People keep journals. Do it aggressively enough and eventually your soul gets version-controlled on GitHub.
Once an archive passes a thousand articles in a few months, “I wrote a lot” stops being an adequate description. What remains is not just prose, but fragments of what once felt strange, funny, suspicious, unfair, or worth understanding, along with the reasons behind decisions and the places where opinions later changed.
The blog starts to resemble something stranger:
a cursed-object knowledge base built from fragments of past thought.
In Harry Potter terms, it is getting dangerously close to a Horcrux-shaped life log.
1. What is actually stored in 1,000+ articles?
A large archive preserves events and patterns of judgment. Why did a rule feel unreasonable? Which sentence in a conversation caused friction? Why did a trivial purchase trigger a cost-per-unit calculation? Why did something incomprehensible years ago finally make sense later?
Each incident is small. Together they form a map of what a person tends to notice and how they explain it.
A diary is good at storing what happened. Explanatory articles derived from questions and conversations can store what happened, what it meant, and why that interpretation seemed reasonable.
So the important fact about a thousand articles is not that someone typed a thousand times. It is that they captured more than a thousand snapshots of how they interpreted the world.
Voldemort stopped at a handful of soul fragments. A modern archive scales horizontally by article.
Please do not horizontally scale the soul.
2. An article archive can become external memory
Humans routinely move memory work outside the brain. Appointments go into calendars, groceries into notes, and locations into photos. Instead of keeping everything internally available, we build reliable paths back to information.
Cognitive science calls this cognitive offloading (Risko & Gilbert, 2016). Clark and Chalmers (1998) went further, arguing that under the right conditions external artifacts such as notebooks can function as parts of a larger cognitive system.
A searchable blog can play a similar role. Instead of trying to remember “What did I think about this years ago?”, you search the archive. Instead of reconstructing why you made a decision, you read the reasoning recorded at the time.
Your biological memory did not become larger. Your routes back to forgotten reasoning did.
3. Why the Horcrux metaphor is disturbingly accurate
A blog does not literally contain a soul. Still, the metaphor works because Horcruxes—and especially Tom Riddle’s diary—are fictional objects that preserve something of a past person and allow that past presence to speak again.
Old writing can produce an ordinary version of the same effect. Open an article from years ago and a former self boots up: “Right, this is what bothered me.” “I still use that joke.” “I would completely change this conclusion now.”
The container is Markdown rather than dark magic.
A Horcrux-style life log, deployable through a CDN.
At this point the Ministry of Magic probably needs a platform engineer.
4. Git gives past selves version history
A notebook can already serve as external memory. Git adds provenance and change history: when something was added, what changed, what disappeared, and how one version differs from another. Past selves acquire diffs.
When an opinion changes, the story no longer has to be “I used to be wrong.” You can trace: “At this point I believed X; then I learned Y; after that, this paragraph changed.”
Git is not a machine for forcing personality consistency. It is a machine for preserving change accurately.
Split the soul enough times and eventually you need a SHA.
Do not put CI/CD on a Horcrux.
5. Search first, then let AI read the relevant fragments
This is where the archive becomes especially useful with AI. When a question arrives, first search the archive for relevant articles. Then give those articles and the question to a language model and ask it to answer using that material.
The technical name for this “retrieve evidence first, generate second” pattern is retrieval-augmented generation, or RAG (Lewis et al., 2020).
Now the archive can answer questions such as:
- How did I think about this problem before?
- When did this value start changing?
- Have I described a similar discomfort elsewhere?
- Which older article contradicts my current conclusion?
Once the archive grows beyond what one person can remember, AI can act as the librarian of the restricted section.
Voldemort hid his Horcruxes. This system builds a search index over them. Opposite operational philosophy.
6. A “personal AI” is not a copied personality
Feeding thousands of articles to an AI does not reproduce a person. It does not transfer consciousness, guarantee future decisions, or create a digital continuation of the author.
What it can create is less mystical and more useful:
a system that retrieves earlier judgments, language, questions, reasons, and changes of mind, then answers with that history available as evidence.
That is less “AI clone” and more “AI that keeps summoning earlier versions of me from the records room.” Not cloning. Search with advanced necromancy.
For practical use, this may be better. A model that merely sounds like you can confidently invent things. A system that can point to the dated articles behind its answer gives you something inspectable.
7. Horcruxes have curses too
More stored material does not automatically mean more truth. Old articles preserve outdated knowledge, biases, mistakes, half-serious jokes, and conclusions written without later context. A retrieval system can also drag an ancient joke into a serious answer with the confidence of a tax auditor. Storing everything indiscriminately creates privacy risks too.
Large external memory therefore needs operational rules: preserve dates and sources, mark superseded pieces, remove identifying information before publication, show which articles were retrieved, and re-check important conclusions against newer evidence.
If you insist on making Horcruxes, put the curse-removal procedure in the README.
Volume makes memory richer. Without quality control, it also makes a larger maze.
8. Conclusion: not immortality, but queryable past selves
The strongest reason to maintain a huge article archive is not digital immortality. It is making past thought searchable.
You can recover what you noticed, doubted, laughed at, and concluded. With retrieval-based AI, you can bring the right fragments back when needed and compare them with the person you are now.
Humans change. Contradictions between old and new writing are not necessarily corruption. They can be the most valuable diff in the repository.
You do not become immortal. You become able to send search queries to your former selves.
Voldemort tried a handful of Horcruxes. A modern individual crosses a thousand articles and adds Git, search, AI, and quality gates.
The magical world had lighter operations.
Do not attach an information passport and version control to your Horcruxes… actually, do.
References
- Clark, A., & Chalmers, D. (1998). “The Extended Mind.” Analysis, 58(1), 7–19.
- Risko, E. F., & Gilbert, S. J. (2016). “Cognitive Offloading.” Trends in Cognitive Sciences, 20(9), 676–688. https://doi.org/10.1016/j.tics.2016.07.002
- Lewis, P. et al. (2020). “Retrieval-Augmented Generation for Knowledge-Intensive NLP Tasks.” NeurIPS 2020. https://arxiv.org/abs/2005.11401
- Harry Potter, “Horcrux” fact file. https://www.harrypotter.com/fact-file/objects/horcrux
- Harry Potter, “Tom Riddle’s diary” fact file. https://www.harrypotter.com/fact-file/objects/tom-riddles-diary
