-
- Private work.
- Reached out to suitsupply (email) for alteration update.
- Crossword hints.
- Singular, plural. Always look for Ss.
- Can be a sentence without spaces (vs a single word).
- Alternate meanings of clues, often smart-ass (like taking a famous quote with a known meaning, but then interpreting it literally).
- Question mark means pun or wordplay (don’t take face value).
- Still hate them. Require specific knowledge (unlike most games being inferable), half the answers are bullshit, half the answers aren’t real words, not satisfying.
- Magnus wins on-demand to survive and push tiebreaks with keymer.
- Stephen King’s Rest Stop was made into a movie (The Passenger). Didn’t know.
- White hat security vulnerabilities on points.com: https://samcurry.net/points-com/.
- Almost all vulnerabilities were unauthenticated endpoints or access to auth.
- Still very manual. Poke around, find an admin endpoint, check how auth works, inspect the js, register or fake auth or spoof from a real account or intercept real users.
- Look for “gift” or “transfer” options in APIs. They usually allow actions on accounts you don’t have auth for.
- Lol: “The Flask session secret for the website that was used by points.com employees to manage all rewards profiles, loyalty programs, and customer orders, was the word “secret”.”
- What to run on the server vs client: https://www.mux.com/blog/what-are-react-server-components.
- Old-everything-server-approaches. Then move to client side rendering, SPAs. Then server-side-generation of next, where server generates html for initial load as well as well as sending js for interactivity.
- Then react server components and client components. Explicitly define what you want to run, and where. That way, both sides don’t do ALL html and js work. You define what you want to gen where.
- Does Money Stuff affect market volatility? https://papers.ssrn.com/sol3/papers.cfm?abstract_id=4386256
- Did some thinking about AI.
- What’s the key for intelligence? Is it inference? Connect-ability? Knowledge (data)? The answer should be the same for natural/artificial. (This is pretty generic, and maybe obvious, but worth stating) Turns out it’s learning.
- Then what’s the key for learning? Turns out it’s information.
- And the key for information is simply machine ingestion of everything. Scrape the whole internet – earth’s best (single) repository for knowledge.
- But then how do you use intelligence? How do you interact with it? How do you get it to provide solutions, creations, etc? The key is simply communication. If you can have the intelligent-fully-digested-knowledge-base speak the same language as you, you can extract answers from it. Just teach the AI english.
- And the key to that is simply text completion. Guess the next word that makes the most human sense, based on the context of the sentence/paragraph/question/etc and the dataset of language that you are trained on.
- This is the key for E2E AI. It’s not just “what’s the capital of germany” and searching for wiki articles until a most-likely statistic is determined. It’s “how do I discretize a desire into step-by-step probabilistic solutions until I can create anything, based on what we already know?”
