The drama of language and technology

Humanity’s journey from the trees, tundras, jungles and deserts, to cities, parallel technological worlds and geopolitical complexities is something we are prone to taking light. This might be attributed to us living it every single day. Most of us are anything but observers - we’re in the meat of it, building and driving the machinery of modern complexity. I’d like to invite you, for the duration of this post, to take a step back with me, and try becoming the observer.

To properly frame our observation, we’ll need to adjust some terminology, specifically the concept of technology. What is it, what’s its use, and how do we interact with it? I’d like to suggest a broader framing for this term: a development beyond the basis of what we might consider “archaic nature”. This expanded framing reveals an interesting angle: the method of making fire by striking one stone at another, or rubbing two pieces of wood; the discovery that planting a seed will produce a living plant in that same spot moments later; the arrangement of material to construct a shelter; yes, even the organisation of sound to produce language! All of this can be considered some form of early technology.

We have of course moved well past the basics since we roamed around looking for berries and fresh meat, but the ground truths - the general approach to discovery, inquiry and development - remain in place. This might seem like an oddly philosophical thing to discuss on the blog page of Balladic, but as we move on, it’ll become apparent how these ideas connect, and why this back-to-basics approach to thinking is relevant.

The technology of technologies

See, the main problem we’re tackling here surrounds language. The ever-elusive descriptor, complex enough to describe pretty much anything if mastered, and approachable enough to be picked up by a child. This has proven to be a major problem for computers, the technology of technologies. These machines are fantastic at tackling hard, well-defined problems at fantastic speeds and enormous scale - but, until very recently, have fallen short of the dynamic, intricate structures that shape the basics of human interaction.

At this point, it might be becoming clearer why and how this relates to Balladic. The development of natural language processors such as Large Language Models has paved the way to completely new frontiers of organization, understanding and cooperation, as computers are now no longer limited to providing structure to carry our messages - they read them; write them; parse them and connect them.

Where information thins

This is the development Balladic seeks to exploit, to pave the way for a new way of implementing machines in our work. We’ve so far discovered that it is possible to synthesize extensive insights from the data produced by work - not to be sold to the highest bidder, but to support, accelerate and ease the hard edges of communication that has always plagued any cowork involving more than one person. Our early experiments indicate that this is a more or less universal application; as soon as any sort of boundary arises, which cowork has aplenty, communication becomes difficult, and information tends to degrade as it moves across these boundaries. Be it a language boundary between a team of developers and a team of designers, between a freelancer and their clients or even between a construction worker and their leadership, the same rule applies: the point becomes difficult to get across, and the further it has to travel, the thinner it arrives at its final destination.

This is a new frontier, and we’re only getting started. As I write this, the team behind Balladic is carrying out preliminary trials on self-writing and self-reading documentation that surfaces itself when it needs to - a sort of auto-generated memory bank of vectors across an entire organization - and we have much more to come.

Pillar after pillar

If we now step back into the observer role, one thing becomes apparent: this is not a software problem. We’re facing the same problem as the hunter-gatherers we discussed in the opening; on a much grander scale, of course. This issue has been solved time and time again. Language solved it, and did a good job of this; but turned out to have all sorts of sharp edges. Writing enhanced language with permanence. Smoke signals, morse and the telephone gave it nonlocality. The email made it global. Each iteration produced a new pillar on which human communication could rest its growing weight.

And once again, our toolbox has grown. It now no longer supports, sustains or increases; it actively participates, discovers and understands. We now have the opportunity to once again extend language; this time, with abstraction and distillation.

Whether that changes everything or shifts the edges remains to be seen. We’re building as if it does, and paying close attention to the places where it doesn’t.