HNHacker News
TopNewBestAskShowJobs

plasticeagle

694 karma · joined November 4, 2022

submissionscomments
plasticeagle··on The realities of being a pop star
I feel absolutely confident that Charlie XCX would never use generative AI in any form. And this sentence is lovely;

"...let some random person you’ve just met in the bathroom try on the necklace around your neck that is equivalent to the heart of the ocean"

Like you I always look for signs of AI in writing I see online, and it's incredibly disappointing how often it's there. There's no personality, no charm, nothing unique - just the same flawless grammar and overuse of cliche. This piece is filled with the quality of humanity that we once took for granted. This is what we are losing.

plasticeagle··on Why can't transformers learn multiplication?
Humans, and many other creatures, learn. While they are performing a task, they improve at the task.

LLMs are trained. While they are training, they are not doing anything useful. Once they are trained, they do not learn.

That's the distinction.

plasticeagle··on The day my smart vacuum turned against me
Most, if not all, of the article is written by AI.
plasticeagle··on How to Draw a Space Invader
Stop it.

Checks

Oh my god. 11/10.

plasticeagle··on How to Draw a Space Invader
May I say that this website is one of the loveliest and most pleasant to read that I've ever seen.
plasticeagle··on Parsing Protobuf like never before
We use flatbuffers extensively in our message passing framework, and they are extremely fast if you take care with your implementation. They have a few features that make them especially useful for us

1) The flatbuffer parser can be configured at runtime from a schema file, so our message passing runtime does not to need to know about any schemas at build time. It reads the schema files at startup, and is henceforth capable of translating messages to and from JSON when required. It's also possible to determine that two schemas will be compatible at runtime.

2) Messages can be re-used. For our high-rate messages, we build a message and then modify it to send again, rather than building it from scratch each time.

3) Zero decode overhead - there is often no need to deserialise messages - so we can avoid copying the data therein.

The flatbuffer compiler is also extremely fast, which is nice at build time.

plasticeagle··on The Real GenAI Issue
Certainly Gen AI is being marketed to business leaders as being capable of reducing their payroll, but I don't believe that's what it's for as such. And as several comments have mentioned, its energy use is not even especially significant.

Gen AI exists to wrest control of information from the internet into the hands of the few. Once upon a time, the Encyclopaedia was a viable business model. It was destroyed as a business model once the internet grew to the point that a large percentage of the population was able to access it. At that point, information became free, and impossible to control.

Look at google's "AI summaries" that they've inserted at the top of their search results. Often wrong, sometimes stupid, occasionally dangerous - but think about what will happen if and when people divert their attention from "the internet" to the AI summaries of the internet. The internet as we know it, the free repository of humanity's knowledge, will wither and die.

And that is the point. The point is to once again lock up the knowledge in obscure unmodifiable black boxes, because this provides opportunity to charge for access to them. They have literally harvested the world's information, given and created freely by all of us, and are attempting to sell it back to us.

Energy use is a distraction, in terms of why we must fight Gen AI. Energy use will go down, it's an argument easily countered by the Gen AI companies. Fight Gen AI because it is an attempt to steal back what was once the property of all of us. You can't ban it, but you can and absolutely should refuse to use it.

plasticeagle··on A brief, incomplete, and mostly wrong history of robotics
Of the one or two successes I have experienced in my career, preventing the adoption of ROS was one of the most satisfying. Watching another group adopt it, and then fail spectacularly was also reasonably gratifying.
plasticeagle··on Generative AI coding tools and agents do not work for me
ChatGPT and Gemini literally only know the answer because they read StackOverflow. Stack Overflow only exists because they have visitors.

What do you think will happen when everyone is using the AI tools to answer their questions? We'll be back in the world of Encyclopedias, in which central authorities spent large amounts of money manually collecting information and publishing it. And then they spent a good amount of time finding ways to sell that information to us, which was only fair because they spent all that time collating it. The internet pretty much destroyed that business model, and in some sense the AI "revolution" is trying to bring it back.

