HNHacker News
TopNewBestAskShowJobs

romaniv

2,750 karma · joined February 2, 2011

submissionscomments
romaniv··on Why senior developers fail to communicate their expertise
>We could call this the ‘Speed’ version of the system. It’s not meant to be understandable, the goal is getting things good enough to take it to the market for feedback.

AI is actually quite awful for prototyping, because it makes it far too easy to add random crap to your "prototype" without any specific intention. This quickly transforms the prototyping process from something that's high-level and geared towards building the mental model of the real system into something akin to copy-editing a random piece of software without any coherent mental model involved. Moreover, prompting allows to to glaze over some essential complexity of the task without getting any notions of the scope of the effort of actually doing it. In other words, people end up failing to make necessary decisions and simultaneously get bogged down with unnecessary ones.

In short, fast feedback loops are only useful if there is actual feedback involved.

romaniv··on Mythos Finds a Curl Vulnerability
"I signed the contract for getting access, but then nothing happened. Weeks went past and I was told there was a hiccup somewhere and access was delayed.

Eventually, I was instead offered that someone else, who has access to the model, could run a scan and analysis on curl for me using Mythos and send me a report. To me, the distinction isn’t that important."

Really? We're talking about (essentially) a product demo from a trillion dollar industry fueled by debt. Clearly, blog posts like this have an immense influence on the perception of usefulness of the particular model and AI in general. With so much staked on this for the company, wouldn't you want to be sure that you're using the actual product without anyone messing with the results in any way?

romaniv··on Cloudflare to cut about 20% of its workforce
This seems to be the new normal in Big Tech. They regularly announce massive layoffs, but if you look at their size over time it stays stable or grows. Cloudflare size grew every year. Microsoft size was stable for the past 4 years despite all the layoffs. Google had lost some (4%) employees in 2023, but has grown back to 2022 size last year. Meta shrunk by 22% in 2023, but has been growing in size since then and is probably back to 2022 size right now.

These companies overhire and then downsize. This is covered up by the moronic narrative about AI.

Side note. You know who is steadily shrinking, though? Intel. Wild, eh?

https://www.macrotrends.net/stocks/charts/NET/cloudflare/num...

https://www.macrotrends.net/stocks/charts/MSFT/microsoft/num...

https://www.macrotrends.net/stocks/charts/GOOGL/alphabet/num...

https://www.macrotrends.net/stocks/charts/META/meta-platform...

https://www.macrotrends.net/stocks/charts/INTC/intel/number-...

romaniv··on Motherboard sales 'collapse' amid unprecedented shortages fueled by AI
PCs and higher-end IoT devices (like RPi) are becoming less and less affordable. The web is saturated with slop and bot traffic. I wonder how many people think about what this means. We are rapidly losing several of our major technological ecosystems that have driven economic growth in the past few decades. It's not at all clear what economic benefits we will be getting in return.
romaniv··on When everyone has AI and the company still learns nothing
> "Where is the ROI for the 2 mio € we paid Anthropic last year?"

The bias in the assumptions here is absolutely bonkers.

Problem: GenAI is not generating any visible return on investment.

"Solution": rearrange your entire development organization around the technology and start inventing new tooling.

What's entirely obvious is that the point of such articles is not the stuff they purportedly discuss, but the normalization of assumptions those discussions are based on.

romaniv··on The Zig project's rationale for their anti-AI contribution policy
This seems like a sensible long-term strategy, much better one than entering into token-fueled AI arms race against slop. It's not even clear what's the end goal of such race would be for an open source project. Open source software was traditionally about growing knowledgeable communities and giving users ability to examine and modify software they use. LLMs quite obviously blow that up on several levels. For starters, if you hate dealing with code and prefer prompts, it's unlikely that you will be generating code that's enjoyable to work with for people who do read it directly.
romaniv··on Microsoft and OpenAI end their exclusive and revenue-sharing deal
Some people thought SHRDLU was basically AGI after seeing its demo in 1970. The hype around such systems was so strong that Hubert Dreyfus felt the need to write an entire book arguing against this viewpoint (1972 What Computers Can't Do). All this demonstrates is that we need to be careful with various claims about computer intelligence.
romaniv··on Spinel: Ruby AOT Native Compiler
One of cool things about Factor (and part of why I brought it up) is that it basically does something similar out of the box. There is a full-featured optimizing compiler and a simpler, faster non-optimizing compiler for eval-like functionality. They work seamlessly together in the interactive Factor environment:

https://docs.factorcode.org/content/article-compiler.html

romaniv··on Spinel: Ruby AOT Native Compiler
I will eat the downvotes to say what I actually think.

Unless this gets back eval, metaprogramming and threads this isn't all that interesting as an actual language. There are plenty of compiled languages out there. Metaprogramming is what makes Ruby interesting and expressive.

I know that this is just an experiment, but I've seen plenty of cases where stuff exactly like this gets forced into use in production because someone established in the company thinks some new experimental tool is "cool" and "the future".

