HNHacker News
TopNewBestAskShowJobs

potholereseller

41 karma · joined August 2, 2024

submissionscomments
potholereseller··on Novel Accidentally Leaves AI Prompt Requesting to Copy Other Writer's Style
I wonder how a prompt could wind up in the final product; this may indicate that the author lacks more than just writing skills; "slopdash" seems a relevant neologism.

> as book 1 of this series released 1/24/25, book 2 on 3/13/25, and book 3 on 3/23/25

That is indeed suspicious, but it is also bad business. Book-reading is uncommon in the US, so it makes no sense to release a book series that frequently; many readers take a long time to read through a single book; it's more entertaining to spread releases out.

> which leads into my paranoid theory that a percentage of readers are just skim reading

That may explain some of the appetite to buy three books in ~3 months. I'm not saying that's fast (I could read a 300-page novel in a day or two, without skimming), but it may be fast if you don't skim, you are working >40 hours/week, and you have a family to take care of. I wonder how much of the book industry is propped up by skimmers.

> The internet at large is also facing an existential threat in the shape of an AI slop tsunami.

That's a bit much; AI slop is a problem, but it isn't a tsunami-level event. It remains unclear if any managers have successfuly replaced any employees 100% with AI; somebody has to write the prompts, and it isn't the managers, who have neither the time nor, in most cases, the understanding of the jobs purported to be replaceable.

For people who like trendy things, AI slop is a problem, because AI is actually good at churning out trendy things, because trendy things are usually slop. But even for trendy slop, AI isn't a tsunami; trends will ebb-and-flow with or without AI.

potholereseller··on The Bargain of Working Hard and Getting a Job Simply Doesn't Hold Anymore
> “Learn to code” is, after all, used as a pejorative against people in famously less stable fields, like journalism.

FWIW, most of the people who repeated that cliche were not programmers. It's odd that some journalists are still bitter about being belittled with that phrase by randos on social media.

The article continues with cliched economic analysis, my commentary on which I will elide. Instead, I will point out that the big man himself wrote about starting a startup during the early part of the Great Recession: <https://paulgraham.com/badeconomy.html>: "The truth is more boring: the state of the economy doesn't matter much either way."

potholereseller··on Ask HN: Is anyone else burnt out on AI?
> It has at least been interesting to me to reflect on how I can still appreciate media that humans make when I find AI media so repulsive.

A relevant analog is the arguments about whether independent bands were more real than those who had signed with labels -- about whether money/popularity corrupts art. I never took a side on that, but I do think that most music isn't worth listening too, simply because it's so saccharine and cliched.

So to generalize, Sturgeon's Law is evenly distributed: 90% of everything -- including sci-fi, music, and AI-generated stuff -- is worthless. 90% of AI content is slop, because it is prompted by people who have no taste whatsoever; not bad taste, just zero taste; not everyone is gifted. In the hands of people with refined taste (whether good or bad), they can use AI to produce that 10% of worthwhile AI stuff; but those with refined taste know how to keep AI content from distracting from the larger work, so you never know they are using AI at all.

I don't think society-wide refinement of taste is possible; Sturgeon's Law is here to stay. Instead, we need a corrolary to Sturgeon's Law, which provides a solution to the problem: you can't overturn Sturgeon's Law; you can only build filters to avoid the crap. I can't say how to build such filters, but we can start thinking about them.

potholereseller··on Activating AI Safety Level 3 Protections
They claim that their new LLM, unfettered, is so good that they are concerned that it might be able to provide instructions to make things like nuclear weapons. A mediocre physicist could provide similar, or better, instructions, but nobody is trying to shut down physics departments at universities.

This is a sales pitch wrapped in over-compliance and presented as an alert. This is pathetic, but I am somewhat entertained by their desperation; I will be even more entertained when LLM companies start using sex-appeal to sell subscriptions ("women find our latest model weights to be irresistible, so sign up now for our $1000/month plan").

potholereseller··on UK study: Almost half of young people would prefer a world without internet
A friend is listening to every recording of Coast to Coast AM, which began in 1988. He notes that it was a remarkably interactive format; people would call in and try to trick Art Bell with preposterous conspiracy theories; Art Bell would play along, because it was fun (and it was his job).[0] The 6pm news was never interactive; what they talked about was never fun; it was and still is an endless stream of terror.

[0] I'm not defending Coast to Coast AM; I'm noting it's interactive format.

potholereseller··on AI Is a Nothingburger [video]
I've noticed a pattern:

(1) People assert that some event will happen (e.g. someone getting elected, some technology changing the world);

(2) They do zero analysis of the feasibility of that event happening;

(3) They divide themselves in regards to whether the consequences of the event will be good or bad;

(4) After months or years, the event doesn't happen; and

(5) They are confused, but eventually re-focus on the next event to get hyped about.

