HNHacker News
TopNewBestAskShowJobs

roadside_picnic

1,721 karma · joined July 8, 2025

submissionscomments
roadside_picnic··on No more O'Reilly subscriptions for me
As an avid reader (and sometimes writer) of technical books, it's sad to see the, perhaps inevitable, decline of the space. I still remember in the early 2000s Barnes and Noble would still have massive shelf space devoted to every technical topic you could imagine. I could spend hours just exploring what languages and topics there were I didn't even know existed. Powell's Technical Books used to be an entire separate store filled with books on every technical topic imaginable.

The publishing industry veterans I've worked with told me it was even more incredible during the height of the dotcom boom: book sales in the 100,000 copy range was not that rare.

Today I can only think of two truly technical book stores that still exist: The MIT Press Bookstore in Cambridge, MA and Ada Books in Seattle, WA. The latter, while a delightful store, has relegated the true technical book section to the backroom, which unfortunately doesn't seem to get refreshed too often (though, part of the beauty of this is it still has many of the weird old technical books that used to be everywhere).

roadside_picnic··on No more O'Reilly subscriptions for me
> Unfortunately, I cannot read technical books fast and definitely not fast enough to make the subscription be worth $500 per year.

For me I find the $500 to be a pretty clear win as far as value goes. My shelves are already overflowing with, while not "timeless", much slower aging technical books. But quite often, throughout a year, I'll want a deeper dive into a current topic than I can get from online resources + Claude. Quite often that dive involves wanting to look through multiple books (even if only using a few chapters).

I know I'm a dying breed, but, while I love AI for interactive exploration and learning, I find books more valuable in the era of endless YouTube tutorials and AI slop blog posts. Technical topics benefit from "big picture" thinking that basically doesn't exist in modern short-form web content.

roadside_picnic··on Wall Street races to protect itself from AI bubble
> who are actually working with it only become more bullish

I have a feeling the word "actually" is doing a lot of work with this. I shipped AI facing user products a few years ago, then worked in more research focused AI work for awhile (spending a lot of time working with internals of these models). Then seeing where this was all headed (hype was more important than real work) decided to go back to good ol' statistical modeling.

Needless to say, while I think AI is absolutely useful, I'm bearish on the industry because current promises and expectations are completely out of touch with reality.

But I have a feeling because I'm not currently deploying a fleet of what people are calling "agents" (real agents are still quite cool imho), you would describe me as not "actually" using AI.

roadside_picnic··on Wall Street races to protect itself from AI bubble
It's national defense! Imagine if China had more slop than us!
roadside_picnic··on The RAM shortage comes for us all
I know this is mostly paranoid thinking on my behalf, but it almost feels like this is a conscious effort to attempt to destroy "personal" computing.

I've been a huge advocate for local, open, generative AI as the best resistance to massive take-over by large corporations controlling all of this content creation. But even as it is (or "was" I should say), running decent models at home is prohibitively expensive for most people.

Micron has already decided to just eliminate the Crucial brand (as mentioned in the post). It feels like if this continues, once our nice home PCs start to break, we won't be able to repair them.

The extreme version of this is that even dumb terminals (which still require some ram) will be as expensive as laptops today. In this world, our entire computing experience is connecting a dumb terminal to a ChatGPT interface where the only way we can interact with anything is through "agents" and prompts.

In this world, OpenAI is not overvalued, and there is no bubble because the large LLM companies become computing.

But again, I think this is mostly a dystopian sci-fi fiction... but it does sit a bit too close to the realm of possible for my tastes.

roadside_picnic··on The End of the Train-Test Split
> You make an LLM decision tree, one LLM call per policy section, and aggregate the results.

I can never understand why people jump to these weird direct calls to the LLM rather than working with embeddings for classification tasks.

I have a hard time believing that

- the context text embedding

- the image vector representation

- the policy text embedding(s)

Cannot be combined to create a classification model is likely several orders of magnitude faster than chaining calls to an LLM, and I wouldn't be remotely surprised to see it perform notably better on the task described.

I have used LLM as classifier and it does make sense in cases of extremely limited data (though they rarely work well enough), but if you're going to be calling the LLM in such complex ways it's better to stop thinking of this as a classic ML problem and rather think of it as an agentic content moderator.

In this case you can ignore the train/test split in favor of evals which you would create as you would for any other LLM agent workflow.

roadside_picnic··on The Junior Hiring Crisis
Larger scale war happens when the lives of young people are more valuable as fodder for the war machine than in a field or behind a desk.
roadside_picnic··on The Junior Hiring Crisis
> Just thinking maybe we're not seeing the end of software engineering for those of us already in it—but the door might be closing for anyone trying to come up behind us.

