HNHacker News
TopNewBestAskShowJobs

horsawlarway

11,569 karma · joined May 23, 2017

submissionscomments
horsawlarway··on Disney Raising Prices Days After Confirming Ads Will Now Show Up on Every Tier
Yup - my take is this is coming for absolutely everything. (it's already mostly here for tv).

Rent-seeking behavior is too good to give up, so companies will move all content to streaming only services (games, movies, tv, music, books, etc...) in the majority of cases.

Physical media will only be available at very high prices, for a very limited selection of content.

We will go from "doctrine of first sale" around content licensing to the utterly crippled "you own a temporary, non-transferrable, and ephemeral license to stream this content from us on exactly this service using only approved devices until we decide otherwise".

The moral action in this case is to simply stop paying. Period.

Long term - we should be fighting to enshrine a new flavor of "doctrine of first sale" for the digital age.

Ex - maybe purchasing any content license sold with the word "buy" should enshrine your right to use, transfer, sell, or otherwise consume in any reasonable form that content for the duration of the copyright + 7 years. And we will not enforce copyright for items that can't be bought (only rented).

horsawlarway··on Woman charged with computer fraud due to scam CashApp account being in her name
Or we can all just agree that maybe basic transit in a city should be subsidized, maybe say... To the same degree that road construction is?

At which point no one would need to pay a fare, and we'd all still save money trying to prosecute broke teenagers trying to move around a city?

horsawlarway··on DeepInfer: High-performance inference framework for DeepSeek models
This is a pretty strange take on an article explicitly pointing out the profitability of chinese models.

Further, even if this is your take (and I won't touch the geopolitics here, maybe you're right, maybe you're not regarding ulterior motives) - go use non chinese open models that are also good.

Laguna, Nemotron, Mistral, Arcee, Llama, etc...

horsawlarway··on DeepInfer: High-performance inference framework for DeepSeek models
I know this gets a lot of flak because it ended up being obviously incorrect for consumer markets.

But the under-reported corellary of this is that a TON of industries essentially operate on hardware right out of that time period (or slightly after, circa 2000s)

banking

Insurance

Healthcare

Gov

retail

Transportation

etc...

Hell, I was writing code to run on IBMs Z/arch as recently as 2021 for banks with mainframes deployed around 2004. They were just then finally considering an upgrade to the later models.

horsawlarway··on DeepInfer: High-performance inference framework for DeepSeek models
define "we".

Because again - I think lots of businesses are already taking a good look at models well off the "cutting edge" because they're good enough for what that business is trying to do. They're already there.

If you're doing categorization, tagging, form-filling, customer support routing, etc... you don't need anywhere close to cutting edge expensive models.

Hell - Jev being so popular is basically exactly that group going "what's the cheapest way we can make this decision".

Meanwhile, there are plenty of spaces where you still want real model improvement. But a lot of traditional business was basically process and paper pushing. You don't need a Phd level model for that, you need an intern level model.

---

But that's the trap for the large providers - intern level models are cheap and easily available with current open weights. It's going to be a very tight market, and margins will go down to essentially the floor (especially once we exit the current constrained-compute market).

Somewhat ironically - the technology itself makes switching costs basically zero. It's not just that there's no big lock in, AI is like an anti-moat. It actively fills in and paves over whatever moat you might have for price conscious customers.

horsawlarway··on DeepInfer: High-performance inference framework for DeepSeek models
Nobody meets expectations of a 2 trillion dollar IPO selling a fungible commodity like tokens for old models.

And that's the rub - it doesn't matter if inference is profitable, the US labs have promised the whole kitchen sink, and taken serious investment on that pitch. They won't make it back in a cut-throat market for inference. They have to continue pitching the idea that only their models are worth using, for quality/safety/performance/etc reasons.

Which means continuing to develop and release new models. At huge expense.

Side note - while deepseek seems to be doing fine, I'll note that the emphasis on cost effectiveness is very much not a thing for the US labs right now... and it was for non-us labs. I'd be quite interested in knowing if anthropic is pulling in the same returns, or what the margin on their subscription plans actually is.

---

I'm already seeing this directly at work - lots of devs still happily use claude for development, but the products we're making aren't. They'll use kimi, glm, gpt-oss, etc... basically - if cloudflare is hosting it, we're exploring it (mainly for contractual/security reasons): https://developers.cloudflare.com/workers-ai/models/.