Also, he's specifically talking about having a coding tool write the code for you, he's not talking about using an AI tool to answer a question, so that you can go ahead and write the code yourself. These are different things, and he is treating them differently.

plasticeagle··on Yes-rs: A fast, memory-safe rewrite of the classic Unix yes command
Oh, Hi Mark.
plasticeagle··on AI in my plasma physics research didn’t go the way I expected
Does anybody else find it peculiar that the majority of these articles about AI say things like "of course I don't doubt that AI will lead to major discoveries", and then go on to explain how they aren't useful in any field whatsoever?

Where are the AI-driven breakthroughs? Or even the AI-driven incremental improvements? Do they exist anywhere? Or are we just using AI to remix existing general knowledge, while making no progress of any sort in any field using it?

plasticeagle··on Data centers contain 90% crap data
> 90% of libraries consist of books that are never opened

I'm reasonably certain that this statistic is completely made up. The best number I can find for the proportion of library books that are never borrowed was from a university library, and was 25%.

plasticeagle··on Recent results show that LLMs struggle with compositional tasks
Indeed.

https://stackoverflow.com/questions/9252656/einsteins-riddle...

plasticeagle··on Bypass DeepSeek censorship by speaking in hex
I would imagine that the difficulty lies in finding effective ways to remove information from the training data in that way. There's an enormous amount of data, and LLMs are probably pretty good at putting information together from different sources.
plasticeagle··on You could have invented Fenwick trees
There's several things like that you can accidentally re-invent. I re-invented the Trie, and Markov Chaining whilst at University. What I didn't do is give them a thorough and rigorous description in any kind of publication.

That's quite a big part of the difference, I think.

plasticeagle··on Static search trees: faster than binary search
As a computer science article, it should have mostly been pseudocode. Concrete implementations can be given in an appendix, where they belong.
plasticeagle··on Trellis – 3D mesh generative model
Do not worry.

Art is much more than pictures on your monitor. If you want pieces of art, made by your peers, visit your local galleries and buy it. I don't know who you are or where you live, but I'm willing to bet that where-ever it is those local galleries exist - and the artists that exhibit there would love to sell some of their work.

And you can be sure that human-made art will remain, and be valued, because art is what humans love to make most of all.

plasticeagle··on Show HN: Cut the crap – remove AI bullshit from websites
I mean, that's fair to some extent.

But on the other hand you could have just purchased any cookbook that covers the basics, instead of taking all this web-scaped content without attribution or compensation. I mean, look, I totally get it and I'm certainly guilty of this too - but let's not pretend that we're not basically stealing other people's content here. Much of the time those people running those recipe websites are just trying to cover their hosting costs and make a squeak of money on the side.

A friend of mine tried to set up a website that would host open-source recipes for people - he called it The Open Sauce - but in the end there just wasn't enough input from recipe creators.

Also, and by the way, the top google hit for bechemal is this : https://www.allrecipes.com/recipe/139987/basic-bechamel-sauc.... Few ads, and the recipe is at the top of the page. No life story in sight.

plasticeagle··on Everyone is capable of, and can benefit from, mathematical thinking
Two thoughts

1) It's tragic that being "bad at math" is often positioned as some kind of badge of honour.

2) It's definitely not the case that everyone is capable of mathematical thinking. Having spent a certain amount of time trying to teach one of my kids some semblance of mathematical thinking, I can report confidently that his ability in this area is almost non-existent. His undeniable skills lie in music and writing, but definitely not in maths.

