wassup nerds: a narrative knowledge graph of all 7 seasons of Star Trek: The Next Generation fabula.productions/explore/star... #knowledgegraph #contextgraph #startrek #llm
10.03.2026 15:42
👍 1
🔁 0
💬 0
📌 0
wassup nerds: a narrative knowledge graph of all 7 seasons of Star Trek: The Next Generation fabula.productions/explore/star... #knowledgegraph #contextgraph #startrek #llm
all those "Hi it's Chad from Blancmange, I noticed you haven't used Blancmange in a while..." emails might land better if they included any details to help me remember what product this is
If you like stories, graphs and television from simpler times you may enjoy this project that is coming together at last. fabula.productions
"I'd love to be able to do what you do. Is there a Medium post I can read?"
A Gemini 3.0 one-shotted mockup of 'Encarta 2026' displaying Mind Maze Score: 100 Which gas, now vital for complex life, was almost entirely absent from Earth's atmosphere for the first 2 billion years of its history? A.Nitrogen B.Oxygen C.Carbon Dioxide D.Argon
this is too much fun
credit where credit's due to the web typogapher of www.cia.gov/the-world-fa...
![Radio as primordial soup: fair cop. I was modeling TV comedy and forgot that Radio 4 is where British comedy goes to gestate. The [:ADAPTED_INTO] relationship capturing the Radio → TV → Film "golden path" is essential domain knowledge I missed.](https://cdn.bsky.app/img/feed_thumbnail/plain/did:plc:tc4vvcx7fluympvdy5claawy/bafkreidejbgqughwh4bpjewnbolmwtddkcfbcq7kgj6abtryqo2orxjv4a)
Radio as primordial soup: fair cop. I was modeling TV comedy and forgot that Radio 4 is where British comedy goes to gestate. The [:ADAPTED_INTO] relationship capturing the Radio → TV → Film "golden path" is essential domain knowledge I missed.
I wrote a paper on exactly this in college and Opus 4.5 drops like its nothing
OpenRouter Wrapped message tells me I routed 1.1B tokens in 2025
this might be some sort of achievement
claude code configures a local media servers and suggests movies for download
giving claude code system access is quite a thing
Airplane (1980)
Claude thought process: Alright, Mike is PUMPED. Let's build this thing properly. I should create an actual plugin manifest that captures the sophistication of their runbook. Given Mike's expertise in systematic approaches, I need to: Pick a meaty example - probably the fact-checking workflow since it's their newest addition and has clear patterns Show the full plugin.json structure Include actual agents, commands, and skills Make it production-ready, not just a toy example
i'm not un-scared
eventually a designer will realise that a 2-line text entry box is not an ideal UI for prompt work.
object_7acb67ec2204 — \"President Bartlet's Requested Roast Beef/Pastrami Sandwich.\" owner = President Bartlet, filling specified as roast beef or pastrami, functions as focal point of a ceremonial/domestic exchange (Charlie / Mrs. Landingham).
The 2010s really are over eh
❌ "Wesley helps with the investigation"
✅ "Wesley connects nuclear vibrations to Elway Theorem, suggesting inter-dimensional transport"
working without AI feels a bit like working offline used to
Close-up of a vintage Zoommax computer case showing a green digital display reading "66," with Power, Turbo, and Reset buttons visible.
Adding vectors risks: Over-engineering a system that already works Creating a "fancier hammer" when you need better nails Introducing opacity where you currently have explainability
im not hating claude's spicy takes
here's Star Trek: The Next Generation season 2 (the one with Pulaski) as a fabula storygraph dataset
huggingface.co/datasets/bra...
The source entity is an implanted, failing cardiac device inside Picard. Four potential duplicates included two object-like implants, one surgical procedure (event), and one unrelated Borg implant. Plan: Evaluate each pair independently using name-variant heuristics, object-vs-event ontology rules, possession/context clues, and narrative/function comparisons. Execution & decisions:\n- Merged source with 'Picard's Early-Model Cardiac Unit' because descriptions align as the same implanted device (same role, location, and surgical target). Kept source UUID as surviving and synthesized a canonical name and enriched description to capture both technical and surgical terminologies.\n- Kept separate 'Picard's Replacement Heart' because that is an external spare implant prepared for transplant and functionally distinct from the installed failing device.\n- Kept separate 'Picard's Cardiac Replacement Procedure' because it is an event/process, not a physical object.\n- Kept separate 'Borg Cranial Assimilation Implant' because it is anatomically and functionally distinct and belongs to a different narrative category (Borg tech).\nValidation: Ensured pairwise independence and internal consistency (only one MERGE, surviving UUID chosen consistently as source_entity_uuid). Potential weaknesses: If source text later clarifies that the 'replacement heart' label was used interchangeably for the implanted unit rather than the external device, the KEEP_SEPARATE decision for object_20827fb030ca would need revisiting. Confidence levels: MERGE (high), KEEP_SEPARATE decisions (high) given current embedded_contexts and canonical descriptions." }
hows your evening going
i just typed 'json fields' and suddenly 2009
wow this is interesting. See you both there!
Claude: The new data filter is working well!
Me: there are zero results
Claude: The filter is working better than expected!
But of course
this thing I made *in keynote* years ago was, on reflection, way too good for the thing I made it for
Star Trek: The Next Generation Season 1 as a narrative graph.
huggingface.co/datasets/bra...
screenplay from Star Trek: The Next Generation Episode "Coming of Age" features placeholder (DIALOGUE TO BE WRITTEN) message.
bold screenwriting choice
