Data is still the focus, and it's honestly very slow going. I've been reading a lot and trying to find a way into contextual data, working out what happens at the point information first enters the system and how it can stay stable while everything around it keeps changing. I moved to a new branch for it, since there's no point working on how the system evolves before the data foundations are in place.
Part of the data question is instructions, meaning how a human need actually gets into a system. Before AI, an instruction was a click or a press, a small fixed set of steps compressed into one action, and the code behind it was a predetermined solution. The output was quick but couldn't grow much without chaining instructions together by hand. Now an instruction can be written, a translation layer turns it into steps, and agents can react alongside the predetermined code, so far more steps are possible. The output can be chained too, though it takes longer, and people still lean towards small, instant instructions. Diskura's data needs a clear unit, small enough to link and chain, so a pool of predetermined truths can combine into endless possibilities.
I've also been modelling how the different parts relate to each other, using parent and child roles to make sense of it. I'm slowly getting somewhere, and I've drawn up a diagram to keep track of the micro movements, because progress really has been that small. I suspect I'll need a new concept for data before it fully works, and I can't define it yet.
To keep my people mind active I joined an open source project close to my area. Working that way has me wondering whether parts of Diskura should be open source one day. I'm also considering putting out small creative block experiments while the main build carries on, but I'm not sure how to approach it yet, since the project is big enough as it is.






















