News for hackers. Stuff that matters.

octopod12

Re:Can we cut the AI Agent for X thing already on Friday October 04, 2024 @11:29AM (#41742437)
the cycle repeats, in every tech transition, there is a bandwagon, and herd mentality amongst VCs and young tech-bros. which involves FOMO and me-too cloning.

in the sharing economy it was AirBnB for X, Uber for Y etc.

in the blockchain boom, it was BlockChain for logistics etc.

in the crypto boom it was NFT for this or that.

in the SaaS era it was cloud CRM, cloud HR etc.

in the dot-com mania, there was pets.com and webvan.com

Re:Johnny LLM Can't Read Code on Wednesday October 02, 2024 @10:53AM (#41721220)
A turing machine is fixed pipeline too. if you define the "pipeline" as its finite set of symbols and states.

and I agree that a folksy sounding title usually hides fuzzy thinking.

but the interesting point made is whether an LLM can figure out if it has "actually" solved a problem or not (?). Technically, this is the classical "halting" problem.

Re:Pear AI founder: We made two big mistakes on Tuesday October 01, 2024 @11:07AM (#41709412)
there is a culture of "gaming" that is in fact promoted by YC. one of the questions they had was to show an instance where the founders "hacked" or "gamed" a system.

ie, game the test, game the admission, game everything. and some folks see this as "hustle".

eventually their own process gets gamed.

gaming incentivises a culture of "pretension" - you dont know sh* but you fake it. fake it til you make it. copy code-repos, trample over licenses, whatever, who cares as long as you are getting ahead.

Re:Pear AI founder: We made two big mistakes on Tuesday October 01, 2024 @10:58AM (#41709280)
Here is my theory.

1) management types who have never coded love the idea that coders can be replaced or minimized by said tooling. hence hey fund any and all projects.

2) coders excited by llms reach out to the nearest projects they can relate to, namely, code assistants, test generation assistants, doc generation, and generally any automation in the coding workflow. so, you will see a lot of me-too tooling.

Re:Y Combinator Traded Prestige for Growth on Monday September 30, 2024 @12:19PM (#41698923)
the founders were status-signaling on twitter.

it is clear they are in it for the "status" of being YC. and they dont care a whit about solving anybody's problem.

they do this for a while, get it on the resume and go back to their 270K jobs after a few months.

this in and of itself is a hustle, lol. they hustled YC.

Re:Y Combinator Traded Prestige for Growth on Monday September 30, 2024 @11:43AM (#41698484)
The founders showed hustle, as every founder must. nothing wrong with that in my book.

But they need adult-guidance on communication. You dont go around twitter boasting about your 270K job etc. They need to show grown up hustle (grit, perseverence, etc). Not high-school (mine is bigger than yours) hustle.

Re:Ask HN: What are you working on (September 2024)? on Monday September 30, 2024 @12:20AM (#41693526)
wont your strategies incur short-term cap gains ? so, they will have to outperform the S&P 500 index to account for it.

great start, and good luck.

premium brands like coach, rolex thrive on exclusivity. same with the premier institutions. they want to keep their brand allure so someone can boast that they are from stanford, or that they own a rolex.

If every tom dick and harry has a stanford cert in their resume, the value of the brand diminishes. so, you are from stanford, big deal, i got a certification too.

fair, but is there are too many such courses, the exclusivity of the brand dilutes.
The same can be said of Stanford which charges ridiculous amounts for "certification" courses.

Basically cashing in on the institutions reputation and student's hunger to get the institution's stamp on their resumes.

Glad to see the pushback. eventually stupid money runs out.

the title can be restated as "outcomes are produced by incentives" and applies to anything including: college admissions, climbing the corporate ladder etc. any system is gamed by its participants, and so, the system design is critical to ensure whatever the system wants to sustain.

the design of disincentives (or punishments) is harder - coz you cannot jail a scientist for poor me-too publications, or papers that self-claim "an innovative approach to this or that"

Just as in corporate orgs, or societies, the culture that is fostered plays a critical role. Once rot takes over, it is hard to eradicate.