How to read a Stave Forecast

Around January 27th, Berlin is expected to have a significant protest- that is the message to take away from this forecast, and a low chance of something happening between now (January 12th) and then.

How do we know?

Stave has build the most advanced system of sociolographic analysis ever- but the general process for how this system generates a forecast is easy to understand. We begin by collecting open source information about a region of interest. No, not public github repos, not that kind of “open source”; here, “open source” means “not secret”. All of Stave’s intelligence products are derived from publicly available information. That is because Stave operates in a radically different fashion from every other threat analysis service. We do not track people. We don’t track any individual in a region of interest- what we do is more akin to a scientific longitudinal study of a society. Operating primarily with social media posts, we can identify a cohort in a region, and parse how they use language. While we do identify what groups are talking about, for the purposes of forecasts, what matters is how people talk about anything.

From this, we can model how conversation intensity is changing over time. Different rates and patterns in our multidimensional language space are tightly correlated with different event types in the future. These rates of correlation can be see on the Stave Toolbox Dashboard under “Statistics”.

So, to the original point - what does this curve about Berlin mean? It means that people in Berlin, or related to the Berlin-located community, are beginning to change the way they talk online, and they’re changing in a way that in the past hasforecasted a large riot. Stave follows the conversation, and gives you a heads up as to when those pattern changes occur.

You already check the weather before travel, why not check the future?