It's worth considering how aggressively open the door has been for the last decade. Each new generation of engineers increasingly disappointed me with how much more motivated they were by a big pay check than they were for anything remotely related to engineering. There's nothing wrong with choosing a career for money, but there's also nothing wrong about missing a time when most people chose it because they were interested in it.

However I have noticed a shift: while half the juniors I work with are just churning out AI slop, the other half are really interested in the craft of software engineering and understanding computer science better.

We'll need new senior engineers in a few years, and I suspect they will come from a smaller pool of truly engaged juniors today.

roadside_picnic··on Proximity to coworkers increases long-run development, lowers short-term output (2023)
The principle applies to a world where people work in offices doing serious long term R&D work. The quote is entirely irrelevant to people in working open offices for projects that change direction quarterly building features designed to make PMs look busy.
roadside_picnic··on Proximity to coworkers increases long-run development, lowers short-term output (2023)
I had an office with a door multiple times in my (early) career. An open office door is a universe away from sitting in an open office. Even when everyone has their doors open, a true office setup allows for plenty of focus.

On top of that "closed"/"open" is a false dichotomy, since you can trivially change the state of your office. Have a hard problem that needs to be solved by the end of the day? You can close your door and have absolute focus. After that task is solved, you can just open that office again.

Real offices also entirely change the tradeoffs for remote/in office. A true office feels like your room. It's considered a private space. I knew people that would bring in their own lamps (and keep the florescent lights off), bring in rugs, hang art from the walls, have tea setups, a bookshelf filled with reference material etc.

roadside_picnic··on Proximity to coworkers increases long-run development, lowers short-term output (2023)
I love Richard Hamming but

> But 10 years later somehow you don’t know quite know what problems are worth working on

Is clearly a quote from a different era. Not only have most engineers I've known never had a tenure at a job close to 10 years, I've found the foresight/planning window of companies I've joined is shrinking each year. In the era of "AI", leadership in most companies I've been at seem to think 3 months ahead is a bit too forward looking.

Also... how many people on HN even remember having an office? I had multiple jobs early in my career where I had an actual office with a window and a door. An open door office is nothing close to the misery of sitting at a desk in an open floor plan. The fact that you could close the door means you do have the opportunity for pure focus. Even when the door was open, it was customary to knock gently on the frame after very checking if it looked like the inhabitant was focused.

Richard Hamming describes a world of research that frankly doesn't exist any more today (I know because I briefly got a taste of the old world of research 20 years ago).

roadside_picnic··on Instagram chief orders staff back to the office five days a week in 2026
> Throughout hundreds-to-thousands of years of history your options have been

You might want to brush up on your anthropology a bit.

roadside_picnic··on Instagram chief orders staff back to the office five days a week in 2026
> Pre-covid - an entire working team is clustered in the same office

Just a reminder that there were plenty of great distributed teams long before Covid. I had had only a few "in office" jobs in my pre-covid career (and generally found collaboration better on remote teams, though goofing off has a bit harder to organize).

I also find it funny that somehow software engineers completely forget about the existence of open source software which worked in a more extremely distributed fashion than any remote company I've worked for at a time when IRC and email where the only tools for remote communication. Most of the most used and most successful software was written by globally distributed teams that only met face to face every few years if ever.

This timeline you are presenting applies only to large corporate jobs, where my experience has been "work productivity" always takes a back seat to organizational theater.

roadside_picnic··on Instagram chief orders staff back to the office five days a week in 2026
> What's the benefit of working remote from your team but next to random, noisy people?

The illusion of control? I mean we can pretend we don't know what this is about (well it's probably also about encouraging a reduction in force), but we do know right?

By far the people who bemoaned working from home the most were people whose job doesn't typically involve any actual "work". Not saying that there weren't exceptions, but the vast majority of working engineers I knew rejoiced in finally getting heads down time, while everyone whose job is primarily "performance for leadership" hated how difficult it was to perform visible theatrics on a camera.

Especially in large orgs "leadership" and "team success" are largely about optics. Being seen working in the office late is so much more important than getting any actual work done. It's only in small companies where actually shipping something has any value at all.

What I don't understand is why we still pretend like this is a mystery. Recognizing this I've completely avoided working for large orgs, and continue to enjoy remote work we're I can be valued for the results of what I build (well there's always a little theater) over office productivity performativity.

roadside_picnic··on Bring bathroom doors back to hotels
I honestly think it's more about "things that look better on instagram" that has infected virtually every hospitality related experience I've had in the last few years. A room that photographs well or a meal that looks ridiculous are more important than a room that's actually comfortable or a meal that tastes good.
roadside_picnic··on OpenAI needs to raise at least $207B by 2030
There's no worship here on my part (in fact I got out of the AI space because was increasingly less about tech/solving problems, and more about pure hype), but my experience in this industry has been that the most dystopian path tends to be the most likely. I would prefer if Google search, Reddit and YouTube were closer to what they were 15 years ago, but I do recognize how they got here.

