


Remarkable, Incredible and Amazing nerdiness


Have been working on a comics project for some time based on CoMix repo. This morning I thought – what if we added writer and artist into vlm analysis based embeddings – and other creators too. Did fine with panel text (in general).
So, gemini cli Gemini Pro 2.5 I gave this test page from Murder Inc: Jagger Rose to:


Based on the image, here are the creator names:
So very good!
Also, new Powers, there was a pleasant surprise – and it is good!
Going through my superhero games list – or the ones that were readable – Gurps 4 and Champions 6 apparently not by Notebook LM – but I can just this as a jump start to doing the human analysis of this I was going to do – but may as well post the robot version before I check it.
This is a source-by-source list detailing how the shrinking power is handled in each of the provided game books:
Not your usual topic, but I had the idea to have Gemini analyse what went on here. Basically had a family plan to get this, in Australia it was 10, then went up to 18, then 33, so tripled in price in rapid succession which is where I opted out. Not apparently 40! So 4 times as expensive. Better off buying albums really. Don’t use youtube much at all, so bundle of little interest in that sense.
Here’s the analysis.
The significant price increase of YouTube subscriptions in Australia, which saw the family plan jump from as low as $18 to nearly $40, is not an isolated event. It is the result of a strategic shift within Google/Alphabet and a broader maturation of the global streaming market.
The primary driver is a move away from a low-cost, user-growth-focused model to a revenue-maximization strategy. This is enabled by the unique and powerful value proposition of the YouTube Premium bundle, where the ad-free video experience is the core product, and YouTube Music is a valuable addition. The price now reflects the combined worth of this entire bundle, not just the music component. This change occurs within an industry-wide context of rising prices as streaming services, after years of prioritizing subscriber growth, now face investor pressure to demonstrate clear profitability.
Research confirms the user’s observation of a dramatic price increase. While the exact path from ~$10 to $33 depends on the specific plan and subscription date, the key jumps are verified:
These figures confirm that the user’s experience of a price more than tripling over time (from an early individual or family price to the current family price) is accurate.
The official reason given by YouTube for these global price hikes is to “continue to improve Premium and support the creators and artists.” While this is a factor, a deeper economic analysis reveals a more complex strategy.
The most critical factor is understanding what the subscription is actually for. While Spotify and Apple Music are pure music services, YouTube Premium is a bundle. The core, high-value feature for the majority of its subscribers is the complete removal of ads from the main YouTube platform.
The price hikes are a classic sign of a market moving from its growth phase to its maturity phase.
While YouTube Music competes with Spotify and Apple Music, the YouTube Premium bundle has no direct competitor.
This analysis is based on high-quality financial journalism, industry analysis, and Google’s own public statements (such as earnings call transcripts). As the topic is a recent corporate decision, formal peer-reviewed academic papers are not yet available.
Regarding the user’s comment on the CEO’s performance, while executive leadership certainly sets the overall corporate strategy (such as the shift to profitability), attributing a specific pricing decision to a single individual’s perceived abilities is an oversimplification. This price hike is a logical, albeit aggressive, market-driven strategy executed within a broader industry trend.
The tripling of the YouTube subscription price in Australia is not an arbitrary decision. It is a deliberate and calculated strategic move to transition the service from a low-margin music competitor into a high-margin, premium content bundle.
By leveraging the immense power of an ad-free YouTube experience as the core product, Google has been able to re-anchor its price to a much higher value point. This move, mirrored by price hikes across the entire streaming industry, signals a definitive end to the era of cheap, growth-at-all-costs streaming and a new focus on sustainable profitability.
This is a cool project – attempting to get models to caption and analyse comics – https://github.com/emanuelevivoli/CoMix
https://github.com/emanuelevivoli/CoMix
Visual – Language models should be continually improving.
Try Gemma and Qwen for example and see how they go in a simple chatbot interface.
What might be like what in this less than ideal game pdf hackery, here is an example:

These 2 examples, not too bad.
I made a gloVe embedding model based on my game book collection – 7000 odd, of which 6000 or so managed to make it through a first pass pdf extraction pipeline
This framework is quite good https://github.com/NRCan/geoscience_language_models/tree/main/project_tools
https://github.com/NRCan/geoscience_language_models/tree/main/project_tools and parallelises, which is important for big books
The C version of gloVe is superior:
https://github.com/stanfordnlp/GloVe
With some work you can get a python version going, but I wouldn’t recommend for large numbers.
e.g. https://pypi.org/project/glove-py
and associated hacks..
The Notebook associated with this is here: https://github.com/bluetyson/RPG-gloVe-Model
These days microsoft probably won’t let you see something that big online, so will make a series of post excerpts.
Campaign wiki style.
https://alexschroeder.ch/view/2023-08-13_Random_generators_on_wiki_pages
https://github.com/modality/abulafia
The old random generators – instructions on how to get this running again.
https://github.com/bluetyson/drpg
Python code to download and sync all your purchases