HNHacker News
TopNewBestAskShowJobs

cmrdporcupine

23,652 karma · joined December 5, 2014

Surfin' the Dunning-Kruger wave since the 1990s.

Member of Technical Staff, AI @ Verda https://verda.com/

some src @ https://github.com/rdaum/

submissionscomments
cmrdporcupine··on Tesla winds down Cybertruck production after years of poor sales in few markets
I just imagine what an incredible success Tesla would have had just making a "boring" Tesla branded competition to the F150. "Just" a model 3 with a standard pickup body.

Americans, for whatever reason, only buy SUVs and trucks. They've left the whole truck market untapped by creating a vehicle only a very small niche could/would want.

I don't understand what product marketing was up to, or if they just completely lacked the spine to stand up to Elon.

cmrdporcupine··on Tesla to wind down Cybertruck production
Elon resigning from Tesla and just letting it run as a reasonable car company making just regular incremental quality and price and new model improvements on normal vehicles people want to drive could be the best thing for the company. The Model 3 is a huge success. That they didn't follow that up with just a regular pickup truck in a similar niche, and instead got distracted by this cybertruck nonsense is an indictment of Musk's poor leadership at this point.

They should just be the Toyota of EVs -- mass produce quality vehicles people want to drive -- instead he's trying to be Tony Stark and keep pumping it as a meme stock.

Tesla remains excellent at engineering and manufacturing. They need not seek the exotic.

He's a liability to that company, not an asset, at this point. And his name has become a brand liability.

cmrdporcupine··on The Lisp curse, now coming to your ecosystem
Depends on the project but some actually adopt an outright hostile response to this. Sometimes justifiably. Sometimes not.

It's honestly not worth it. I understand both sides here, but people are going to have to come up with new ways of collaborating. Pull requests are likely not it anymore.

cmrdporcupine··on The Lisp curse, now coming to your ecosystem
IMHO that "career building" part of it was always a bit problematic because of the incentives it could add.

Instead of "here's some small bits of code you might find useful" or "let me share my approach", the personal glory building style of OSS construction can and has tended towards "behold my framework" and a brand, and often even a VC etc funded entity to go with it.

I don't love LLMs just conjuring half-assed things into existence, but I've also been burned too many times by someone else's frameworkization of something that should just be a small modular bit, or part of a standard library. Languages like Rust that have exempted a lot from their standard library that frankly should be there suffer from this the most.

cmrdporcupine··on The Lisp curse, now coming to your ecosystem
"I’d rather rely on proven libraries with well-defined abstractions than generate hundreds of lines of code which solve a local problem for me. “Not generated here” is just a different version of “Not invented here”.

I disagree generally, and the problem is two fold..

I've just seen "well-defined" abstractions go wrong too many times. Over- encapsulation, over-abstraction, or just overall over-cleverness in third party packages that has led me to spend large parts of my career just frustrated trying to glue third party dependencies together in a sensible way... and the abstractions only adding complication.

The joy and temptation of having an LLM "just make it" is that this goes away. The functionality fits my existing structure and abstractions and I'm not fighting a war with the architecture that the third party maintainer decided worked best for them. And getting only the parts I need, not your whole giant framework with it. (And yes, I know the LLM could easily hallucinate something far far worse, but let's just... play pretend here for a minute)

The second part is that since the dawn of Maven, NPM, Crates.io generally there has just been this insane explosion in programs of transitive third party dependencies out into the stratosphere, and the cognitive, technical, and legal debt of so many deps generally is not sufficiently recognized by the community. Running "cargo tree" in my project is often frankly depressing. Multiple linked MD5/SHA256 implementations, random number generators, HTTP protocols, people dragging in tokio deps where I don't want them, multiple versions of the same thing compiled and linked... It's exceedingly frustrating, and with the explosion of security concerns we've seen recently, it's also potentially dangerous.

I don't love the idea of having LLMs "just yeet it" esp if people misapply this to things like security/cryptography domains. But I'm also more than skeptical about the actual real world quality of many hallowed third party packages. I've been burned too many times.

Perhaps potentially what replaces "widely used standard implementation of X in a library/npm/crate form" is an "LLM readable recipe for X along with a formal proof / automated test scenario that I can point my clanker at to get a community proven way of doing this" along with a set of prompts, licensing/credit requirements, etc.

cmrdporcupine··on Bidirectional charging during grid outages
Also that the dealerships etc are super paranoid about any potential damage to the pack. They'll declare total loss even when there's no evidence of battery pack damage rather than risk liability for being wrong about it.
cmrdporcupine··on Bidirectional charging during grid outages
CATL is already producing sodium ion at competitive prices. Expect 5-10 years to end up with a flood of stationary battery products based on them, at prices less than LFP.
cmrdporcupine··on Bidirectional charging during grid outages
I live rural and I'm hoping the next decade will bring a wave of cheaper safer sodium ion battery options. Something I'd feel a lot more comfortable with putting in or around my home since they don't have thermal runaway issues, their materials are theoretically cheaper, and they can withstand colder temperatures.

