Forgot your password?
Close
typodupeerror

Submission + - Nanoblog Index: Budgeted Parallel Lookup & Tweet-Count Ranking (Shyamal Chan (github.io)

Submitted by advancecoder
advancecoder writes: X publishes no global catalog of accounts and no lifetime ranking by post volume. Nanoblog Index builds two
public boards—celebrity and civilian—each capped at 100,000 seats, and ranks verified humans by raw lifetime tweet_count only. The private TypeScript worker looks up seed handles through the official X API in batches of 100, then writes a JSON snapshot the public Next.js site can render without holding a bearer token. This paper specifies the frozen grade covenant, the human filter, the seed-overlay snapshot, and a pre-calculated request budget that fans remaining lookups across at most eight parallel workers after a single probe call. We report local runtime, test-suite cost, and the request arithmetic for the current 2,533-handle seed. The design treats HTTP 402 (credits depleted) as a hard stop, not as evidence that zero accounts exist.

Submission + - MicroMachina: Turbo Music Video Generator (Shyamal Chandra of Crankshaft News) (github.io)

Submitted by advancecoder
advancecoder writes: Programmatic short-form advertising places demanding requirements on a generative pipeline: hyper-dense dialogue that remains intelligible, character consistency across independently generated clips, bounded cost, and resilience to the noisy, non-deterministic behavior of hosted generative-AI APIs. We present MicroMachina, a production web application that synthesizes sixty-second “fast-talk” commercial videos from a single textual pitch. The system couples an LLM-driven script director with a four-stage asset pipeline—text-to-speech, text-to-video, music, and ffmpeg muxing—and adds three properties that are essential for industrial use: resumability (a disk-backed checkpoint manifest that turns any failure into a resume point), resilience (bounded retries with backoff and semantic retry classification), and financial safety (a confirmation gate before spend and a user-invokable hard cancel that propagates abort signals into in-flight requests). We report the system architecture, the failure taxonomy, and the energy (consumption-budget) model. We validate the reliability-critical components with a hermetic test suite that exercises retry, cancel, and resume logic without incurring API cost.

Submission + - Press Release for Domino Data Systems Prompt Bee (github.io)