Yes, music and maths have some things in common. But musical thinking is not mathematical thinking.

plasticeagle··on Hand Tracking for Mouse Input (2023)
That sounds very interesting. I've been needing a filter to deal with noisy A/D conversions for pots in an audio project. Noise on a volume control turns into noise on the output, and sounds horrible, but excessive filtering causes unpleasant latency when using the dials.
plasticeagle··on Hand Tracking for Mouse Input (2023)
This is cool, but a moving average filter is pretty bad at removing noise - it tends to be longer than it needs to be because its passband is so bad. Try using a IIR filter instead. You don't need to deal with calculating the coefficients correctly because they'll just be empirically determined.

out = last_out * x + input * (1-x)

Where x is between zero and one. Closer to one, the more filtering you'll do. You can cascade these too, to make a higher order filter, which will work even better.

plasticeagle··on Trump wins presidency for second time
I'm against Trump for all the obvious reasons, which there's no need to enumerate here.

Being against him doesn't explain that an openly fascist craven liar just won the US presidency. A poorly educated population does.

plasticeagle··on Trump wins presidency for second time
I've also spent plenty of time there over the years, and while most people I interacted with did seem perfectly fine, there were glimpses of something quite wrong.

A woman who worked at the hotel I was staying at had never visited the centre of the city the lived in, because she was afraid of being "knifed". This was Dayton, Ohio. Downtown Dayton is lovely.

A colleague who appeared reasonably intelligent and competent absolutely did not believe that Evolution occurred. I explained that this while this view might be common in the US - and it is - the rest of the world mostly considers this settled science.

Religion is absolutely far too influential a force in people's lives. This is decreasing, but it's still problematic I believe.

The Armed Forces are idolised. Airports have special lines for service personnel. You get to board early if you're in uniform. This is almost unique in the world, to the best of my knowledge.

plasticeagle··on Trump wins presidency for second time
It's well documented that Americans are, on average, quite undereducated. And it's also quite well documented that most of the people that vote for Trump are poorly educated.

So, not stupidity, no. But a lack of education can look similar.

https://www.uneducatedamerica.com/useful-links

plasticeagle··on ChatGPT Search
People already spent upwards of $50 a month for the internet itself, plus they probably pay monthly for one or more streaming services. They likely pay separately for mobile data too.

Separate monthly fees for separate services is absolutely unsustainable already. The economic model to make the internet work has not yet been discovered, but $20 a month for a search engine is not it.

plasticeagle··on Show HN: iFixit created a new USB-C, repairable soldering system
I have a cheap binocular microscope at home, and solder most everything underneath it. A USB microscope is much lower resolution, and the image lags. They're pretty terrible.

At work I get to use a very very expensive Olympus binocular microscope. It is extraordinarily good, but at about 60k it costs more than a car.

plasticeagle··on Show HN: iFixit created a new USB-C, repairable soldering system
This device seems perfectly insane.

I have two Quecoo soldering stations : https://www.quecoo.com/products/quecoo-t12-956-soldering-dig...

They're very cheap, they heat up just as quickly as any other induction iron. They are very repairable. They come with multiple tips, which are cheap to replace.

They don't contain expensive batteries or pointless USB-C WebSerial-based interfaces. You turn them on. They heat up. I've had mine for years, so they're reliable too.

iFixIt have a laudable mission generally, but this product will be an expensive failure.

plasticeagle··on Why Copilot Is Making Programmers Worse at Programming
This article 100% read like it was written by AI.

I will personally never use Copilot, or any other AI code generation tool, for the simple reason that I enjoy writing code.

Even if I were unfamiliar with a new language, I'll still never use it. Instead, I'll consult the documentation and follow examples. I like coding, and I neither need nor want a machine to do it for me.

It's exactly the same as writing English. There is great pleasure to be found in writing, it's worth your time. Just be careful when doing so to not end up sounding exactly like ChatGPT.

plasticeagle··on Our Git Hash Bug
Always use the full hash everywhere.

Never store the short hash anywhere. It's good only for immediate commandline use, and nothing else.

plasticeagle··on Moral implications of being moderately successful computer scientist and a woman
Cf: "honestly the best marketing scheme in history is men successfully getting away with calling women the "more emotional" gender for like, EONS, because they've successfully rebranded anger as Not An Emotion"

It's very much sarcasm, and very much warranted.

← PreviousPage 3 of 6Next →