Hacker News Reader

meltynesson Ventiva's fanless ionic cooling engine is higher performance and more efficient

I wonder if the stuff they expel smells like ozone. Definitely a compelling approach to explore.

meltynesson Curl removes experimental HTTP back end in Rust

I've reached some mechanical sympathy with it, I think.

You get some intuition for which methods to use, many structs have a pretty breadthy pallet, and then you can mostly just ignore it; quickly write your algorithm and let the compiler walk you through any oversights with the way you may have coded processing tasks.

Also helps to know that the docs support search by type signature since functions operate on or return owned, borrowed, or mutably borrowed, and generally just think of borrowed as a pointer for heuristic reasons.

It makes more sense than I do, usually.

meltynesson Curl removes experimental HTTP back end in Rust

I could see rust catching on as a shell scripting language too, honestly. Well documented, typed abstractions over shell utilities is really quite nice.

meltynesson Curl removes experimental HTTP back end in Rust

But if you can manually identify an invariant inside an abstraction it can greatly improve performance for callers/users, additionally, tools like Kani use comprehensible macros to facilitate automatically proving safety of `unsafe` code. Not to mention built in linting, package management, docs, and FP that rust/std provides. Lots has been said about unsafe rust, but the most basic libc tools require the whole cascade of upstream callers to check safety, it's basically backwards from the ground up from a resources and an outcomes perspective.

meltynesson Half My Life with Perl

yes, trailblazing.

yes, enables brevity.

yes, probably comprehensive offline std docs.

,...

https://perldoc.perl.org/perldiag should set off some alarms.

meltynesson The Curse of Recursion: Training on generated data makes models forget (2023)

My intuition given the rapid, informal developments of agent-type systems is that this is obvious insofar as the initial dataset was formed from a huge hidden "data cleaning" task that was human evolution and society. This isn't really that interesting of a claim and is it clear that it holds if you simply loop the LLM back onto the data cleaning task itself as a critic to the new training set? Is this what the author would classify as fine tuning?

Another question is what is the interpretation of the output of an LLM generation when unprompted? Isn't that always effectively garbage when there's not a deliberate bias in the training set?

meltynesson The Curse of Recursion: Training on generated data makes models forget (2023)

On the arguments involving modeling inference as simply some function f, the specific expression OP used discounts that each subsequent application would have been following some backpropagation and so implies a new f' at each application, rendering the claim invalid.

At that point, at least chaos theory is at play across the population of natural language, if not some expressed, but not yet considered truth.

This invalidates the subsequent claim about the functions which are convolved as well, I think all the GPUs might have something to say whether the bits changing the layers are random or correlated.

meltynesson OpenAI Reinforcement Fine-Tuning Research Program

I assume it's more like scaled nlp, which sort of describes the whole thing to begin with. i suspect it will boil down to further generalizing nlp-in-the-loop algorithms, more Tools, Tools between Tools, presumably Expert mixtures or randomly selecting "Axioms" and having an expert forget one and seeing if what remains makes sense still as the Tools are operated, and how that can be encoded better across domains.

It's not nothing, but there's a lot of value stuck up in there, I mean, it's made out of people.

Real special, takes a lot of smart

meltynesson Computer Networking Fundamentals – Learning Series (2023)

GNS3 work pretty good

meltynesson Google reacts angrily to report it will have to sell Chrome

V8 and jQuery won the web, really.

meltynesson Convolutional Differentiable Logic Gate Networks

Differentiable relaxation makes it sound like Rowhammer backpropagation, but I'm fairly certain that's not what they're talking about.

meltynesson Porygon Was Innocent: An epileptic perspective on the infamous Pokémon episode

An optimistic take is that the organization was simply too incompetent to identify and assess the financial risk of injuring someone.

meltynesson Microsoft still not said anything about unexpected Windows Server 2025 installs

ugh, who would

meltynesson Researchers discover a new form of scientific fraud: 'sneaked references'

"I'm In!"

meltynesson Google is killing infinite scroll on search results

It's hard to blame them, most of the well monetized expansive web content is deep web, pay walled, brittle, mixed media, copyrighted, and not enabled for something like microtransactions, subject to GDPR, and doesn't make them a dime.

meltynesson Study identifies first drug therapy for sleep apnea

  Key multiplicity-controlled secondary end points included the percent change in AHI and body weight and changes in hypoxic burden, patient-reported sleep impairment and disturbance, high-sensitivity C-reactive protein (hsCRP) concentration, and systolic blood pressure.

hmm... good?

meltynesson Ask HN: How do you handle the stress of impacting users?