---

Also, I would like to direct people to take a look at Factor programming language that both compiles into fairly efficient binaries and has amazing metaprogramming features inspired by Lisp and Smalltalk. It doesn't have real threads either, though, which is extremely unfortunate.

https://factorcode.org/

romaniv··on Scoring Show HN submissions for AI design patterns
> It's fun to complain about the good ol' days, but I'd rather face the world as it is and find the joy in it.

This is a manipulative combination of condescension, gaslighting and emotionalization.

"It's fun to complain" trivializes and dismisses a valid observation about the content being submitted as self-indulgent whining.

"I'd rather face the world" implies that people who want to see carefully constructed projects and human-written articles about them are refusing to face the world, i.e. delusional.

"Find the joy in it" reduces the whole discussion to the question of self-imposed mindset, as if there is no possible rational reason to be unhappy about what's going on.

romaniv··on Ban the sale of precise geolocation
The problem with all these discussions about banning stuff is that privacy is always on the back foot. It's by design. People who want to surveil and manipulate us are actively investigating new ways of doing it, they get paid for it and they risk nothing in the long run. All of these discussions about specifics are just reactions. They aren't even reactions to the surveillance itself, but rather to a discovery by someone that a new surveillance machine has been constructed and launched.

So the current feedback process involves: construction → exploitation → reporting → public awareness → legislation. This is too slow. Moreover, operating in this environment is exhausting.

We need a different feedback loop altogether. I'm not sure which one would work best, but something different needs to be considered.

romaniv··on Maine is about to become the first state to ban major new data centers
>For people who support this kind of ban, I'd ask if you would support a similar ban on new factories for, say, car parts.

If car parts factories produced nothing, employed no one and were made with equipment that will get outdated in a couple of years... Oh, gee, I dunno, it's a tough one.

romaniv··on Microsoft terminates VeraCrypt account, halting Windows updates
This is a tiresome argument that is based on a pile of unstated and rather shaky assumptions, ignores the very concept of opportunity costs and does not consider alternative solutions to the problems you seem to consider so important.

Fir starters, UEFI Secure Boot is actually rater bad at protecting users from bootkits or kernel-mode malware or anything, really. You can search this very website to get a giant list of bypasses and news about leaked vendor keys. Not to mention the fact that CrowdStrike Falcon incident had clearly demonstrated that Microsoft is more than happy to sign utterly insecure garbage.

Also, the issues with boot malware and kernel verification could be solved in many other ways, many of which are much more sensible or elegant. For example, by storing the bootloader and its keys on a physically separate read-only medium.

The issues with UEFI Secure Boot are actually the main point of the system, just like the issues with Windows executable signing are the whole point of that system.

romaniv··on Microsoft terminates VeraCrypt account, halting Windows updates
I still hope that one of these days people in general will realize that executable signing and SecureBoot are specifically designed for controlling what a normal person can run, rather than for anything resembling real security. The premises of either of those "mitigations" make absolutely no sense for personal computers.
romaniv··on DRAM pricing is killing the hobbyist SBC market
> use less RAM.

The same accounts that defended and promoted LLM use just a few weeks ago are now telling RPi users to use less RAM.

romaniv··on DRAM pricing is killing the hobbyist SBC market
RPis are used in a lot of embedded devices. From industrial IoT to music keyboards. You can't easily use refurbishes laptops for those[1].

--

[1] Korg Kronos with its crazy Intel Atom based architecture notwithstanding.

romaniv··on AI for American-produced cement and concrete
The current strategy of the AI hype machine is to exhaust people's reserves of attention by presenting a never-ending stream of hard-to-verify "positive" claims. It's Gish Gallop done on the Internet scale with a never-ending parade of tech influencers, proxy "journalists" and low-value accounts. The whole strategy aims for saturation and demoralized acceptance.

It's no surprise that people readjust their immediate reactions by expressing hostility and skepticism about anything AI-related without spending much time on analysis. In fact, it's an entirely rational repones.

Complaining about it without acknowledging the larger picture is disingenuous.

In this particular case, using the term "machine learning" would likely avoid the immediate negative reaction.

romaniv··on I'm OK being left behind, thanks
> weaponisation of FOMO

This is in an excellent characterization of the kind of marketing tactic I see all over social media right now and that I find absolutely disgusting.

The keyword here is fear. Despite faux-positive veneer, the messaging around certain technologies (especially GenAI) is clearly designed to induce anxiety and fear, rather than inspire genuine optimism or pique curiosity. This is significant, because fear is one of the most powerful tools to shut down rational thinking.

The subliminal (although not very subtle) message there is something very primitive. "If you don't join our group, you will soon starve to death." This is radically different from how most transformative technologies were promoted in the past.

romaniv··on The “small web” is bigger than you might think
Small Web, Indie Web and Gemini are terminally missing the point. The web in the 90s was an ecosystem that attracted people because of experimentation with the medium, diversity of content and certain free-spirited social defaults. It also attracted attention because it was a new, exciting and rapidly expanding phenomenon. To create something equivalent right now you would need to capture those properties, rather then try to revive old visual styles and technology.

