5,647 Posts
So, the elf hated my academic paper by me, on the subject of me, in the context of me. And rightly so, it is unreadable for a number of reasons. First, it is in the academic style. Second, it was poorly written by AI. And third, the subject matter isn’t that interesting, unless you are me.
The solution? Well I could turn the paper into a patent using my unpatented software. She can read patents alright. I will do that just for shits and giggles.
Or I could just write a summary here:
My blog, Offshore Westerly, has 5,647 posts published between April 2013 and July 2026. That is a lot of information on and by oneself.
What are all these posts? Well, observations, thoughts, critique, attempted humour, and often much more.
In this study what I did was write some code that uses an LLM to catalogue all these blogs into categories, subject, tone, and keywords.
It’s like taking a photo of yourself every day for 50 years and then stringing the photos together into a movie to see yourself age. That’s what I’m doing here in this analysis.
The categories in the study are drawn from the following options – AI, Quantum, Technology, Science, Business, Economics, Politics, Law, Society, History, Health, Travel, Sport, Personal, Humour and satire, Other.
The subject and keywords were extracted from each individual blog. Subject being sub-categories.
The options for tone were – Analytical, Satirical, Observational, Speculative, Personal, Explanatory, Critical, Narrative, Mixed.
The first observation was the frequency of blogs – I started with a bang, lost interest after 4 years and then found interest again 7 years later. Odd, because I didn’t stop thinking in those gap years. Looking back now, I just have no idea what I was thinking about. Hence the point of a diary.
The overall categories of the blogs are shown as follows:
There’s nothing surprising there – it just shows what was on my mind. That is somewhat a reflection of what presents itself to my mind – I will think about whatever comes in. I am not partisan when it comes to dishing out my neuronal capacity.
The top categories obviously imply a bit of personal interest in the subject matter. Otherwise why would I focus on them? These are society, personal, AI and politics. And isn’t that just the core of the question that is answered with 42? We are people and we are impacted by personal matters, societal matters, politics and, recently, AI. It’s the very essence of what both enables and threatens our existence.
My interest in these top 4 categories over the 13 year period shifted. AI was always there because I have been following it for decades. But in 2022 it took off in my blog writing, which is fully aligned to the emergence of the commercial LLM models. As someone who is deeply interested in both technology and words, as well as life in general, LLMs are catnip, baby.
The measurement of my tone is very subjective of course. But the LLM measured it thusly over 13 years:
I ran this test through different models and got less than 50% agreement. It turns out that LLMs are really bad at detecting dry ironic satire. But then so are humans, so that fits.
Anyway, I am not surprised or discontented by that breakdown. Seems about right.
The academic paper also included all this other bullshit about my approach to processing information. GPT claims that I have a fixed process that I use time and time again. It involves the following steps in order 1. Observation, 2. Make assumptions, 3. Create a model, 4. Push the model to limiting cases, 5. Find contradictions, and 6. Revise.
Yeah maybe I do that in important or extreme cases. Like my engagement with Quantum Computing for example. I keep pushing deeper into the science and technology to understand the inventive and commercial opportunities available, if any. But I am not sure I do that on other matters (like politics) where I have no intention of creating physical outcomes.
But having said that I suppose my “book” (Time Enough for Luck) is a physical outcome. So maybe GPT is right, maybe over time I do use this process to understand and improve the stuff that interests me. I remain on the fence.
I can’t help feeling though that this study has more legs in it than this summary suggests.
First, if I can fully extract meaning from it then maybe I can move on to something else. It would appear at the surface level that my approach, and even the subject matter, is somewhat static. I would prefer to move on and find something else rather than just turning the same wheels forever.
Second, the LLM processing is new and allows an awful lot of information to be processed where previously that was not possible. I feel like using that to my advantage to create something new (I like new things).
Third, I am the child of bullshit – we were taught that a life well lived is one that finds meaning. Siddhartha, 42, etc. So here’s my chance to dig into that well in a way that hasn’t been done before.
Fourth, I am also a child of the Renaissance. I prize novelty and invention through reasoning over all else. D’oh. So why not indulge in these addictions until they make me nauseous so I can give them away?
OK, so I am going to further process this data to see what else I can find.
I might ask the LLM to find and collate all the blogs on the same subject and then give me a one-page summary of my views on that subject. Then I will end up with say 100 pages of my life philosophies. Well not philosophies as such, but reasoned opinions.
offshore westerly 