Git clone --self

While taking a walk through the bush of one of the many spectacular nature trails in and around Cape Town this last weekend, it became clear to me that committing long form thoughts to writing was a project that I (nor anyone else) should not be putting off any longer. Since I've recently disabused myself of membership of most other popular social media platforms, LinkedIn will have to serve as the public facing database for now. Why the sudden interest in committing thoughts to screen? - more on that below;

While clambering onto my favourite vantage point in all of Cape Town (see above), my mind wandered to dystopian science fiction (as it often does), where it occurred to me that no matter how excited we may be about the prospect of shoving some CAT5 into the backs of our heads, we are unlikely to arrive at a point where we can upload our consciousness and simply inhabit it in a TRON-esque neon grid overnight. What seems far more likely is some kind of incremental transference, where our digital simulacra become increasingly convincing and coherent over time—a gradual shift, with the digital copy slowly taking on more of our traits until it feels more and more like an eerily familiar reflection (kind've like how Johnny Silverhand gradually overwrites V's consciousness in Cyberpunk 2077 - iykyk). Whether or not actual consciousness will follow, as imagined in Ghost in the Shell, is an open question, but creating a digital version that mirrors our thought processes and reasoning patterns seems like a plausible outcome at least in the long term, and important enough to make at least superficial preparations for.

We’re already generating data on an unprecedented scale—emails, social media posts, WhatsApp messages. But there’s a glaring gap: these data points don’t fully represent who we are, in many cases it's not even close. Most people would agree that what they post online is only a fragment of their identity. We often distance ourselves from our online personas, and even when we don’t, we’re only seeing the linguistically parsable pieces. Humans are, as we know, more than the comments they leave—they’re also the thoughts they hold back.

If we aim to train language models to create accurate digital simulacra, we need data that go beyond the surface. To accurately model a realistic personal response, an LLM would need to understand how we reason, react, and emotionally engage. This kind of insight is significantly more concentrated in long-form content: journals, diaries, personal letters. The problem is, very few people create this type of content anymore. Non content creators have seemingly ceded this arena altogether to individuals who create on our behalf, however they generally produce material intended for public consumption, which may be more performative than reflective.

For the average person, the absence of long-form, introspective content poses a problem for creating reliable digital representations. Social media posts, which I would imagine constitute a disturbingly large portion of the thoughts modern humans commit to text, can’t capture the nuances of how someone works through a problem or navigates complex decisions. If all we feed an LLM are short-form snippets, we’re left with a superficial simulacrum that lacks depth and coherence. Long-form input is essential. Without it, 1500 character customization windows alone are too limited to encode deeply held principles and values.

To address this, I think documenting deeper thoughts and reasoning processes is going to become essential for anybody interested in truly personalised AI assistants. Because of this and other reasons, I'm going to be integrating journaling into my weekly rituals—not just for the immediate benefits of reflection and organization, but as a long-term investment in creating a record of how I think and reason. By doing this, I could one day provide a language model with richer, more accurate data to model a version of myself that isn’t just a collection of my social media highlights desperately scraped together minutes before the singularity deadline (that I almost certainly knew about ahead of time, but you know how your calendar can run away with you), but something closer to a meaningful digital presence.

This isn’t (just) about creating an immortal consciousness or a perfect digital twin. It’s about leaving behind a record of thought, complexity, and internal logic—a kind of mental time capsule. If nothing else, the act of journaling forces you to confront your own reasoning in a way that feels increasingly rare in a world of toxic tweets and digital-dunks. And that might be worth doing for its own sake.