For a while I hoped that VR will become the new World Wide Web, but it was successfully torpedoed by the Metaverse initiative.

romaniv··on Claude's Cycles [pdf]
I would like to note that it would be trivial to definitively prove or disprove such things if we had a searchable public archive of the training data. Interestingly, the same people (and corporate entities) who loudly claim that LLMs are creating original work seem to be utterly disinterested in having actual, definitive proof of their claims.
romaniv··on I'm Kenyan. I don't write like ChatGPT, ChatGPT writes like me
The fact that everyone is now constantly forced to use (oftentimes faulty) personal heuristics to determine whether or not they read slop is the real problem here.

AI companies and some of their product users relentlessly exploit the communication systems we've painstakingly built up since 1993. We (both readers and writers) shouldn't be required to individually adapt to this exploitation. We should simply stop it.

And yes, I believe that the notion this exploitation is unstoppable and inevitable is just crude propaganda. This isn't all that different from the emergence of email spam. One way or the other this will eventually be resolved. What I don't know is whether this will be resolved in a way that actually benefits our society as a whole.

romaniv··on Why engineers can't be rational about programming languages
> Instead of asking “which language is best?” we need to ask “what is this language going to cost us?”

As long as engineering salaries depend on tribal identity markers (i.e. language and tooling preferences) rather than ability to save money, people will entirely rationally choose tools that look good on their resume rather than save their companies money.

romaniv··on AI and the Future of American Politics
I liked Shneier much more when he was arguing against hyperbolic tech claims that were used as excuses for mass control and surveillance.

The notion that AI is reshaping American politic is a clear example of a made-up problem that is propped up to warrant a real "solution".

romaniv··on Comprehension debt: A ticking time bomb of LLM-generated code
No, this is not a pre-existing problem.

In the past the problem was about transferring a mental model from one developer to the other. This applied even when people copy-pasted poorly understood chunks of example code from StackOverflow. There was specific intent and some sort of idea of why this particular chunk of code should work.

With LLM-generated software there can be no underlying mental model of the code at all. None. There is nothing to transfer or infer.

romaniv··on AI-generated “workslop” is destroying productivity?
We call these workers “pilots,” as opposed to “passengers.” Pilots use gen AI 75% more often at work than passengers, and 95% more often outside of work.

Identify a real issue with the technology, then shift the blame to a made-up group of people who (supposedly) aren't trying hard enough to embrace the technology.

Embody a pilot mindset, with high agency and optimism

Thanks for the career advice.

romaniv··on Human-Oriented Markup Language
Reminder: you can use indentation to encode s-expressions. The result is very easy to parse and easy to read provided the underlying data structures are not insane.

https://srfi.schemers.org/srfi-49/srfi-49.html

romaniv··on Shai-Hulud malware attack: Tinycolor and over 40 NPM packages compromised
I've been a web developer for over two decades. I have specific well-tested solutions for avoiding external JS dependencies. Despite that, I have the exact same experience as the above security guy. Most developers love adding dependencies.
romaniv··on Writing by manipulating visual representations of stories
Not a single reference to P. H. Winston's work on story understanding and the Genesis system. Yes, sorry, I do expect people claiming to operate in some field to have knowledge of that field and acknowledge prior wok. This is absolutely fundamental to having an actual institution of science, rather than a bunch of people tinkering with random projects.

https://groups.csail.mit.edu/genesis/

https://www.youtube.com/watch?v=7XvgBI2KV28

romaniv··on SSL certificate requirements are becoming obnoxious
The web today is a rotting carcass with various middlemen maggots crawling all over it and gorging themselves on the decay. The only real discussion to be had is what to replace it with and how to design the new protocols to avoid the same issues.
romaniv··on AI Ethics is being narrowed on purpose, like privacy was
Spot on.

At some point I tried to figure out where the term "alignment" came from. I didn't find any definitive source, but it seems to have originated on a medium.com blog of Paul Christiano:

https://ai-alignment.com/ai-safety-vs-control-vs-alignment-2...

Basically, certain people are dismissing decades of deep though on this subject from writers (like Asimov and Sheckley), scholars (like Postman) and technologists (like Wiener). Instead, they are creating a completely new set of terms, concepts and though experiments. Interestingly, this new system seems to make important parts of the question completely implicit, while simultaneously hyper-focusing public attention on meaningless conundrums (like the infamous paperclip maximizer).

In my view, the most important thing about the three laws of robotics is that they made it obvious that there are several parties involved in AI ethics questions. There is the manufacturer/creator of the system, the user/owner of the system and the rest of the society. "Alignment" cleverly distracts everyone from noticing the distinctions between these groups.

← PreviousPage 2 of 30Next →