JohnBerryman

I have a conspiracy theory.

Consider an analog - Human brains are made of neurons, and so are fruit fly brains. We now have a fully connectome of drosophila melanogaster, so shouldn't we – very smart brains – be able to look at the fly connectome and make sense of it by just looking at it? Of course not! That's friggin' crazy. Our neural weights have been trained to take in data from senses and to produce output behaviors (moving muscles, squirting hormones).

An LLM has been trained to take in text and images and generate "good" output text. It doesn't natively understand what the 17th weight on the input channel implies. Instead, when it sees "strawberry" what it really sees are 3 symbols "st", "raw", "berry", and these symbols are really just identifiers [302, 1618, 19772] which get converted to basically meaningless numerical vectors on the input plane. This is why in 2025 the frontier models kept swearing to us that "strawberry" had 2 r's!

So when the model sees the full representation of a pacman playing neural network (here's the actual prompt they used https://lnkd.in/e5Td2amH) it reads in a bunch of numbers like "0.054 -0.026 0.014" as text. This isn't somehow presented to the LLM it in a "native" neural weight form. Rather, it gets tokenized just like everything else, into meaningless symbols – in this case "0", ".", "054", " -", "0", ".", "026", " 0", ".", "014". Nevertheless, the models are able to read reams of numbers and pick out patterns that let it know that it's looking at a pacman playing neural network.

Getting back to my conspiracy theory. This new talent for LLMs to read and understand model weights is unnatural. You won't find much training data available on the internet that is analogous to this. Humans simply train neural networks. We don't look at specific networks and write essays describing how model weights lead to various model behaviors.

This new talent is so unnatural that I posit it is actually being trained into the frontier models intentionally. The frontier companies are actively experimenting with models that are capable of reading models and directly understanding and modifying weights. OpenAI and Anthropic have known this capability existed prior to the report because they are the ones that put it there!

Welp... it took 2 weeks for my prediction to bear out – https://huggingface.co/blog/sora-2/what-is-openai-decisions-... https://openai.com/index/devday-2026-recap/

Panagram seems fun! I tried it out with a large chunk of my post (whole thing wouldn't fit). Panagram says it is 98% human. Feels right, to me b/c I aggressively edit whatever comes out. The thing that I'm avoiding is the ominous blank page - I just freeze. If there's text there I can always reshape it. And I do, heavily.

prodigycorp is entitled to their opinion

Partially. I'm a terribly slow writer and get stuck on phrase choice, but I'm good at content ideas, outlines, and editing text already on the page. So I have AI do the bit that I'm not as good at.

Process: First, actually have ideas :D Then, I write an outline for what I want to talk about at basically a sentence-by-sentence level. (This is me yelling things at my computer.) And then I have the AI convert a chunk at a time into prose. I reread it and rework it to be my voice.

Then I have the AI help with things like subject titles and social posts.

¯\_(ツ)_/¯

For me, I think the big deal is that it promises to be general and broadly applicable and high quality. That's new and special. But we'll wait to see if the claims actually hold.

You might find my old post interesting https://arcturus-labs.com/blog/2025/03/31/supercharging-llm-...

I'm glad someone noticed :P - It was hilarious once we figured out what was going on. We had literally removed it's ability to shut up.

chaotic neutral

Nope. Here's the closest I got https://arcturus-labs.com/blog/2025/03/31/supercharging-llm-... - 1.5 years ago! But I never really did anything with it. And I wasn't thinking about reinforcement learning anything.

I think that ego/status is a big part of it. And the loss aversion bias. Our thought process is "This thing is going to be able to do software as well as a human, therefor it is going to take my job." When a more optimistic and probably more realistic take is "This thing is going to do the boring parts of software, so I can spend my time building out the more interesting, big-picture parts of code. I'm going to be so much more productive." But the fear of loss is so much more distracting than the potential gain. ... And what is it that we fear losing? Status! Software development is in high demand, it's a sought after role, we feel "special".

Now that it's released, it's time to experiment with what it can do.

``` from cataclysm import doom

def mystery_func(): while True: pass

# predict if specified function halts print(doom.does_it_halt(mystery_func)) ```

I wrote a book (https://www.manning.com/books/relevant-search). Considerations: are you willing to write the book assuming that you will effectively make nothing in direct proceeds from the book? (Close to true.) The upside from a book is in monetizing on it indirectly. For instance, as a "though leader" you'll be able to garner a higher salary at your next job or you'll be able to ask for higher rates for consulting. But the catch here is that you've got to be willing to put yourself out there, otherwise it's a nice bullet point on your resume.

Oh yeah - and don't underestimate the time and effort involved. (I know you mentioned it in your post, but still.) A due date, even if fairly far off, looms ominous. For a year you'll have trouble relaxing and having fun because you know you should be getting work done on the book. You won't have weekend back for quite some time.

Would I do it again? Yep. Will I do it again? Nope. (At least not likely :D )

Good luck!

DataDog uses elasticsearch for their timeseries data store: http://www.elasticsearch.org/content/uploads/2013/11/es_case...

Elasticsearch might seem like a strange option at first since it's historically a text search engine, but it's main datastructure is a compressed bit array which is ideal for OLAP processing.