v2.0 Planning

Discussion from early 2012, a bit outdated now

Just pasting in some of the discussion that's started around this. Will organize into feature sections on an ongoing basis.

Storylines: I like this idea a lot. I think being able to see individual narratives will help both readers and analysts. "Connecting the dots" in a very literal sense. One functionality that I think should be included is the ability to simultaneously view multiple storylines. That would help us to visualize the way different individualized timelines interact with each other, if at all--see where they intersect, find commonalities, recognize similar patterns, etc. One question I had about the storylines: I think it was Taneem who suggested tagging storylines, which I think is a good idea. Would those tags then be included in all of the datapoints that comprise that storyline? For example, if there were datapoints for an attack, an arrest, and bail all included in a storyline, would all three include a tag for "arrest?" Also, it would be interesting to include this as one of a few filters, e.g. have an option for "Show all datapoints," and one for "Show only storylines," something like that. Just another way users can drill down into the data.

Search within Threatwiki: I think Troy mentioned something akin to a "CTRL + F," and I think that'd be great. Something like that would allow us to add keywords but perhaps not include them as tags--something that could make the Correlation visualization a little crowded. Maybe this fits in a little with Chris' thoughts on a more hierarchical structure of tagging. Perhaps some will be official tags, which will be categorized, but we could add other keywords that would show up in search results. The search function might also search the item's title and description as well.

Positive vs Negative events: I really enjoyed the discussion on acceleration vs. deceleration. I started out thinking it'd be great to have a way for people to easily see developments that favor peace and those that do not. The more I thought about it, though, the more difficult I thought it would be to do right, so as of now I'm on the fence. Some concerns: I worry about unforeseen consequences, not of our coding but of the events themselves. The example I was thinking about was this: Say there were a law passed that reversed policy on Baha'i education in Iran and permitted them to openly attend institutions of higher education. This would seem to be a very positive step--it moves toward integration, is a step away from dehumanization, etc. But if that policy causes resentment against Baha'is in particular communities, if some individuals or groups feel that they are receiving special treatment, or if they simply don't want to be seen on equal footing with them, that could be a net negative. I suppose in a situation like that the backlash would be an event itself that we could code, but we want this to be a predictive analytical framework--we want to see it coming before it happens. Would it be a binary analysis? Would all events have to be either good or bad? I know we talked about a scale, which I like better because it's more nuanced, but it does risk being somewhat arbitrary as well. There are ways to mitigate that, though--standardized coding procedures, blind multi-analyst coding, etc. Perhaps scale addresses my concerns about complexity, but then would we have to provide detailed analysis on each datapoint with unclear implications? I know we can get into lengthy analyses in the periodic threat assessments, but I agree that it would be nice if there were a way to easily visualize these points on the map.