sedCapital

@cryptypudgy @RichardMCNgo Link? You’re the only one who gave the real answer.

@RichardMCNgo extremely high dilution. look at the total market cap instead of the index number and you'd know why about 30x of mcap vs 3x in index over past 20+ years

What's the best explanation for why, despite China's economy booming, the Shanghai Stock Exchange has been flat for decades?

The average American goes to work for 8 hours during which they do about an hour of actual work) at a low exertion low risk job, and in return they consume probably 25 man hours of global labor product daily. We are now being told that somehow this amounts to the rest of the world getting over on us, though we’ve had virtually nothing to offer to them in return other than green paper which we print indiscriminately out of thin air. Imagine for a moment that we had an entirely closed economy. Absolutely everything produced in the United States and consumed in the United States. At that point the average American is not going to be able to consume more than the average American’s work product daily. I don’t think people understand the standard of living reduction that would require.

PGA TOUR officials are closely monitoring the weather and play is expected to resume later this afternoon. Updates regarding resumption of play and gate reopening times will be shared via THE PLAYERS Championship social channels and website. #THEPLAYERS | @PGATOUR

@MoonL88537 What would you like to no longer believe this?

honest opinion? we aren't really much past gpt4.

@adonis_singh @aidan_mclau @sumukx @FoundTheCode Do you think grok 3 plus thinking is better than o1 pro?

@aidan_mclau @sumukx @FoundTheCode but not the goodest

grok 3 (base) is MUCH better than sonnet 3.5 - and now is my go-to base model how the tides have turned

@nabeelqu No number 3 is not possible

So the options are: 1) LLMs/AIs are already somewhat sentient, or will be soon 2) Only biological organisms can be sentient, though we're not sure why yet 3) Consciousness isn't a thing Right?

@haralabob @michaelmorrell Thoughts on if sports market makers can be profitable without being able to limit toxic flow (by limiting successful gamblers)?

I've always employed partners or 3rd parties to make my bets for me. The last (only) time I tried betting at a LV casino was the 2014 NBA finals where I had some extra Chips after playing in a poker match and tried to bet the Spurs in G5 handed the guy a stack of chips he asked for my ID and then told me my limit was $500. So ya I would say I would not personally be allowed to ply my trade in Las Vegas.

@voooooogel @doomslide @repligate @aryanagxl @teortaxesTex Interesting. Thank you for your thoughts.

i have read through many, many R1 chains of thought, including skimming through 2MB of extended CoTs from the 32B distilled model. it's hard to describe succinctly, it's unique and def has emergent traits, but as a poor summary: it reminds me much more of a sharpened base model than a proto-paperclip maximizer

what deepseek did to get from v3 to R1 could have been done by openai on gpt4 way back in 2022. am i the only one who finds this odd?

@doomslide Example prompt?

both utterly fail to compose maps between spheres. (just like their ancestors.)

tested o3-mini-high on math and i'm not very impressed. it does feel a bit high at times with how much it likes to bullshit. judging by % of correct answers it's only slightly behind r1. its failure modes are a LOT more severe though.

@slight_blu @doomslide Can you explain this take?

@doomslide Remember when there were complaints that the safety team at openai wasn't getting the resources they wanted? Might be a funny world where these are connected.

@voooooogel @doomslide @repligate @aryanagxl @teortaxesTex Why did it make you worry less?

@doomslide @repligate @aryanagxl @teortaxesTex 😶‍🌫️ i still worry about RLVR but R1/R1-Zero made me worry less... i hope future recipes will still lean majority pretraining compute but we'll see

@Airick100 Taxable event to buy things

So I can make purchases with my pretax gains 😍

@interchaincosta @gregosuri @EffortCapital Are there not benefits to having massive clusters built by centralized companies? How does akash fit in?

honestly I think Akash is in a unique position to change the semi conductor industry Cheaper prices per gpu more scalability with AI Agents. History w NVDIA Brev uses Akash in its back end. Brev got acquired by NVDIA Since NVDIA bought Brev and Brev uses $AKT. NVDIA now holds $AKT simple

Since AKT all time highs: - GPU supply is up ~3x - GPU demand is up 9x - Daily GMV is up ~5.5x By end of 2025 Akash will have: - verifiable compute - confidential compute - one of the largest, fully onchain, decentralized compute incentive programs

@gregosuri @EffortCapital Thanks. In 2 years what is your best case scenario for akash?

@sedCapital @EffortCapital x.com/gregosuri/stat…

@gregosuri @EffortCapital How does nvidia use akash? It’s not clear.

@sedCapital @EffortCapital Was this link hard to find on akash website? akash.network/ecosystem/depl…

@EffortCapital Who

@sedCapital Yeah

@McaleerStephen Build smarter base models not just run time compute searchers

Ok sounds like nobody knows. Blocked off some time on my calendar Monday.

The way is quite clearly to build smarter base models and hope they develop morals to protect us. Most dangerous path is to scale up request time compute with current models. Evidence - talk with Claude.

@Tor_Barstad @liron I think clearly on aspects with a clear feedback loop - company building he has had success. Without it - not so much. To me this shows that he is very smart but his ego is so big (thanks to his successes) that he doesn’t properly examine his own viewpoints

@liron "To what extent is Elon smart?" continues to be a source of cognitive dissonance for me. Not that it's a new phenomenon for people to be successful while also saying silly things. But still.

This looks bad for Elon