Hacker News Reader

mbestoon A $500 RL fine-tune of a 9B open model beat frontier models on catalog review

This is basically the entire argument by Richard Sutton happening in real time:

https://www.youtube.com/watch?v=21EYKqUsPfg

Right now, we keep creating new models with larger and larger weights and throwing hardware at it. His argument is that this is a dead end and ultimately we'll eventually go back to purpose fit algorithms like we've always done in the history of AI development.

mbestoon Stripe and Advent have made a joint offer to acquire PayPal – sources

> Visa or MasterCard is making the bulk of the commission

They are not. Payment facilitators (like Stripe) are.

https://images.prismic.io/sacra/30d7bd58-8149-461e-a3c2-8a4a...

mbestoon Zig Creator Calls Spade a Spade, Anthropic Blows Smoke

You can't unilaterally say all AI blogs are low quality. That's bullshit.

> would spend more time writing than I would have to spend reading

To your point however, the reason people don't like AI generated blogs is because there is a explicit recognition that the author of the blog lacked effort. There is a visceral response for the reader about the social contract "if you didn't spend as much time as I did why should I care about what's written here", it's however NOT that the quality is inherently poor (but perhaps one my insinuate that notion).

mbestoon Anthropic says Alibaba illicitly extracted Claude AI model capabilities

> The proprietary models are better because of proprietary data

Source? Otherwise this is pure speculation.

mbestoon Ubiquiti: Enterprise NAS, Built on ZFS

https://app.opencve.io/cve/?q=vendor%3Acisco

What about Cisco?

mbestoon Nextcloud Hub 26 Spring: Built together, designed for the future

What does that have to do with scalability?

mbestoon Nextcloud Hub 26 Spring: Built together, designed for the future

> Nextcloud is written in PHP, which doesn't sound scalable to me.

Lol what? Facebook (pre Hack), Tumblr, Wordpress, Etsy...

mbestoon xAI is looking more like a datacentre REIT than a frontier lab

And they are planning (well "planning" if you believe Elon) to start building their LLM over from scratch, which means they need a HUGE ass training data center, i.e. not a data center for inference to do so.

mbestoon Age verification tech could put children at greater risk, says think tank

The first link is the FTC punishing Pornhub for something they did wrong. That is the parent commentator's main point (e.g. it can be regulated), which you keep ignoring.

> You wouldn't say we should stop age verifying people for online gambling just because there are "dark web" gambling websites.

False equivalencies. Gambling and porn have VERY different dynamics to them.

mbestoon Anti-social: It's fads, not friends, which now dominate social media feeds

> trip also knows where I live and vice versa.

Do those people also have access to your travel schedule? Mine don't.

Maybe you're just not as globally social as me? I've lived in 5 different countries and have friends all over the world and in probably 20 different US states that I can name off the top of my head.

Do I have close friends that I regularly contact? Do I send them a message when I'm in town to see if they are there? Absolutely. But it's not mutually exclusive with a cohort of people I will link up with when I'm traveling.

> well enough to link up

It seems bizarre to me that you only limit yourself to these people. I regularly try to meet up with people I don't know super well but want to get them or their city better. Social media has absolutely helped facilitate this.

mbestoon Anti-social: It's fads, not friends, which now dominate social media feeds

There are legitimate benefits. I just think its very easy to argue (which I agree) that the benefits don't necessarily outweigh the harm for most people.

mbestoon Anti-social: It's fads, not friends, which now dominate social media feeds

I am very critical of social media but this is far too of a myopic take. There is a ton of real life social benefit to these platforms.

Simplest example - someone posts a picture/video of them in a city that I also am in and now I know they live there / traveling there and I can meet up with them.

mbestoon Artificial intelligence is not conscious – Ted Chiang

> It takes humans a really long time to accept the fact they are no longer as special

We ARE special because we determined we are special. Name another animal that can determine it's also special or more special than us?

mbestoon Launch HN: Rudus (YC P26) – AI for concrete contractors

Travis Kalanick had zero taxi experience and Brian Chesky had zero hospitality experience.

Now they created new models to existing paradigms, because I do tend to agree that founders that have verticalized experience tend to be far more successful (but perhaps arguably less 'disruptive')

mbestoon Alphabet announces $80B equity capital raise to expand AI infra and compute

(1) There is currently $1T in dry powder in private capital markets

(2) Middle east oil money (Saudi Aramco's profit every year is $100B+)

(3) Public traders have been and are looking to cycle out of other investments into higher growth areas.

mbestoon Accenture to acquire Ookla

Undercounts it by what barometer?

mbestoon Accenture to acquire Ookla

> I’ve had sales people tell me to my face that they are the most important part of the business and the actual product or services is not that important.

