Blending word- & number-oriented docs
Our pragmatic approach to "deep learning"
In the early days of mainframe computing, Artificial Intelligence (AI) was to train computers to act and think like humans. However, available technology favored reductionist solutions. The result, as anyone trying to contact modern central authorities knows, is AI now expects humans to act and think like computers. There are high-end deep learning tools (see some here) that may swing the pendulum back, with a big IF. If they see the game as a relay race where some runners are human and some are robots. The trick is then to ensure smooth passage of the baton--defined by rules specific to a race.
Deterministic and stochastic models won't suffice. Models must be self-referencing and adaptive; with a whiff of Calvinball.
Equally important, central authorities will have to be main funders of such models, since costs will start high; yet recognize they can't own/control the team. Today's central authorities, corporations and governments, have a comparative advantage in capturing the value added such a 'smart grid' approach will generate. However, like managers of coal-fired power plants, they must see beyond the write-down of systems designed solely for centralized distribution of power.
OconEco's toolkit (M-LA, FIND, ISPP, SRoI) is designed to clarify key synapses, where humans intervention--particularly by those with local knowledge--is crucial. Whoever can recognize and value such knowledge in a 'smart grid' will have the best chance of capturing the value added to wealth creation process by such 'renewable energy'.
Mitigate lopsided IT
Central authorities' search engines are getting better at finding patterns in user preferences, particularly when backed by 'big data' syntheses of information users may not even know are out there. Most such authorities are open about their priors so, in theory, we as individuals can be selective about whose search engines we trust, how open or veiled we are about queries, etc.
In practice, lopsided IT means we face Too Much Information (TMI). As with the old crystal radio receiver, we tend to stick with whatever signals our IT manages to separate from TMI noise. OconEco's solution is a web-crawler delimited by a client's 'hits' on M-LA, FIND, etc. Once operational, mid-2020, clients will receive overnight "because you hit ... you might like" notices based on most recent and also cumulative use of our tools.
Upgrade abductive reasoning
From its beginnings in 1995, OconEco has been concerned with the IT interstice between deductive and inductive reasoning, each of which can now be handled within robust IT system. The problem is each tends to become a Sorcerer's Apprentice, able to carry buckets of conclusions to decision-making workspaces that may need them or be overflowing with similar results.
We offer operational guidelines for Virtual Workspaces that let clients inform prospective stakeholders how and why they will consider 'changing gears', given stakeholder input. The goal is then an iterative and ideally upward spiraling process, not a simple cycle.