From Google Docs to Plain Text

I recently moved from using Google Docs to creating plain text files using TextEdit on my Macbook. I now have a simple content management system with all my files that lives on my desktop, and is backed up remotely using Github.

Some of my files include things like: watch lists, ideas, notes, journal entries, old writings and more. I was inspired to move over to plain text after reading a blog post by Derek Sivers, who gave me these four good reasons for doing so:

1. They’re portable and will be the most likely file type to stand the test of time.
2. They’re uncommercial, which means no platform locked content.
3. They’re available offline, whenever and wherever you want.
4. There are no dependencies on other companies, who could at any moment shut down, or shut you out.

However, aside from these four reasons for creating plain text files, I think that this is currently the best way to start building what I’m calling Context Pods,

What is a Context Pod?

A Context Pod is exactly what it sounds like, a pod of information that provides context to your life. For now, I’m starting with simple text files, but I do see my context pod eventually growing into storing expansive financial and health records, videos, photos, entire chat logs, and more. So, practically a digital memory disk.

Why would I need a context pod in the first place?

Prior to the rapid expansion of using large language models like Open AI’s ChatGPT or Anthropic’s Claude, there was no way to process large amounts of information at conversation like intervals.

My reasoning for thinking context pods will become an essential component of incorporating an AI assistant in your life, is because, context is everything. When you chat with an assistant like ChatGPT or Claude, it knows nothing about your life, well at least to an extent that is. ChatGPT does a memory feature, but I don’t believe it provides anywhere near enough context on who you are as a human being, to provide accurate enough insights for impactful change.

LLM’s lack of personal memory

Over the past few months I’ve been using Claude as a psychotherapist and counsellor to analyse and breakdown self-limiting beliefs I’ve had about myself. Whenever I do this, I need to warm up the assistant and usually provide enough context before we can get into our reprogramming session.

Some chat logs with these assistants can go on for days as the model learns more and more context about me. Until, unfortunately, the model runs out of memory and ends our chat right there with all our history. Which means I have to start up another chat to rebuild all the context I just lost. Practically making this the equivalent of having to go through the talking stage all over again.

Which is where Context Pods come in.

The vision for Context Pods

If you had your own context pod, you could plug it into your LLM of choice, and it would immediately make your interactions a hundred times more impactful. Imagine, if the LLM instantly had a record of your health history, unhealthy habits, and your personal story? You could hop right into the deep end about why you act a certain way, or how you could remove that one unhealthy habit that’s plaguing your life.

Privacy concerns

However, I do understand how exposing huge chunks of your life to AI models, could become problematic, and I do not have a solution to this. Perhaps the answer might be running your own private models at home, that are encrypted in some way, but as I said, I do not have the answers, and am nowhere near as smart to give them to you right here.

I do know however, that this information is the oil of the internet, and should be protected as such.

Reclaim ownership

I believe in people taking back ownership over their data. For too long has big tech practically owned each and every one of us and sold us off to advertisers. The future I envision is a world where, if they want our data, they’re going to need to pay up. No longer is access to your mind-numbing service enough compensation in exchange for putting in twelve hour shifts in the data coal mines. Hence why I’m advocating building out your own context pod that you own, encrypt, and store somewhere safe for future use.

Existing implementations of Context Pods

Tim Berners Lee, the creator of the World Wide Web, was building something similar to a context pod with his project, Solid (Social Linked Data), that allowed users to have selective control of who could access their data. However, I’m unsure on the progress of this project, and haven’t heard or seen anything as of late.

Do it yourself

Which is why I’m taking it into my own hands to begin building out my own context pod, which in reality, is just a folder on my computer, with hundreds of text files about my life. But this doesn’t mean this glorified folder won’t be useful in the future.

Long live you

Your context pod is how we will be able to rebuild you, and get your tone of voice just right, so your great-great-great-grand-kids can interact with you, and perhaps learn more about their inherited family trauma, or maybe just chat with you about what it was like living through the AI boom. And yes, I’m also aware of the implications something like this could have, but that’s a conversation for another day.

For right now though, as Citizens of The Internet, we need to reclaim ourselves, and take back what’s rightly ours, our data. Today it’s plain text files, but in the future we will have financial statements, videos, chat logs, viewing preferences, and every other imaginable piece of data. Which means, it’s in our best interest to make sure we shepard the data that makes the web go round, the right way.