Same here. This happened when I was 22 years old I didn't want to believe it.

BUT, I've seen far more shitty codebases win marketplaces with strong sales & marketing, than I've seen stellar codebases with shitty sales & marketplaces win marketplaces.

It's unfortunately really that simple.

mbestoon Accenture to acquire Ookla

Or https://speed.cloudflare.com/

mbestoon Accenture to acquire Ookla

> but it took me 18 years to grow the user base , figure out entreprise sales strategy and exit.

The audience here has never wanted to admit that the codebase doesn't really matter. Now that codebases can be created in a weekend, people are opening their eyes to this sentiment - the hard part is the sales, the code is easy.

mbestoon Robinhood now lets your AI agents trade stocks

This is essentially what AlphaZero is doing: https://nof1.ai/

mbestoon I think Anthropic and OpenAI have found product-market fit

> And I didn't even say "profitable", I said "credible business model".

What's the difference? I would assume a business model that rewards profit (and therefore can sustain) would be considered credible...and not much else.

mbestoon Private equity bought America's essential services

> would people stop doing PE even though it's profitable?

No but there would be a lot less PE dry powder available. To the parent's point - there is currently a trillion dollars in dry powder that is allocated to acquiring businesses. If that trillion dollars drys up then less businesses get acquired - it's that simple.

mbestoon Private equity bought America's essential services

> As long as customers choose services based on quality.

If the market doesn't reward this then maybe quality isn't important to the customer. Could be price, location, availability, etc. - PE can absolutely create that value even when they roll up 70% of your local HVAC market.

mbestoon I think Anthropic and OpenAI have found product-market fit

> My core point is more that April 2026 was the point when Anthropic and OpenAI finally appeared to have figured out a credible business model.

How so? What's specifically changed? We still don't know what their unit economics are and everything you've documented is basically speculation at this point.

mbestoon I think Anthropic and OpenAI have found product-market fit

> but I suspect there’s a more important factor here: I think they’ve finally found product-market fit

Ahhh the classic startup term that's definition is nebulous. But also, since when does any definition of product/market fit mean a product is profitable? And profitable in what sense? Unit economics? Overall company?

mbestoon The real cost of owning a home

Hi - homeowner here who is NOT in an fixer-upper. "every single weekend" is clearly hyperbole however here is regular maintenance for me:

- once a quarter clean the pollen off the A/Cs

- once a week clean the pool (okay I pay someone but still its part of maintenance)

- inevitably something on the pool goes wrong

- once a week clean my grill

- Blow leaves (seasonal) off porch

- pull weeds

- power wash the deck

And then most homeowners, no matter how new your house is, inevitably find an endless amount of "projects" to improve their home experience. Wife wants a table for the grill made, I add automated sprinklers, we put new planters for a garden in our backyard...the list goes on..

mbestoon Why Japanese companies do so many different things

> is because the author claims Japan's success in precision parts is driven by 'horizontal' and 'collaborative' practices. That just isn't true.

> In reality, this system is largely sustained by the ruthless squeezing of subcontractors (for the record, I am Korean, but I actually like Japan), which is a massive social issue there.

Well that's just like your opinion...man. I think you're both singularly wrong. Trying to attribute a single factor to a highly complex system is a fool's errand.

If your conclusion is correct "ruthless squeezing of subcontractors" are there other cultures whether there is true and the country has been successful in precision parts? Otherwise, it's pretty impossible to conclude the causality.

mbestoon Was my $48K GPU server worth it?

> I could easily say that everyone who says its profitible is msking unsubstantiated claims lol.

And people who don't understand the difference between capex and opex are making uneducated claims. It's not basic math.

Running an inference data center is a mix of variable and fixed costs. The fixed costs are currently in the billions of billions of dollars for pretty much any investment in this space. Many of those fixed costs have (currently) unknown refresh cycles. So, unless you have access to the financial books of these companies it's currently just speculation whether inference is profitable.

mbestoon Flipper One – we need your help

Did the PR achieve it's stated goal or not? Thats what we should be focusing on.

> because one of the main issues with AI content is the ability to generate an overwhelming amount of it.

So then let's focus on that, and not whether it's generated by AI. Yeesh you people are hard to please.

mbestoon Google changes its search box

I know, but if Google was replacing their search with AI today doesn't that mean this should already be happening? otherwise is google going to make zero revenue starting today?