I mean, look at all this "alignment" research. I think the people working in this space sincerely believe they are protecting humanity of a "misaligned" AGI, but I also strongly believe the people paying for this research want to figure out how to make sure we can keep LLMs aligned with the interests of advertisers.

Meta put so much money into the Metaverse because they were looking for the next space that would be like the iPhone ecosystem: one of total control (but ideally better). Already people are using LLMs for more and more mundane tasks, I can easily imagine a world where an LLM is the interface for interacting online world rather than a web browser (isn't that what we want with all these "agents"?) People already have AI lovers, have AI telling them that they are gods, having people connecting with them on a deeper level than they should. You believe Sam Altman doesn't realize the potential for exploitation here is unbounded?

What AI represents is where a single company control every piece of information fed to you and has also established deep trust with you. All the benefits of running a social media company (unlimited free content creation, social trust) with none of the draw backs (having to manage and pay content creators).

roadside_picnic··on OpenAI needs to raise at least $207B by 2030
> It also estimates that LLM companies will capture 2% of the digital advertising market, which seems kind of low to me.

I'm not super bullish on "AI" in general (despite, or maybe because of working in this space the last few years), but strongly agree that the advertising revenue that LLM providers will capture can be potentially huge.

Even if LLMs never deliver on their big technical promises, I know so many casual users of LLMs that basically have replaced their own thought process with "AI". But this is an insane opportunity for marketing/advertising that stands to be a much of a sea change in the space as Google was (if not more so).

People trust LLMs with tons of personal information, and then also trust it to advise them. Give this behavior a few more years to continue to normalize and product recommendations from AI will be as trusted as those from a close friends. This is the holy grail of marketing.

I was having dinner with some friends and one asked "Why doesn't Claude link to Amazon when recommending a book? Couldn't they make a ton in affiliate links?" My response was that I suspect Anthropic would rather pass on that easy revenue to build trust so that one day they can recommend and sell the book to you.

And, because everything about LLMs is closed and private, I suspect we won't even know when this is happening. There's a world where you ask an LLM for a recipe, it provides all the ingredients for your meal from paid sponsors, then schedules to have them delivered to your door bypassing Amazon all together.

All of this can be achieved with just adding layers on to what AI already is today.

roadside_picnic··on Python is not a great language for data science
> (n - 1)

It's also funny that one would write their own standard deviation function and include Bessel's correction. Usually if I'm manually re-implementing a standard deviation function it's because I'm afraid the implementors blindly applied the correction without considering whether or not it's actually meaningful for the given analysis. At the very least, the correct name for what's implemented there should really be `sample_std_dev`.

roadside_picnic··on Python is not a great language for data science
> that's not real-world coding

It's pretty clear the post is focused on the context of work being done in an academic research lab. In that context I think most of the points are pretty valid, but most of the real world benefit I've experience from using Python is being able to work more closely with engineering (even on non-Python teams).

I shipped R code to a production environment once over my career and it felt incredibly fragile.

R is great for EDA, but really doesn't work well for iteratively building larger software projects. R is has a great package system, but it's not so great when you need abstraction in between.

roadside_picnic··on Python is not a great language for data science
So I've been writing Python for around 20 years now, and doing data science/ML work for around 15. Despite being a Python programmer first I spent a good 5 years using R exclusively. There's a lot of things I genuinely love about R and I strongly believe that R is unfairly maligned by devs... but there's a good reason I have written exclusively Python for DS work for the last 5 years.

> Python is pretty good for deep learning. There’s a reason PyTorch is the industry standard. When I’m talking about data science here, I’m specifically excluding deep learning.

I've written very little deep learning code over my career, but made very frequent use of the GPU and differentiable programming for non-deep learning specific tasks. In general Python is much easier to write quantitative programs that make use of the hardware, and you have a lot more options when your problem doesn't fit into RAM.

> I have been running a research lab in computational biology for over two decades.

I've been working nearly exclusively in industry for these two decades and a major reason I find Python just better is it's much, much easier to interface with other parts of engineering when you're a using truly general purpose PL. I've actually never worked for a pure Python shop, but it's generally much easier to get production ML/DS solutions into prod when working with Python.

> Data science as I define it here involves a lot of interactive exploration of data and quick one-off analyses or experiments

This re-iterates the previous difference. In my experience I would call this "step one" in all my DS related work. The first step is to understand the problem and de-risk. But the vast majority of code and work is related to delivering a scalable product.

You can say that's not part of "data science", but if you did you'd have a hard time finding a job on most of the teams I've worked on.