These people suffer significant personal harm (wasted careers, ruined relationships, endless dread of the future, increasing debt etc), because they are trapped in a cycle of unjustified exuberance. This isn't any way to live; it's an empty life.

I'm not quite saying that everything is a nothing-burger (though I do agree with OP that LLM->AGI is not going to happen), but I know that most predictions are incorrect -- because it is difficult to accurately describe things when you are excited, and predictions are usually made due to excitement. Many companies and politicians believe they have a chance of success if they can get enough people hyped.

I do not look forward to seeing how people deal with an LLM industry crash. I remember reading about a young man who borrowed $10k via student loans and donated that to a US presidential candidate, because that young man believed the candidate was going to win and wipe away all student loans. The candidate never got close to winning; that young man desperately begged for the money back; he was ghosted.

potholereseller··on What interesting things low spending people do, that others know nothing about?
>In my head, every piece of hardware, broken furniture, or random screw has a future use.

For me, this mentality has lead to a sort of hoarding, in which everything I own has a potential future use, but little of it has been used, even after years of being in my possession. This is partly due to the fact that the quantity physically obscures many of the items and I forget that I own them. It is also because many of the items have near-zero use-value; I can imagine a use for almost anything; I can only actually use a small fraction of that.

Recently, I have been aggressively throwing away possessions; if I see it and I can't remember using it (or I have replaced it with something I use more frequently), then I throw that thing in the trash ASAP. I don't care how much it's worth; space, time, and labor; every possession has mental load.

potholereseller··on JavaScript Style Sheets (1997)
You can modify the styling of elements in Javascript with modern browsers, but I'm not sure it is exactly "standard", in the way that CSS is "standard". A modern form of JSSS might look like this: an HTML document with no styling/CSS, but some JS that styles each element when it is loaded. This might cause some flicker [0], because the HTML would be rendered un-styled and then re-styled by your JS code. You can mask that flicker with a sort of "loading screen" -- oh, now I know why I see those tacky pre-loaders on so many websites.

The Internet is a flat circle, in which old technology is constantly re-invented by semi-literate people. This will surely be the sole Horseman of the Internet Apocalypse: a black-and-purple checkered horse, the rider wearing a hoodie that says "my prompts bring all the ladies to the yard" and wielding AR goggles. He speaks a dozen words to command his AI legions -- they do his bidding in a thousand-thousand jobs: getting hired, attending meetings, writings emails, getting fired; each in a cycle fast enough that he is ever employed, ever influencing, ever disrupting. We have assuredly earned all that he does to the Internet.

[0] = This is alluded to in TFA: "You'll also want to be careful with your style-setting functions: get too complex and it will take your document forever to render itself in the user's browser."

potholereseller··on To Save America's Democracy We Should End Elections
Sortition (the formal term for this) is very attractive, because it seems it would eliminate corruption and other political ills. The problem is that the bureaucracy (he part of the executive branch that has some legislative power) would grow, become more corrupt itself, and become very effective at manipulating the citizen-legislators. I don't think you can eliminate the Federal or State bureaucracies without also parallelizing the legislative process.

Now, don't get me wrong, sortition may improve things for a time. But like code, people figure out how to take advantage of laws. Re-writing the US Constitution on a regular basis is very risky; instead, SCOTUS changes it's opinion on what the Constitution means periodically; it's like updating the interpreter's code instead of changing user-submitted code.

You might also be interested in the districting method in which you draw district boundaries through the middle of high-population areas. Half the population would be on either side of each boundary. You keep subdividing until you have as many districts as needed. (Sorry, I don't remember the name for this, but it was the best option I could find when I last looked into this topic more than a decade ago.)

potholereseller··on Just Fucking Use React
I'm waiting for an article entitled: "Just Use X" Considered Harmful. The Web is complex (i.e. has many parts) because it is meant to satisfy a very wide range of needs. Most websites use WordPress, because that's an especially easy option to host, configure, and add/update content. But "Just Use WordPress" would be nonsense; a lot of people need more (e.g. Google Maps) or less (e.g. static personal pages) than what WordPress is appropriate for.

I know, I'll write an article entitled "Just Fucking Use C, You Yellow-bellied Sapsuckers". It'll be about using good-old CGI with C. Imagine a web framework in C, which generates HTML, CSS, and JS. It'll be sleek, easy-to-deploy, portable, fast, and you can optimize to your heart's content. Plus, it will future-proof your career, because your boss will make you spend a chunk of your career re-writing it in Rust: CGI + Rust is the future that nobody is aware of (read: dreading) yet.

potholereseller··on Rogue communication devices found in Chinese solar power inverters
Getting two sources is the standard in journalism; that is what journalists are taught in university. The fact they are anonymous sources is also considered okay by most, if not all, journalists.

