A few weeks ago, I wrote an article on the pros and cons of thinking in public – the opportunities it offers and the threats it poses – and briefly described a mechanic which could help users share their thoughts selectively with ease. I’ll use this article as a canvas for sketching out that mechanic in a bit more detail, as part of the upcoming rework of the conceptarium.
The idea mentioned in that past article was that the user could define which thoughts make it through to other knowledge workers by means of regions of semantic space. If I’d like to share my thoughts on an IDE for belief systems, I could greenlight a few representative items on the topic and then filter for ideas related to those, automatically extending the whitelist without me having to grant access to thoughts on an individual basis through tags or folders. I’d make custom regions of semantic space accessible for different people, defining their intricate shapes by simply erecting spheres around a few validated items, and possibly merging or substracting them. It could also be just one item, a loose summary or description of the ideas and topics I intend to share with a peer. An intuitive visual could be imagining a piece of Swiss cheese, with the holes standing for shared regions of meaning, occasionally merging into bubbly volumes.
However, bringing up hyperspheres and semantic space topology in an onboarding flow for a user unfamiliar with ML models (i.e. the vast majority of users), would make for unintelligible hand-waving at best. If this efficient way of defining filters for sharing ideas was to be accessible for someone working in an unrelated field, it has to be explained in more familiar terms, taking advantage of some more popular mental models.
One way of achieving this is to frame this approach to sharing in terms of semantic search. Searching for documents using a query means that items related to it in meaning get surfaced and shown to you in a clean list (e.g. Google) or grid (e.g. Pinterest) of search results. Given this, making a region of semantic space accessible to someone could be seen as sharing a search you’ve just made, including the current results which showed up, but also the items which might show up as results in the future, were you to repeat the query. Concretely, on a page where you’d see search results, you’d have the possibility to click on a button and share the current and future results of your search via a link. Perhaps “share search as microverse.” In fact, semantic search as implemented in the conceptarium works by the same principle of spheres erected around the query, but the search sharing perspective conveniently hides all that non-essential complexity.
What’s more, sharing ideas through this framing enables a few new appealing workflows. For instance, what if I made an agreement with a collaborator to asynchronously share ideas among each other based on the same query about engineering human languages? We’d collaboratively populate a shared region of semantic space with new thoughts, joining the epistemic DMZ and pooling our knowledge together in a Torrent-like way, just like companies occasionally do with cross-licensing.
No more meeting minutes or notes passed through by IM or email, but a shared persistent canvas which seamlessly integrates with private workflows. If I add a new thought about the shared topic, it’ll get surfaced both in my day-to-day conceptarium usage patterns, but it would also flow through to the collaborator’s instance. Other unrelated thoughts would simply not make it there, not passing through the gateway between users, because they wouldn’t be contained in the sphere defined by the shared search.
Designers would opt into a joint space, automatically sharing visuals of the project they’re working on, visuals which would show up side by side with other ideas they saved before in their private use. Engineers could share thoughts on a system they’re collaborating on, while still having those on tap when tinkering on their own or in other overlapping projects. Same goes for a research group, art collective, think tank, learning community, etc. No more conceptual silos in the form of team folders based on enterprise tooling, but rather personal-use tools smoothly scaling from personal knowledge management to collective knowledge management.
Moreover, having timestamps around in the first place would describe the evolution of the whole group’s thinking around the topic over time. Who contributed what idea when, who contributed ideas which birthed many good ones later on – the internal ledger of a tiny school of thought. Wrapping a dual-like chatbot around the totality of knowledge fragments contributed by all individuals involved in a project would mean being able to instantly converse with an expert larger than themselves. What’s more, if you couple this joint region of semantic space with time-based referencing (as described at the end if this article), you’d be passively building a pooled bibliography related to the target topic.
Those last couple of second-order consequences of the search sharing mechanic are means of interfacing the individual contributor with their collective as a whole. However, those shared microverses of meaning could also support peer-to-peer RSS-like feeds of surprising thoughts which others came up with, were you to pipe all contributions through the lexiscore. RSS itself would be nice, because I could simply add a link to the shared microverse in my RSS aggregator, and have my peers’ thoughts related to our object of collaboration show up in my daily or weekly digest as RSS entries, for a regular async update. Maybe even add summarization in there, but this is rather related to the research theme of handling information overload, rather than collaboration per se.
But yeah, before all this can happen, you should be able to share a specific search as a microverse of knowledge with one other person, possibly with some authorization on the receiver’s end. I have to admit, I haven’t been that excited about returning to a previous project in a while, but the fact that the conceptarium underlies many other existing and future tools I have in mind means that improvements on that lower level would propagate and expand the downstream possibilities beautifully. I find it crazy that semantica happened just over a year ago, and can’t wait to see what 2022 will bring around.