely context, but object of study.

This framing is intentionally open. A concrete entry point for collaboration are datasets and pipelines already being developed within the partner institutions, including: (i) a corpus of 65 million tweets from 10 EU countries with automated stance classification on migration, climate change, and gender equality; (ii) BERT-based ideology estimation for elected representatives and mass accounts; (iii) the Euromanifestos dataset, extended via automated classification into the digital era; (iv) search and participation platform traces as indicators of political engagement and trust; (v) raw observational data on social media recommendation dynamics, as a basis for an initial study of how algorithm design affects political attitudes (relatedly: Piccardi et al. 2025, Science).

In this workshop, we aim to identify researchers interested in contributing to one or more of these challenges — whether from computational methods, political science, communication studies, psychology, or adjacent fields. We want to map who is working on what, and identify 1–2 concrete starting points for joint work.

Looking ahead, TRACE could serve as a core shared research infrastructure in a future collaborative research programme, providing a longitudinal, cross-national platform for the computational study of political communication and democratic resilience across the partner institutions.

Many socially important changes first appear in places: a strike reshapes movement, a heatwave alters public-space use, a port slowdown signals wider disruption. An observatory for place-based change could detect such shifts earlier and at finer spatial resolution than traditional statistics. The challenge is that most digital and geospatial traces were not designed as scientific measures of social phenomena. The real task is therefore not just sensing the world, but turning raw traces into valid social-science constructs — a point underscored by past big-data failures when proxies drifted or lost validity.

SCOPE would be a scientific instrument that combines satellite imagery, mobility and POI activity, event and news streams, and other georeferenced data to detect meaningful changes in places and interpret them through social-science concepts such as accessibility, resilience, exclusion, disruption, and recovery.

Building SCOPE raises fundamental challenges of measurement, integration, and validation: how concepts like disruption or exclusion can be inferred from geospatial and behavioral traces; how heterogeneous signals can be aligned across space, time, and uncertainty; and how the observatory can be grounded in surveys, interviews, administrative records, and local knowledge — not only in machine-generated data. A second set of challenges concerns explanation and governance: how detected changes can be linked to plausible causal mechanisms, and whose perspectives, privacy, and forms of representation shape what the observatory makes visible — and what it does not.

The abstract above is intentionally open. To start grassroots collaboration we can also start by focusing on existing datasets/pipelines that relate to (i) hourly POI-level activity/mobility levels, (ii) Google Maps reviews (including information about where else people leave reviews), (iii) daily low-res (3m resolution) satellite images, (iv) ground-temperature or vegetation health inferred from satellite imagery.

In this workshop, we want to identify who is working on what — and find 1–2 concrete starting points for joint work as quickly as possible. Existing datasets are a natural entry point: others might have ideas for what to do with Google Maps reviews and reviewer locations, with live POI-level busyness data, or with a dataset of AI bots. These are all low-hanging fruit that could lead to a submittable paper within six months.

Looking ahead, SCOPE could serve as one of the core shared infrastructures in a future collaborative research programme: a longitudinal, multi-signal platform for place-based social observation across the partner institutions.