While I'd like to use my EV as backup, the cost and integration issues right now are just not sensible. The utilities and manufacturers have made it too cost prohibitive.

cmrdporcupine··on Bidirectional charging during grid outages
CHAdeMO had that spec, but of course is obsoleted now.

If I understand things correctly tho many of the V2L solutions just work with pretty much any car that does CCS, regardless of whether the manufacturer officially supports this or not. My Polestar2 for example doesn't official have any kind of V2H/V2L option but I've heard of many success stories.

cmrdporcupine··on Serbia just elected the world's brainiest parliament
Yep, working class people's lives have sucked after 2008, though it was a continuation of a longer trend and is still continuing. (And COVID made it worse. All us tech people sitting comfortable in our home office while mass wealth transfer happened out of blue collar pockets...)

Worth remembering... Anti-globalization and anti-liberalization were slogans and policies of the socialist left in the late 90s and 2000s. Battle of Seattle, Occupy, etc. Voices marginalized and mocked by the neo-liberal centrist "left" (not left) in the Democratic party, Labour party, mainstream European govt's, Canadian Liberal party, etc. Manufacturing jobs were dumped overseas, union membership declined, middle class incomes plummeted, 2008 drained people's financial well being, while housing prices skyrocketed etc. And same "moderate" forces also went to war in the middle east, etc. etc.

Obama ran on left populist slogans but actually governed from the neo-liberal centrist right.

Actual socialist left voices and formerly moderate social democratic policies generally had no voice or power in government anywhere, Corbyn was run out of the room, Bernie isolated, etc.

So it's no surprise that the right and far right saw the vacancy and exploited it. The rise of Trump, of UK Reform, etc. ties directly into this sense of hopelessness and anger people had.

Now the "conservative" right -- which was formerly pushing liberalization the strongest -- talks in angry tones about "globalism" instead of "globalization", and it's caught up with anti-Semitic dog whistles, national chauvinism, etc. etc. That shit sells.

cmrdporcupine··on Serbia's Vucic concedes election defeat in historic win for student movement
So? Mark Carney is to the right of the centre by the standards of the G7. Just slashed corporate taxes, environmental regulations, doing privatization, federal service layoffs, and slashing labour rights. Only looks "left wing" to culture warriors. Sits right in the centre of European politics, same vibe as the CDU in Germany or Macron in France.

There's nothing "far left" about the provincial NDP parties. They're centrist neo-Keynesians, pretty pragmatic, they don't even call themselves socialists. Again, it's only smelling "far left" if you're obsessed with cultural politics or "woke" stuff and are frankly politically illiterate and never taken a basic political science intro course? I didn't seeing Eby nationalizing/socializing industries and crap. "Far left" is such a bizarre characterization that it boggles the mind.

The federal party goes through phases of more old school NDP/CCF social democracy. But never governs, so.

cmrdporcupine··on Serbia's Vucic concedes election defeat in historic win for student movement
"far left"

OMG. Peak hackernews comment.

cmrdporcupine··on Why is AI generated language so hard to understand?
My experience is: The LLM fails to "understand" or "intuit" who the audience is. It tends to reward its own recent context and its interaction with the prompt writer, rather than "considering" documents from the viewpoint of a hypothetical reader. You can often prompt them to go more in that direction but it gets tiresome.

The best approach is just to write the outline and initial text in a very messy fashion yourself. And then have it clean it up and add detail. And then edit again by hand.

cmrdporcupine··on Kimi K3.1 Pro and K3.1 Pro Highspeed
That's been done with the decision models, as a benchmark.

https://opper.ai/jevman-benchmark/

cmrdporcupine··on Canadians are running out of time to oppose Bill C-22
All of this can be true without stupidly calling it "state sponsored."

The US mainstream media is equally stupid -- "liberal" or "conservative" -- CNN or Fox both make me want to gouge my eyes out.

Those of us on the (real) left have been pointing out the problem with the broadcast media for decades. Concentration of ownership and private interests are the problem, and frankly the CBC is the least worst of the bunch here in the country.

Canada is currently under a mass nationalist fervour under which the Carney gov't is getting away with a lot. The thing is the stuff he is getting away with... is exactly the stuff the Conservatives also wanted to do anyways. So people on the right frothing at the mouth about the liberals just baffles me. He is literally implementing the full CPC platform from the last election.

cmrdporcupine··on Canadians are running out of time to oppose Bill C-22
It has the same problem that the US has, which is it is stuck in a two party duopoly, with both major parties sharing fundamentally the same political economic outlook. Just with different takes on culture wars stuff.