I'm not saying this is good. Rather, I'm saying that this is normal: it's rumors masquerading as revelation. Reality is far too boring (and occassionally too horrifying) to be news-worthy.

potholereseller··on What Happens When There's No Job to Learn On?
> The proof that AI has already ruined life is that I sent that rainbow photo to a friend of some 40 years with lens data and it had full EXIF info attached, less location, which can be seen from only one place on earth. He asked me if it was real.

AI may have ruined his life, but it seems to have not ruined your's; it certainly hasn't ruined mine. This phenomena might be more unevenly distributed than we know.

> My philosophy is not that we seek meaning in life, but that we live to create meaning, as we write stories with our lives, and exerience the stories of others, stories that touch our knowledge (filtered, not a giant bag of dissociated bytes) and life experience, and as we process them with our minds and emotions.

If something can destroy that meaning-creating in a person without altering their way of life (as war alters lives), then it might be the case that the person's ability to create meaning was under-developed the whole time. It's somewhat related to consumerism and trend following; one makes so little meaning in one's life that an external, endless stream of vapid things feels meaningful. The current generation of AIs (and social media) seem to target that exact sort of person.

It will be worth paying attention when the AI companies crash and burn during the next recession. Our friends/families/co-workers who currently rely on those vapid streams might not take it well, and may need someone who can show them that it's not difficult to create meaning.

potholereseller··on Google Starts Scanning All Your Emails After Gmail Upgrade
The title is misleading, because scanning emails is normal. The update adds AI to help with writing emails. To me, it's a superfluous addition, because so few people read emails (the meetings will continue until people start reading their emails).

All that's necessary is for Gemini to summarize emails to close the loop and create AI-in-AI-out Telecommunication (AAT?). AI generation of meetings and AI summarization of meetings (AAM?) is all that's necessary to close the last communication loop in business.

Our forebears promised that the future would be cool and fun and sexy; academics of yore warned that the future might be horrifying; the future has turned out to be a multitude of preposterous contrivances.

potholereseller··on Claude's system prompt is over 24k tokens with tools
When you've trained your model on all available data, the only things left to improve are the training algorithm and the system prompt; the latter is far easier and faster to tweak. The system prompts may grow yet more, but they can't exceed the token limit. To exceed that limit, they may create topic-specific system prompts, selected by another, smaller system prompt, using the LLM twice:

user's-prompt + topic-picker-prompt -> LLM -> topic-specific-prompt -> LLM

This will enable the cumulative size of system prompts to exceed the LLM's token limit. But this will only occur if we happen to live in a net-funny universe, which physicists have not yet determined.

potholereseller··on How are cyber criminals rolling in 2025?
> Almost seems like there's room here for a grey hat to come in and ...

... call the scam numbers to tie up their staff and prevent them from talking to potential victims. Someone like Kitboga could do this at scale. Where there's a phone number, there's a way.

potholereseller··on [dead]
>and that its Palestinian population will “leave in great numbers to third countries”

Where would the Palestinians be sent to? TFA says Egypt and Jordan won't accept them. And this isn't a few thousand people to be moved; it's millions. This is yet another case of people skipping analysis of a plan's feasibility, and only judging the outcome of the plan. This is why conflicts so often escalate so far; analysis seems to be anathema to leadership. Many people involved could use a dose of mockery for the preposterous things they propose.

potholereseller··on Starlink Terms of Service
Marketing finds a way. I just wish the writer had picked something more interesting than "self-governing principles" and "good faith"; every democratic government on Earth claims to be built on those principles. There's lots of good anarchist and non-anarchist fiction and non-fiction they could have been inspired by, if only books weren't so arcane. For lack of inspiration, Mars may just be a copy of Earthgov, but with more underground structures.
potholereseller··on I tried resisting AI for programming. Then I tried using it. Both were painful
If the hype is true, then AI-generated code will soon need no review at all. Programmers will be totally useless. Managers will be able to spend ten hours each quarter doing the work of a hundred programmers. You would be better off moving into management now, because that's the only place code will be seen. Open-source projects will also be abandoned, since AI will generate the kernel tailored to your CPU, motherboard, peripherals, and applications. The AI companies say that AI will eliminate all jobs. They sincerely mean all jobs.

But how do you know that AI companies will stick around? Those companies rely on excitement among people like programmers to pay subscriptions. But when the next recession puts a lot of those people out of work, they will cancel those subscriptions in order to pay for rents/mortgages and food. AI hasn't delivered on the promise of a work-free world, so we still have to pay our bills. No subscriptions means the AI companies will lose their only revenue stream. Why woudl a VC invest in companies whose revenue vanished? During the next recession, a lack of programming skills and a lack of AIs will be the worst scenario for many programmers. Whatever company you work for will struggle under the weight of you and other programmers who can't do anything except review code, debug code, and write complex code. The simple, repetitive code will be un-writable -- the simple things will become arcane.

There are already posts on HN bemoaning the loss of programming skills due to AI use. A generation for whom the simple and boring things might already be arcane.

← PreviousPage 2 of 2