Because we are price sensitive, and we already have to do all the work of creating very robust evaluation frameworks for our products, so we might as well choose the cheapest model that meets our standards.

whiiiiich is why we're seeing things like Haiku and gpt-luna priced down near the open models at ~.10/mm in, ~.50/mm out. They've been forced to cut prices to keep commercial customers on their cheaper models instead of moving away.

This will continue. If inference is profitable and easy, prices will continue to erode. Go browse openrouter, plenty of providers happy to undercut the competition. Bigger shops with auditing requirements will avoid them, but smaller ones won't.

horsawlarway··on How employment is changing at Anthropic
Your last point is the one that I find most compelling.

Humans are autonomous, distributed, and adaptable. In a package that weighs ~150 lbs, with very low energy requirements.

You literally cannot wipe them out without a sustained and incredibly expensive ground campaign, and even then it's not likely.

AI in a data center on the other hand... So easily disabled.

horsawlarway··on Tesla Roadster
Yeah, and even with all the qualifications here, the man is still an absolute asshole.

All you need to do is see how frequently he posts on Twitter to know he's broken and trying to fill the hole with money and attention.

horsawlarway··on Redesigning source control at Anthropic
No.

Because tokens have a cost. CI jobs have a cost. Organization objectives and goals have a cost. Features have a cost. Code has a cost.

---

So in a world where agents do all the work... why are they not able to control operating costs AT ALL?

Why is the right operating method the one that happens to blow an absolute shit load of cash on - wait for it - internal tooling that the customers don't give a fuck about?

Customers do not care how the internal business operates. They care about value delivered. I'm absolutely not convinced that this "churn churn churn" method of operating is actually delivering enough extra value for the cost it incurs.

---

Now - that said, I think there are some fair points here that are worth considering, mainly around whether a distributed version control system (git) is really the right call for agents that need to coordinate much more closely than most human developers did.

I think that's a directionally appropriate insight, and worth considering.

What I don't find compelling at all is the idea that many (any, really) companies should expect to be running millions of ci jobs a day, and mergings hundreds of thousands of prs/commits. That reeks of "I have cash to blow and don't care".

horsawlarway··on Redesigning source control at Anthropic
I genuinely don't understand the core value they're pitching here. The tech is interesting, and I'm not trying to under cut that, but I question the idea that this is directionally valuable advice to nearly any company.

This amount of churn feels like a thing only possible with no real profit motives and a lot of money to waste.

Lots of activity is not necessarily synonymous with productivity, and it carries real cost.

So I think the design here is interesting, and the concepts seem applicable to where some agentic development may end up, but I question the underlying principles that drive this being the right solution space at all. At least if the goal is a profitable company.

horsawlarway··on 'AI' Is Not a Tool by Rich Hickey
So functionally, this take is mostly incorrect.

You usually use entirely different techniques for hand vs power tool.

The way you'd do the work, the methods of joinery you'd choose, the style and choice of cuts, even the preferred size of boards - all completely different.

Learning hand tools is an entirely separate skill set. Down to the planning stage.

It's not merely doing power tools more slowly. And trying to use power tools like they're merely "faster" hand tools also doesn't work. You waste a ton of time because it's just a different operation.

horsawlarway··on Solar panels that plug into your wall
I'm in the US and I did my own install.

It's the regulation cost that's really hurting us.

Just the required paperwork for the permit in my region cost ~2.5k. structural engineer was 800, architecture diagram was 800, certified electrician stamp on the plans was 800.

Add in the extra cost to facilitate RSD (~5k after mlpe devices, and extra wiring) and tack on the cost for utility interconnection agreements.

Then deal with the NEC requiring UL listed items with no/very little cross certification with things like CE which effectively means you can't buy equipment from China, at least not without a US shop taking a huge markup for branding/certs.

---

Overall, it's just a mess. I walked away from the project (completed, all approved) with a huge dose of skepticism for the permitting/certification/approval process.

11.2kw in panels, ~12kwh lfp battery, split phase 240v install - ~28k usd.

Of which ~10k was regulation driven, and doesn't change much by install size.

horsawlarway··on 'AI' Is Not a Tool by Rich Hickey
> And similar to with coding, you've lost things from automating the reading, that reading just a summary does not provide.

I actually find this a very interesting line of discussion, because in many ways it mirrors the transition from hand tools to power tools.

Ex - you're often trading the finesse and control of a hand tool for the speed of a power tool.

The way you process and deal with problems shifts, because certain things become much harder, and certain things become much easier.

horsawlarway··on YouTuber Says Cops Visited Him After He Built a Flock-Style Camera to Track Cops
If we're talking about license plate numbers, it seems fairly reasonable to just have the device download the entire target list.