The only significant opposition to this bill has been coming from the NDP, which was decimated to a small number of seats by people piling into the Liberals because of nationalism and by its previous pathetic fealty to the Liberals in the last two parliaments.

cmrdporcupine··on Canadians are running out of time to oppose Bill C-22
what are you even on about, this is complete nonsense

nevermind half the population doesn't even pay attention to traditional "the media" these days

the bulk of "the media" in Canada is US owned, not "state funded" (which is only the CBC) and I haven't gotten any particular pro-vibes around this bill from them

cmrdporcupine··on Kimi K3.1 Pro and K3.1 Pro Highspeed
Yep, and often actually very expensive to use. More expensive than the frontier models because there's no realistic coding plan.
cmrdporcupine··on DeepInfer: High-performance inference framework for DeepSeek models
Anybody could fire up vLLM now to do exactly what you're talking about. And they do? There's dozens of places hosting DS4.1 Flash, at good speeds. So I'm not sure what your point is.

What this provides is the give contributors to vLLM, sglang, etc. a reference implementation of some of DeepSeek's own hard-won inference serving techniques.

No doubt we'll see patches to vLLM inspired by bits in here.

cmrdporcupine··on DeepInfer: High-performance inference framework for DeepSeek models
Not sure why you're being downvoted. It's very clear to me in the order that even the open weights would be forbidden, even served through vllm or sglang etc.
cmrdporcupine··on DeepInfer: High-performance inference framework for DeepSeek models
It's not US labs that it targets, but just US gov't civilian agencies.

... so far.

cmrdporcupine··on DeepSeek just open-sourced its inference engine
I don't see how this proposition holds. It's just part of an inference engine for serving DS4.1 Flash. Which vLLM and SGlang can already do. Less efficiently, I'm sure, but on much broader hardware, and with other models in the mix.

People will take the novel bits here and port them to the above. The community as a whole will benefit. Including infrastructure engineers inside OpenAI and Anthropic.

But a lot of what's in here is already in published papers out of DeepSeek already.

cmrdporcupine··on MiniMax M3.1 Flash
I mean, memory prices are high. They ain't getting cheaper. And KVCache is held in memory. The more people are using cache, the more memory you need. It's not free to cache it.

Plus there's complicated cluster-wide KVCache distribution schemes, built around RDMA networking, etc etc.

By far the most expensive hardware cost now of inference is the memory cost. KVCache is still using it. Though obv GPU compute is using more energy

It's not just like... "stick it in Redis, all done!"

cmrdporcupine··on Putting Linux on my RTX Spark laptop because I don't have a Microsoft account
If it's anything like my DGX Spark, the "premium" is almost 100% just the memory cost, not the GPU or ARM core.

10 months ago when I got my Spark it was competitive in price with a similarly spec'd out Mac Studio. As RAM prices have gone even more parabolic, so has its price. And so has the Mac's. And the Strix Halo competitors.

cmrdporcupine··on Mercury 3: Diffusion at scale
Crazy are the times when we consider it slow to spend 3 seconds to conjure a whole pelican into existence.
cmrdporcupine··on DeepSeek banned from U.S. federal civilian agencies under new DHS order
It's also odd for them to single out DeepSeek in particular, and not just any model from the PRC?
cmrdporcupine··on Mercury 3: Diffusion at scale
How fast did they generate? That's the specific sell with this model

EDIT: I see, "3337ms"

How does that compare to your other artistic works?

cmrdporcupine··on Mercury 3: Diffusion at scale
"Quality: On par or better than frontier speed-optimized models like GPT-6 Luna and Claude Haiku 5.5. Speed: 1,068 tokens per second on NVIDIA GPUs. Context: 1M tokens, up from 260K for Mercury 2.5 Price: $0.20 per million input and $0.70 per million output tokens. At launch, Mercury 3 is 50% off at $0.10 per 1M input and $0.35 per 1M output. Cached tokens are 90% off the input price."
cmrdporcupine··on DeepSeek banned from U.S. federal civilian agencies under new DHS order
Thing is, is it seems to say a ban on use even when the weights are hosted via other providers.

"third-party products and services developed using DeepSeek implementations"

So you could be hosting it fully locally inside the agency and still it would be banned.

It's not surprising to me, but it's pretty blunt. Not that I care what US agencies do, but this does seem like the first step of many to come.

cmrdporcupine··on How the Claude Code team does code review
> you know it's at least coherent enough that someone understood it

30 years working in this industry and I'd say that's the exception rather than the rule. When it comes to large systems.

← PreviousPage 2 of 34Next →