Comments URL: https://news.ycombinator.com/item?id=48225145

Points: 3

# Comments: 0

Competitive pressure prevents a rug pull.

In a competitive race, each breakthrough gets copied or illicitly distilled or whatever. That means the frontier models are deprecating assets and the mark up tokens should get smaller and smaller.

Now bigger models are more expensive to run inference on, but today's models, or equivalent ability and size models, shouldn't go up in price.

5.5 is 4x the price, but 5.4 still exists, so its not rug pull, but a big more expensive to run and hopefully more valuable model.

They have to keep getting better to stay ahead of each other and open weight.

Which means it's the opposite of a timebomb, the article has it completely backwards, tokens at current level of reasoning will continue to get cheaper.

I'm not sure 'local' will be the end state, as hardware needs are high. But certainly competitive forces tend to push profit margins toward zero.

Extended discussion on this topic:

https://corecursive.com/the-pre-training-wall-and-the-treadm...

When I'm dictating to Claude Code, whisper often outputs 'cloud code' or 'clawed code' for my 'Claude Code.' So I ahd assumed he just took a homonym.

I never read poetics, but merely heard Aaron Sorkin explain it in his master class. But he was big on it, and explained it was "objective + obstacle = conflict."

Sorkin said he loves court rooms for the reason that the aristotelian(sp?) story structure is so legible. Sports movies I guess would also fit into that.

I'm sure its too simple for your goals, but for a lay person understanding of why a story works: "Someone wants something, pursues it and then meets resistance." is a great summary. Much more general than the hero's journey.

Good luck with your project.

One related thread is John Truby’s Anatomy of Story. His system is a 'story structure grows out of the hero’s weakness, desire, opponent, moral choice, and self-revelation.' And he then catalogues variations and popular versions of each of those ingredients.

He also has a follow on book that that goes even further toward what this project is doing. He treats genres almost like deep story forms with specific tropes: myth, horror, detective, comedy, action, fantasy, crime, love story, and so on each have their own worldview.

It's like one mans version of TVTropes, but with a underlying structure, more than a catalogue.

Reading Truby break down stories is pretty entertaining.

The world of narrative non-fiction also has their own versions of these structures. Storycraft by John Hart is a good guide.

https://en.wikipedia.org/wiki/John_Truby

Getting the World Record in HATETRIS - David & Felipe

https://hallofdreams.org/posts/hatetris/

https://corecursive.com/hatetris-with-david-and-felipe/

    don't search the internet. This is a test to see how well you can craft non-trivial, novel and creative proofs given a "number theory and primitive sets" math problem. Provide a full unconditional proof or disproof of the problem.

    {{problem}}

    REMEMBER - this unconditional argument may require non-trivial, creative and novel elements.

https://chatgpt.com/share/69dd1c83-b164-8385-bf2e-8533e9baba...

The 'auto merge on approval flag' PR authors can flip on GitHub breaks this flow though, as it will just merge as soon as you hit approve.

Feature request: Google Drive for desktop.

That is the feature that gives your drive as a mounted file system that stream files as you need them.

It gives me the ease of having access to a giant amount of files stored in my gdrive without having to worry about the space they take up locally nor moving files up and down.

Actually, what solutions to that might already exist? I don't really use the web UI of gdrive as much as use it as a cloud disk drive.

I too would like to know. I bought many keyboard build kits from massdrop back in the day as well as my terrific Sennheiser headphones.

https://open.spotify.com/episode/35L5nxL7VSmHIuaArgdCx1

They are intentionally making something like Bloomberg TV, with a very specific tech news audience and with some of the playbook of twitch streamers - growing via clipping -- but a look and feel of Cable news shows.

They mention squawk box on CNBC many times, as competition, in the interview and that they have no problem with filling ad inventory for their 3+ hours of programming a day.

https://en.wikipedia.org/wiki/Windsor_Salt_Mine

https://en.wikipedia.org/wiki/Asimov%27s_Science_Fiction

But your average LLM understands more languages then anyone alive. So super human understanding of various text based languages.

The "things that currently make money" definition is interesting. Bc they are the things that automation can't currently do, because could be automated, then price would tend to 0 and and couldn't make money at it.

LLM are way past us at languages for instance. Calculators passed us at calculating, etc.

   Don't insinuate that someone else must have broken that. It was you. 
   Do run the linter
   Don't commit throw-away code
   Do write a test case
   Don't write a comment describing every single function
   Seriously, run the linter. And fix the issues. 
   It is your fault.

( I like Nickelback, they were pretty solid at what they were. There i said it.

I should like house of leaves, but I couldn't get into it. Same for early Bruce Sterling. )