Longest plates are 8 chars. It seems exceedingly unlikely that the target list is longer than 1,000,000 plates. They all use characters that fit in a single UTF-8 code point. So 8mb initial download, and very small incremental updates.

Likely much more power efficient than sending every plate over the network, and would use considerably less bandwidth over a couple months.

---

Is that how they do it? No freaking clue.

But it seems like a very solvable problem, with correctly aligned incentives (ex - it's both cheaper and more private to NOT send the plate over the net, vs sending the list to the device).

horsawlarway··on Court throws out killer's sentence after judge said he loved AI video of victim
Sure, not a jury recommendation in this case. But it absolutely is supported.

The whole resentencing judgement came in part because the original judge stated how much he liked the ai video of the victim.

horsawlarway··on Court throws out killer's sentence after judge said he loved AI video of victim
fair - I've lived in places where juries can recommend sentencing. You're right here.
horsawlarway··on Court throws out killer's sentence after judge said he loved AI video of victim
> The attorney thinks the best way to get the maximum sentence is to make an AI video of the victim asking for leniency in sentencing?

Ab-so-fucking-lutely. This paints the victim in the most positive light. And makes killing them seem even worse.

Again, attorneys understand that a lot of times you're looking for an emotional hook, not straight up "facts". This video made the victim seem moral, empathetic, full of forgiveness... as close to freaking jesus as you can get.

And that other person killed them.

Now go sentence that person.

---

Instead imagine the video was of the victim smoking a cigar, laughing, and suggesting that he'd never forgive him and he should be put to death?

---

Which video makes the jury more likely to actually come back with a tougher sentence? The one where he's a bastard, or the one where he's a saint?

horsawlarway··on Court throws out killer's sentence after judge said he loved AI video of victim
You assume the intent of the video was to be listened to and accepted as actual input.

My assumption is roughly the opposite - the intent of the video was to prejudice the jury as far as possible against the accused (he killed a person honorable enough and empathetic enough to suggest a more lenient sentence! To the pit with him!)

It wasn't about the content of the video - it was about the emotional response to an AI generated facsimile of the victim.

horsawlarway··on Ask HN: What do you run on a $5 VPS that's worth keeping online 24/7?
bookstack

keycloak

immich

jellyfin

several personal sites

home assistant

unsloth studio

several synology tools on a nas (drive/security/backups)

etc...

If it's "just you" your laptop is likely fine. Hosted tools really start to shine when you've got a group of folks using them. Ex - Wife/Parents/Kids/Extended Family.

Simple k3s cluster and the shallow end of kubernetes actually saves you a ton of time and effort (don't go into complex orchestrators or helm charts - just write small deployments in static yaml). You can get to roughly the same spot with just docker/compose if you want, but the capabilities cap out earlier (ex - multi-machine, resilient to any one machine going down, gpu sharding, etc)

Maintenance is basically "sudo pacman -Syyu" on a cron job.

horsawlarway··on PS5 Jailbreaks Are Escalating at an Unprecedented Pace
> Markets find equilibrium based on demand

This is decidedly untrue non-fungible item markets (ex - the market where you can only play Sony exclusives on a playstation).

> in markets for non-fungible goods - price formation and demand equilibrium do not rely on a single, continuous market-clearing price. Instead, they rely on discrete matching, asymmetric information, and social coordinates.

Basically - no. Fuck off. Sony is acting in their best interests with asymmetric information (ex - they know how much profit they make of those game sales and you don't, they know how much profit/loss they make from that playstation sale and you don't).

You know how we know? Because absolutely nothing is stopping you from still opting into a model where you don't pay costs up front.

> The problem is assuming everyone wants the hardware costs upfront model.

We have an absolutely smorgasbord of financial services that would LOVE to let you pay over time, or give you a moderate loan, to buy a playstation with low upfront costs... to just list some of the bigger ones: affirm, afterpay, klarna, paypal, zip, bread, upstart, albert, withcherry, etc... there are SO many of these.

horsawlarway··on PS5 Jailbreaks Are Escalating at an Unprecedented Pace
By your own argument:

Sony prefers charging a tax on game sales that's high enough to recoup the cost of the hardware discount.

This means playstation owners are getting a discount on hardware, and then paying so much more for each game - to the point Sony prefers this because they get higher revenue than just selling the hardware at cost.

---

You are literally advocating for the worse model for yourself.

Jailbreaks don't ruin anything other than a "predatory loan" style arrangement.

Personally, I fail to see the problem.

horsawlarway··on Qualcomm licenses patents on Huawei’s LogicFolding chip tech
I think it's a more complicated discussion.