Submitted by advancecoder
advancecoder writes: FOR IMMEDIATE RELEASE
Domino Data Systems Launches “Prompt Bee,” an Interactive Challenge for Learning How to Think, Reason, and Communicate with Artificial Intelligence
New web-based project from Solo Technical Founder Shyamal Suhana Chandra transforms prompt engineering into an accessible “guess the prompt” experience designed to encourage experimentation, AI literacy, and human-machine understanding.
PITTSBURG, Kansas — August 11, 2026 — Domino Data Systems today announced Prompt Bee, an interactive web project that challenges people to explore one of the most important new skills of the artificial-intelligence era: understanding how prompts shape the behavior and output of AI systems.
Prompt Bee is available publicly at:
https://domino-data-systems.gi...
The project presents itself as “Prompt Bee — Guess the Prompt” and is structured around a challenge corpus that provides material for the interactive experience. The central concept is deliberately simple: instead of merely reading instructions about prompt engineering, participants are encouraged to examine a challenge, reason backward, and determine what kind of prompt could have produced the result.
Prompt Bee turns prompt engineering into something that can be explored as a puzzle, game, educational exercise, and research concept.
“Artificial intelligence is creating a new kind of literacy,” said Shyamal Suhana Chandra, Solo Technical Founder of Domino Data Systems. “We have spent decades teaching people how to operate computers by learning commands, menus, programming languages, and graphical interfaces. Generative AI changes that relationship. Now people can describe what they want in natural language—but the quality, structure, context, and precision of that language can dramatically affect what an AI system produces. Prompt Bee is an experiment in learning that relationship through discovery.”
Learning Prompt Engineering by Working Backward
Most interaction with generative AI begins with a person writing an instruction and waiting for the machine to respond.
Prompt Bee explores the process in reverse.
A participant encounters an AI-oriented challenge and attempts to reconstruct or infer the underlying prompt. That reversal encourages users to think about the ingredients that may exist inside a successful instruction: subject, context, objective, style, constraints, structure, details, relationships, and desired output.
Rather than treating a prompt as a single sentence, Prompt Bee encourages the idea that prompts can be understood as structured descriptions.
A conceptual representation might be expressed as:
Prompt = Goal + Context + Constraints + Details + Desired Output
A more advanced interaction can be thought of as:
AI Output = f(Prompt, Model, Context, Data, Parameters)
Prompt Bee focuses attention on the first component—the prompt—while helping users recognize that seemingly small changes in language can influence what an artificial-intelligence system understands and generates.
“Prompt engineering should not have to feel like memorizing magic words,” Chandra said. “It can be approached as reasoning. What information would an intelligent system need? What assumptions are missing? What attributes matter? What can be specified more precisely? When people start asking those questions, they begin moving from simply using AI toward understanding how to communicate with it.”
From AI Consumers to AI Thinkers
Domino Data Systems sees Prompt Bee as part of a broader transition in computing.
Traditional software generally requires users to adapt themselves to interfaces designed by programmers. Generative AI increasingly allows the interface itself to become language. A user can describe an objective conversationally and ask a machine to create text, software, analysis, images, plans, simulations, explanations, or structured information.
That freedom creates a new challenge: users must learn how to express intentions effectively.
Prompt Bee approaches that challenge through experimentation.
Each guessing exercise can encourage a user to examine clues, identify important concepts, distinguish essential information from decorative information, formulate a hypothesis, test that hypothesis, and refine it.
The process resembles elements of scientific reasoning:
Observe Hypothesize Prompt Compare Refine Learn
This feedback loop can potentially make prompt literacy more intuitive because users learn by attempting to reconstruct the instructions behind an outcome instead of only studying abstract rules.
A Project Built Around Curiosity
The bee and hive theme reflects the idea of many individual pieces of knowledge contributing to a larger intelligent system.
The Prompt Bee interface refers to loading the “hive” and fetching its “challenge corpus,” reinforcing the concept of a growing collection of challenges that users can explore.
For Chandra, the project also represents what a solo technical founder can create using modern web development and AI-oriented technologies.
“Today an individual developer can experiment with ideas that once might have required an entire software organization,” Chandra said. “AI can help people write, design, research, analyze, prototype, debug, and iterate. The role of the human becomes increasingly focused on imagination, architecture, judgment, and defining the problem. Prompt Bee is both an AI project and a demonstration of that new development environment.”
Applications Beyond a Game
Although Prompt Bee uses a playful challenge format, Domino Data Systems believes the underlying idea has applications in education, prompt-engineering training, AI evaluation, computational linguistics, human-computer interaction, and the development of better human-AI workflows.
Students could use reverse-prompt exercises to understand descriptive precision.
Artists and designers could study how visual concepts are translated into language.
Software developers could explore how requirements become instructions.
Researchers could investigate relationships among prompts, interpretations, and generated outputs.
Businesses could use similar exercises to help employees understand why vague instructions and structured instructions can produce substantially different AI results.
Prompt Bee therefore asks a larger question:
How well can humans understand the language that guides artificial intelligence?
As AI systems become more capable, Chandra believes the ability to formulate problems may become as important as the ability to solve them manually.
“The computer is no longer only a calculator or a machine waiting for rigid commands,” Chandra said. “Increasingly, it is becoming a reasoning and creation partner. But partnership requires communication. Prompt Bee is about practicing that communication.”
An Invitation to Experiment
Domino Data Systems is inviting students, developers, AI enthusiasts, researchers, creators, educators, prompt engineers, and curious members of the public to explore Prompt Bee and experiment with the challenge.
The project is accessible through a standard web browser and represents an evolving exploration of prompt reasoning and interactive AI education.
Users can visit:
https://domino-data-systems.gi...
“Prompt Bee is ultimately about curiosity,” Chandra said. “Look at an outcome. Ask how it might have been created. Break the problem into pieces. Build a prompt. Test your assumptions. Improve your understanding. That cycle is fundamental not only to prompt engineering, but to invention itself.”
About Domino Data Systems
Domino Data Systems, based in Pittsburg, Kansas 66762, is led by Shyamal Suhana Chandra, Solo Technical Founder. The organization develops and explores software, artificial intelligence, data systems, interactive computing concepts, and experimental technologies focused on new ways for humans and machines to create, reason, learn, and work together.
Media / Project Information
Shyamal Suhana Chandra
Solo Technical Founder
Domino Data Systems
Pittsburg, KS 66762
United States
Prompt Bee:
https://domino-data-systems.gi...
###

Submission + - Introducing...book2screenplay from Crankshaft News from Shyamal Chandra (google.com)

Submitted by advancecoder
advancecoder writes: Abstract:
We present book2screenplay, an end-to-end system that transforms a PDF book into (1) a production-formatted screenplay rendered with the screenplay.cls LaTeX class and (2) a narrated 4K picture-video with per-scene imagery, multi-voice text-to-speech, and optional rich caption overlays. The pipeline is implemented in Rust with tokio-based concurrency and orchestrates a hierarchy of local large language model (LLM) agents served by multiple Ollama instances. A Director agent produces a story bible and beat structure; parallel Writer agents draft Fountain-structured scenes; Continuity and Polish passes reconcile tone and formatting; image and speech stages materialise the screenplay for audiovisual output. We report architectural findings from development on a 96 GB Apple Silicon Mac Studio: memory is dominated not by the Rust orchestrator but by concurrent model residency and 4K ffmpeg encodes; explicit model eviction and KV-cache quantisation reduce peak resident set from 70–90 GB to 48 GB without sacrificing throughput on a two-server Ollama pool. We document failure modes—JSON truncation, schema drift in model output, and optional-filter absence in minimal ffmpeg builds—and the mitigations adopted (context budgeting, lenient deserialisation, capability probing). The system demonstrates that fully local, agentic book-to-film pre-visualisation is technically feasible for feature-length targets, though page-count fidelity and cross-scene continuity remain open research problems.