I make this horrible unpleasant grunting and grinding sound, particularly as there's a chance to jump from the dock to the departing ferry.

meltynesson Program synthesis

Is anyone familiar with how this differs from 'animation' ?

meltynesson Bing Copilot seemingly leaks state between user sessions

Yes, but it's also generative, so at each time step it will be basing those predictions off of its own recent behavior so it's also chaotically, unpredictably performant in the quality of its predictions, but nothing else.

meltynesson Ask HN: Software Engineer Certifications That Mean Something?

I don't know about the certification, but the content is interesting, at least:

IEEE Professional Software Engineering Master

I haven't taken the exam, but started an online course, and it's been interesting.

Very textbook stuff, but towards a more succinct reference than lots of books about how software gets built. Not that formalism isn't important, I just haven't had a season of interest in it, particularly since it's so sparse and mostly without principles.

meltynesson Billions stolen in wage theft from US workers

It is, it's a state-level thing, and that serves as a check on companies that scale to be national or international.

Makes lobbying over 50x more complicated.

The fact that a UK publication is on about it should be a clear signal that the authors are being influenced by an international interest would like to simplify the process.

Probably chipotle

Here's a page that lists like, a third of the state agencies, I guess, hope this helps! : https://en.wikipedia.org/wiki/Ministry_of_Labour

meltynesson Lag in Cybertruck's drive-by-wire system

uhhh, 100ms isn't particularly stringent for a local system... nor should be the tolerance for the impact of control in a system that's literally at rest.

unless we're playing soccer / free-kicks, in which case, yeah, overengineer the hell out of it, sport.

ps, chapter 5 precisely specifies tolerances and minimums for things like latency of specific controls, for a control system for heavy machinery intended to be controlled by a human... and thanks for your input.

meltynesson Lag in Cybertruck's drive-by-wire system

The car with autopilot?

There's standards for control like MIL-STD-1472 which I'll bet this meets.

I'm sure none of them are "the highly redundant sensor fusion system with absurd levels of automation and sensing, the operator should not be able to notice any lag"

meltynesson Bard is now Gemini, and we’re rolling out a mobile app and Gemini Advanced

It's a hint about how to use it if you want to circumvent the censorship.

meltynesson Ask HN: What happened to normal-sized smartphones?

Came here to lament Sony like everyone else, but instead:

Consequences matter, so what is the consequence of having a screen that is just a bit too large?

- inadvertent swipes right, lands you on the Google Home page, but you can simply disable this or use a different launcher.

- can't fit as many thumb-addressable QS options, but Google-flavored android makes these larger, also provides one-handed mode gesture

- need to grip phone so that the corner extends beyond the palm of your hand, some phones have the camera on a lip that the index finger can use as a lip to hold

Obviously we're all ad supported entertainment focused media monsters so, screens aren't going to be smaller, you're just going to have to adapt

meltynesson Ask HN: 9-yo son wants to build a game, I'm lost. What can I do?

OOP, Events, and Algebra.

Unity has a pretty good debugger and graphical language.

Also somehow need to help develop tolerance for a great deal of thumbing through giant directories of names of things, skimming abstracts, and uncertainty in general.

meltynesson The Wolfram Prompt Repository (2023)

They sell Mathematica, see https://news.ycombinator.com/item?id=38934226 for more commentary about it

meltynesson Google Assistant will lose 17 features in the coming weeks

Hello!

meltynesson Google Cuts Jobs in Engineering and Other Divisions

Amazon is priced for massive relentless growth.

Apple, Microsoft, Meta priced for modest growth.

meltynesson Maestro: A Linux-compatible kernel in Rust

https://rust-lang.github.io/async-book/03_async_await/01_cha...

Is this all information that should be communicated to a programmer, or is all this info something that should be automatically assessed by a compiler with clear responses automatically to all possible situations specified, black and white, in the code?

If so, if it is the responsibility of the compiler, why release a version of the language, stabilize a spec, where that's not the case?