My read is that the US has over-indexed heavily on "intellectual property" and that position is profitable, but an overall net drag on innovation and research.

China is much less indexed on IP, and often outright ignores it. This allows them to be much more competitive in innovation over the long term, but at short term reputational damage.

---

From the US side - if you're a company operating in a restrictive, patent encumbered space... it absolutely feels like an unfair field. Those other companies are able to ignore the restrictions you're hemmed in by (steal).

From the China side - if you're a company operating outside of compliance with licensing and IP, you get a reputation for "just stealing" even when your research is novel and impactful.

Of the two - I believe the US is "more incorrect". The current IP laws feel a lot less like encouraging innovation and a lot more like government enabled and funded rent-seeking.

horsawlarway··on Why don't more developers “use the platform”?
QT base alone is 66mb on arch, not incl deps.

Chrome is ~100mb.

They're right along the same order of magnitude for size. Install size has very little connection to standards/api complexity. It's just a bad argument.

The right discussion (which you're nicely skipping by talking about install size) is "how does the complexity of QT widget selection compare to html?"

And... I find the spaces incredibly similar - this feels decidedly like browsing html components: https://doc.qt.io/qt-6/widget-classes.html

QT gets the perks of being able to remove deprecated items - so I fully expect theirs to be a more minimal set. That's a luxury not afforded to browsers.

So again...

---

> What the system gives is exactly a canvas to render stuff on.

Render fucking how? The deliverable wasn't code, it was an html document. That's the defining history of browsers.

horsawlarway··on Why don't more developers “use the platform”?
Look - even coming from a spot where I mostly agree that the end spot isn't wonderful and the standards are probably too complex, I think this glossed over the "conceptual shift" that has happened for browsers.

Browsers got complicated because they weren't seen as a platform to ship code to... They were seen as document renderers.

There wasn't anything to "compose with" for a meaningful history of browser development. You shipped a document (html) with some optional styling, and the browser handled EVERYTHING.

JS was a very minimal, underpowered, runtime that was decidedly NOT the thing to use for any real composition that resembled backend/desktop coding. The tools of the time for that were flash and activeX and silverlight.

Javascript was just a thing you used if your checkout form needed to be more complex than normal, or you wanted to show off a bit.

---

Then Ajax came along in the early 2000s, sites got increasingly complex and interactive during the web 2.0 push, js got progressively more powerful (to the point where the js runtime is functionally not that far off a real OS in complexity and capability) frameworks hits the scene, and people finally had the right tools to actually implement a single page app using web tools.

---

So...

1990 to 2005: Browser is solely a document renderer, every new cool thing has to be implemented by the browser.

2005 to 2015: Browser is mostly a document renderer, but you can do cool new interactive things, and you needed to paper over lots of rough patches with js.

2015 to 2025: Browser can ship meaningful applications that run entirely in the client, and the "document" model has faded a bit (although it's definitely NOT gone) while new complexity largely lives in expanded system access and capabilities (ex - usb access, crypto, complex networking, async processing, web-assembly, etc).

---

The challenge is that people who want the "document" model still want it all implemented by the renderer, and there's enough historical momentum there that you basically can't undo it without starting from scratch to deliver applications instead of documents.

Further (and this one is entirely on you) the real comparision isn't "unixes" it's a graphical env. So compare the complexity and attitude to win32 dialog apis, QT, etc...

Often they also ship with a huge library of components built in. Because the users of those systems don't really want a canvas they have to render everything on... rendering is hard. Rendering performantly is really hard.

horsawlarway··on Lawmakers look to require AM radio in new cars despite automakers' pushback
Yes. The intent is not that AM radio be "free" in the abstract. It's that charging for it means it's an optional upgrade.

One which consumers likely won't buy.

They don't care if the sticker price goes up by the cost of the am radio. They care that manufacturers install it, and don't make it optional.

And they don't care because it's like $5 worth of parts (honestly probably less).

horsawlarway··on The End of a Fair Price: Dynamic Pricing and the Normalization of Gouging
> Sure, everyone would prefer to have all the stuff they want for low or zero cost. That’s Santa Claus economics.

This - this again implies you're not actually engaging in the conversation. I have zero interest in "free" things.

I do have interest in their being at least some real negotiation between parties signing a contract.

For pretty much all consumers - there is no negotiation. The contract says what the company wants it to say, and if you don't like it - fuck off. Same is mostly true for small business as well.

And yes - my WHOLE comment above is pointing out that often... they are literally the only provider. Or one of two or three providers that have essentially identical terms. (See - Apple and Google for mobile).