More details coming soon!

Submission + - Introducing...Agentic Parallel Book Writer from Domino Data Systems. (google.com)

Submitted by advancecoder
advancecoder writes: Abstract:
Generating a coherent full-length novel (300 printed pages, 90,000–300,000 words) from a single prompt is well beyond the effective context window of every publicly available large language model in 2024–2026. We present AGENTICBOOKWRITER, an industrial-grade system written in Rust that achieves this by decomposing the problem into a four-tier hierarchy of specialized agents—a Director, a Planner, a pool of Writers, and a pool of Reviewers—each implemented as a prompt template against a pool of locally hosted Ollama instances [Ollama contributors, 2024, Gerganov and contributors, 2023]. A shared “canon” data structure (synopsis, world bible, character sheets, scene beats) is serialized to a single JSON file that doubles as an atomic, scene-granular checkpoint, making a multi-hour run crash-safe and resumable. We engineer aggressive parallelism (up to 16 concurrent writer requests on a 28-core Apple Silicon Mac Studio) through a port-sharded pool of Ollama servers, guarded by a per-request num_ctx auto-sizer and a transient-error retry classifier that neutralizes the known GGML_ASSERT KV-cache bug in llama.cpp. On llama-3.1:8b we generate a readable 90,000-word first draft in 48–80 minutes wall-clock; on a 300,000-word target the same pipeline completes in 3–4 hours with a graceful fallback that has never, in our measurements, lost a chapter to a model crash. We contribute: (i) a canon-plus-fallback prompting protocol that tolerates small-model schema drift, (ii) a tier-specific concurrency bound that keeps the KV-cache stable under long prompts, (iii) scene-granular resume semantics, and (iv) a reproducible single-binary Rust implementation.

More details coming soon!

Comment Questions for GNU Kind Communications Guidelines (Score 1) 448

by advancecoder (#57524003) Attached to: Richard Stallman Announces GNU Kind Communication Guidelines
On another note, here are my thirty questions due to the loaded, lossy, and opinionated terminology dictating the GNU Kind Communication Guidelines (https://www.gnu.org/philosophy/kind-communication.html). Please let me know if there is any confusion because I am totally confused and as a user and modder of GNU software in the past, present, and future, I feel it is not only my obligation but my duty to respond to this post on the gnu.org website. 1. What do you mean by advance? 2. What do you certain patterns of communication mean? 3. What do you mean by conscious effort? 4. What do you mean by specific? 5. What do you mean by good faith? 6. What do you mean by wrongs? 7. What do you mean by harsh tone? 8. What do you mean by accept it? 9. What do you mean by actually say and actually do? 10. What do you mean by preferences? 11. What do you mean by personal attacks? 12. What do you mean by hit back? 13. What do you mean by feelings? 14. What do you mean by private? 15. What do you mean by peace? 16. What do you mean by anger? 17. What do you mean by typical desires, capabilities, or actions? 18. What do you mean by kind? 19. What do you mean by conscientious? 20. What do you mean by mistakes? 21. What do you mean by nonfree? 22. What do you mean by not allowed? 23. What do you mean by exaggerations? 24. What do you mean by constructive criticism? 25. What do you mean by real views? 26. What do you mean by tangent? 27. Why not have a special hotline for GNU software w.r.t. that repository? 28. What do you mean by comfortable? 29. What do you mean by cater? 30. What do you mean by friendlier?

Comment Questions for Malcolm Gladwell! (Score 4, Interesting) 111

by advancecoder (#48411921) Attached to: Interviews: Ask Malcolm Gladwell a Question
Ten-thousand hours (~3.4 years if a regular job) is Gladwell's estimate of temporal mastery. With that being said, the Mozarts like Carlsen or Fischer learn faster and become World Champion. What is the difference between the Mozarts and 3.4 years? Is it there some passionate rage to absorb and decipher patterns that magnetizes them to a particular domain or is it their consistent, well-designed regimen for reaching the upper echelons (like Lalzo Polgar's systemic approach with Judith and Susan)? If it is "pure" passion, then will people who find their true calling and invest appropriate time (e.g. have an OCD mentality) always see the unquestionable results? If it is "pure" regimen, then will following the same systematic approach always see the overarching performance? One thing to keep in mind is are these skills transferable to other domains? Is there a way to tackle a number of domains in the same 10,000 hours with an abstract approach? What about the time to create "new" domains rather than to "solve" problems in a particular domain? Is there some sort of estimate for that? Malcolm could possibly use those clues for his sequel to "Outliers" appropriately called "Pioneers". Any thoughts?

Slashdot Top Deals

Gravity brings me down.

Close