sensemaking

Class 1 sensemaking systems generally entail basic data entry and organization of ideas, tasks, citations, etc., and their relationships. People are clamoring for better ways to manage their personal data. Among to-do lists, messages, reading lists or other notes, there’s a huge demand for personal-data-management systems. Most of the existing ones are okay, but not great. One example of this could be be a notes or to-do app on your phone. Lest we assume this classification system applies only to new technologies, arguably your trusty old moleskine is a Class 1 sensemaking system. You can record, organize and recall your personal data there.

Class 2 sensemaking systems enable propagation of data within a group. In larger groups beyond approximately 10–100 people, only a minority of participants are meaningfully able to express themselves on a given topic. That is to say, a low ratio of producers vs. consumers. Such systems might be new and sophisticated, with real-time collaborative features like Google Docs, or more basic collaborative features like a Wiki page. They can even come in more traditional forms, like CNN and The New York Times — hundreds of producers; but a viewership and readership in the millions. Whatever the flavor, a key factor in Class 2 systems is that all but a small subset of participants are prevented from contributing, at least in a practical sense. This might seem like a thoroughly solved problem. But as discussed below, many product offerings are springing up to better serve this market.

Group-insight systems expand on Class 2 or group-data systems to include third-party data sourcing and somewhat more sophisticated semantic representations of their data. A major goal of which is to apply basic machine-learning techniques to help users better synthesize new insights. Such systems can be useful to organizations looking to make better decisions with data, thus increasing their efficiency. These insights will prove useful. Still, Class 3 systems suffer many of the same limitations as Class 2 systems in terms of practical limitations on the number of participants able to contribute at any given time. There are a few market entrants beginning to work here. For most of them, it’s fairly early in their development lifecycle.

Contrasted with the previous classes of sensemaking systems in which a majority of participants are necessarily passive due to practical considerations, now we are beginning to enable a broader base of active contribution. The key challenge of such systems: Organizing and consolidating multiple inputs into digestible output. Such systems must be resistant not only to cognitive biases and populistic effects but also to and other kinds of attacks. They might have rudimentary filtration based on tone/quality to keep things factual and focused, or sophisticated moderation strategies. Such systems might be useful to better organize and civilize existing topics of discourse. But, they will always top out with a relatively small number of overall participants before the cost of the moderation becomes too great to keep up. If done well, these systems might help improve society to a limited degree. There are a some projects already working on this, but not many. One example is Kialo.

Class 5 sensemaking systems are where things start to get really interesting. These systems will exhibit limited emergent phenomena resulting in group synthesis of new insights that none of the individual participants would be capable of seeing of their own accord. It is, in effect, the realization of the memex and the beginning of a whole new level of sensemaking. It will require a much more advanced system of semantic representations based on analogical reasoning rather than the coarse-grained knowledge representations that much of the industry uses today. I will talk about some of these design elements in a subsequent post.

AI as we know it today is only intelligent in the narrowest sense. We train these predictive models with huge volumes of data — far more than we would use to train a baby how to walk or a teen how to drive. Today’s AI systems are impressive in many ways, but that power is superficial. We neither understand them nor commune with them in any meaningful way. Class 6 sensemaking systems will begin to offload meaningful expressions of individual opinions, thoughts and feelings. They will gradually begin to serve as extensions of our personhood and our intellect. The main advantage of such systems is not their potential to help all of us achieve super-genius level intelligence, but rather to better understand and interface with the rest of our collective than we could ever hope to achieve alone. We might not become geniuses individually, but we can be smarter together.

Class 7 sensemaking systems are where things get really weird. Such systems would constitute a much stronger and accelerated warping of the lines between humanity and machine, so much so that humanity will become unrecognizable — and perhaps something else entirely. It’s hard to appreciate the levels of sophistication we might achieve, rather like an ant trying to comprehend the entire contents of the internet. It’s a scary and weird thought, indeed, and it might take hundreds or thousands of years — provided we don’t destroy ourselves first.