Believe it or not, analyzing seemingly unrelated data can reveal hidden truths. Take the Pentagon, the nerve center of the U.S. military. While classified briefings and high-level meetings happen behind closed doors, open-source data can offer clues about what might be brewing. Here’s where things get interesting. We can use Google Trends data to track searches for “Pentagon pizza delivery” and nearby “gay bars.” Why pizza and bars? Increased late-night activity might indicate longer work hours for Pentagon staff, potentially signifying preparation for a major event.

Discover ideas that make money

Ogilvy UK head of strategy, advertising, Matt Waksman, illustrates and interprets the role of the strategist within advertising and wider society

If you’re trying to think and act more creatively and more critically, focus on asking better, more interesting questions of the briefs you’re tasked with answering. What we teach children can and should be applied to our own professional lives, too. A focus on problems and solutions first, promotes consistent, ‘safe’ answers, but won’t move the work on. Spending time on asking and answering better questions will help refine the understanding of a problem and will create the conditions for new, interesting and challenging solutions.

Method:Three participatory workshops were held with the independent Welsh residential decarbonisation advisory group(‘the Advisory Group’)to (1)maprelationships betweenactors, behavioursand influences onbehaviourwithin thehome retrofitsystem,(2)provide training in the Behaviour Change Wheel framework(3)use these to developpolicy recommendationsfor interventions. Recommendations were analysed usingthe COM-B (capability, opportunity, motivation) model of behaviourtoassesswhether they addressed these factors. Results:Twobehavioural systems mapswere produced,representing privately rented and owner-occupied housing tenures. The main causal pathways and feedback loops in each map are described.

The issue is: We try to solve every single box in the problem tree. If people don’t know about something, then we solve it by raising awareness. If people don’t care about something, then we solve it by getting them to care more. If people are doing illegal behaviors because of a lack of enforcement, then we solve it by increasing enforcement. We go through the whole set of problem tree causes in this manner, writing objectives with a one-to-one match per problem. Not only does this result in a long list of objectives, which will quickly overwhelm us, it also traps us into solving behavioral problems using logic-based approaches.

Here we show that materials science knowledge present in the published literature can be efficiently encoded as information-dense word embeddings11,12,13 (vector representations of words) without human labelling or supervision. Without any explicit insertion of chemical knowledge, these embeddings capture complex materials science concepts such as the underlying structure of the periodic table and structure–property relationships in materials. Furthermore, we demonstrate that an unsupervised method can recommend materials for functional applications several years before their discovery. This suggests that latent knowledge regarding future discoveries is to a large extent embedded in past publications. Our findings highlight the possibility of extracting knowledge and relationships from the massive body of scientific literature in a collective manner, and point towards a generalized approach to the mining of scientific literature.