---

So circling around again to the point - My stance is there are two functional paths forwards:

1. Break up the offenders into small enough pieces that meaningful competition around service levels, pricing, terms & conditions, and features exists again. When we go back in time - arguably this was the drive that ended the age of "Robber Barons" - laws like Sherman Anti-trust were a direct response to concentration occurring in the early to mid 1800s in railroads and oil.

To directly quote Sherman: "If we will not endure a king as a political power we should not endure a king over the production, transportation, and sale of any of the necessaries of life." Ring familiar when applied to Apple/Google in mobile today? (Be honest....)

2. Nationalize the fuckers. You and I probably disagree quite a bit here - but this is the route I'd argue is more productive long term. If a monopoly is natural (aka - market forces make it hard to avoid) then the people need to own it or regulate the absolute hell out of it. See - utilities.

---

But no - continue blaming me for wanting "santa claus" economics instead of actually engaging or participating in the conversation. I'll leave it here. I don't think we're really conversing so much as you're ignoring what I'm saying and focusing on words where your presumed definition matters more than the context in which the discussion is happening.

So no - I don't want Santa Claus. I'd much prefer strong labor unions, a focus on consumer rights, and a political party in the US less interested sucking billionaire dick. Historically speaking - we'll get there, but its going to get ugly first.

horsawlarway··on The End of a Fair Price: Dynamic Pricing and the Normalization of Gouging
Go actually read Kreitner - then consider the standard EULA you're required to accept for all services.

Consider whether his argument is at least a valid description of the issue of "free people transact[ing] business" when one side is literally the ONLY service provider, and the other is a utterly disposable customer.

---

Your comment implies you don't actually understand the discussion.

horsawlarway··on The AI Race Just Got Awkward
Right, which is why I'm happy to see China continue to innovate in the open, and increasingly wary of the US stance given articles like

https://www.anthropic.com/research/glm-5-3-and-the-spread-of...

It's VERY clear that the US companies are trying to push for regulation to kill open models and open weights. I see this as much more hostile and authoritarian response than what we're seeing come out of China right now.

So is China going to always publish in the open? No clue. But right now they're modeling much better behavior.

horsawlarway··on The AI Race Just Got Awkward
Yes, we already discussed the US.
horsawlarway··on The End of a Fair Price: Dynamic Pricing and the Normalization of Gouging
Yes - because "additional regulatory burdens" aren't the answer... breakups are.

Somehow we've settled into a spot where it's acceptable to have incredibly powerful monopoly/duopoly structures in American culture, sometimes with a "third wheel" of a small scattering of options to make it appear as though there's competition. Usually, the terms these companies operate under are substantially "the same" from a consumer point of view (whole related can of worms there - see: "tyranny of contract - starting with Engels and leading to Kreitner")

Want a computer? Choose: Microsoft or Apple.

Want a phone? Choose: Apple or Google.

Want to advertise? Choose: Google.

Want a credit card? Choose: Visa or Mastercard (third wheel: Discover/Amex)

Want a browser? Choose: Google (third wheel: Firefox/Safari/Edge)

Want a grocery store? Choose: Walmart or Kroger.

Want internet? Choose: ATT or Comcast.

Want to watch TV? Choose: Comcast or Disney (third wheel: Fox/Paramount/WB)

Want to stream a show? Choose: Disney or Paramount (third wheel: Amazon/Netflix)

Want to bank? Choose: JPMorgan Chase or Bank of America (lots of third wheels still, but they hold more than the next 6 combined in assets)

Want a political party? Choose: Republican or Democrat.

Want to sell a good online? Choose: Walmart or Amazon.

Want to do home improvements? Choose: Home Depot or Lowes.

---

It's not about regulatory burden, its about a clear failure to provide competition by allowing a small number of dominate players to functionally own a market.

They're just competitive enough to shut down any real competition (by undercutting them via scale, cash-on-hand, and monopoly effects) but the impact is short-lived.

And I partly blame our response in 2008 - we appear to have decided that it entirely ok to be "too big to fail" and that's an acceptable state of the country.

And it's definitely "efficient" (in some respects - much like cancer is incredibly efficient at replicating and capturing resources) but I don't think it's particularly healthy for society. A very small number of folks are sucking up more and more wealth via a set of largely entrenched conglomerates. Meanwhile - any feedback mechanism or regulation that might stop that siphoning has been knocked away.

It feels like we had a nice breeze for a bit, but we're headed directly back to the "Robber Baron" age.

← PreviousPage 2 of 34Next →