All that said, my R vs Python experience has boiled down to: If your end result is a PDF report, R is superior. If your end result is shipping a product, then Python is superior. And my experience has been that, outside of university labs, there aren't a lot of jobs out there for DS folks who only want to deliver PDFs.

roadside_picnic··on Roblox is a problem but it's a symptom of something worse
HN is heavily biased towards silencing any content that makes any subset of its readership uncomfortable. This policy exists under the mistaken (I hope) assumption that silence is not fundamentally biased, which is usually not the case.

Posts about climate change, Israel/Palestine, uncomfortable concerns about the state of the industry etc will very often disappear quickly once they hit the home page.

Of course silence about climate change is beneficial for those contributing the most to it, silence about Israel/Palestine is very beneficial to Israel, and clearly silence about lack of concern for child safety by major tech companies benefits those companies.

roadside_picnic··on What OpenAI did when ChatGPT users lost touch with reality
Funnily enough I was just reading an article about this and "my boyfriend is AI" is the tamer subreddit devoted to this topic because apparently one of their rules is that they do not allow discussion of the true sentience of AI.

I used to think it was some fringe thing, but I increasingly believe AI psychosis is very real and a bigger problem than people think. I have a high level member of the leadership team at my company absolutely convinced that AI will take over governing human society in the very near future. I keep meeting more and more people who will show me slop barfed up by AI as though it was the same as them actually thinking about a topic (they will often proudly proclaim "ChatGPT wrote this!" as though uncritically accepting slop was a virtue).

People should be generally more aware of the ELIZA effect [0]. I would hope anyone serious about AI would have written their own ELIZA implementation at some point. It's not very hard and a pretty classic beginner AI-related software project, almost a party trick. Yet back when ELIZA was first released people genuinely became obsessed with it, and used it as a true companion. If such a stunning simple linguistic mimic is so effective, what chance to people have against something like ChatGPT?

LLMs are just text compression engines with the ability to interpolate, but they're much, much more powerful than ELIZA. It's fascinating to see the difference in our weakness to linguistic mimicry than to visual. Dall-E or Stable Diffusion make a slightly weird eye an instantly people act in revulsion but LLM slop much more easily escapes scrutiny.

I increasingly think we're not is as much of a bubble than it appears because the delusions of AI run so much deeper than mere bubble think. So many people I've met need AI to be more than it is on an almost existential level.

0. https://en.wikipedia.org/wiki/ELIZA_effect

roadside_picnic··on You can see a working Quantum Computer in IBM's London office
Before LLMs/AI became the obvious "next big thing in computing" I remember coming across a fair number of opportunistic devs on LinkedIn trying to promote themselves as "quantum software engineers", and even a just a year or two ago I would see "quantum machine learning" on people's profiles. I remember thinking maybe I had missed something and seeing how many qubits we could even have... needless to say it was (and still is) not enough for any meaningful ML work quite yet.

If you search you can still find some, and, as someone who has spent more than a decade doing actual machine learning, I find the audacity to claim that you're doing any kind of serious software engineering, let along proper ML work, on a quantum computer to be almost impressively audacious.

roadside_picnic··on Google Antigravity
I used to be really excited about "agents" when I thought people were trying to build actual agents like we've been working on in the CS field for decades now.

It's clear now that "agents" in the context of "AI" is really about answering the question "How can we make users make 10x more calls to our models in a way that makes it feel like we're not just squeezing money out of them?" I've seen so many people that think setting some "agents" of on a minutes to hours long task of basically just driving up internal KPIs at LLM providers is cutting edge work.

The problem is, I haven't seen any evidence at all that spending 10x the number of API calls on an agent results in anything closer to useful than last year when people where purely vibe coding all the time. At least then people would interactively learn about the slop they were building.

It's astounding to watch a coworker walk though through a PR with hundreds of added new files and repeatedly mention "I'm not sure if these actually work, but it does look like there's something here".

Now I'm sure I'll get some fantastic "no true Scotsman" replies about how my coworkers must not be skilled enough or how they need to follow xyz pattern, but the entire point of AI was to remove the need for specialize skills and make everyone 10x more productive.

Not to mention that the shift in focus on "agents" is also useful in detracting from clearly diminishing returns on foundation models. I just hope there are enough people that still remember how to code (and think in some cases) to rebuild when this house of cards falls apart.

roadside_picnic··on The time has finally come for geothermal energy
> a holy grail of energy production

Since you're comparing it to nuclear, I'm assuming you mean electricity production here, not energy production?

It's always worth remembering that electricity only accounts for ~20% of global energy consumption (in the US it's closer to 33%).

I suspect people confuse these two because in a residential context electricity plays a huge part of our energy usage, but as a whole it's a smaller part of total energy usage than most people imagine.

But any serious discussion of renewable energy should be careful not to make this very significant error.

The Lawrence Livermore National Laboratory publishes a great diagram of US energy use: https://flowcharts.llnl.gov/sites/flowcharts/files/2024-12/e...

roadside_picnic··on Canada loses its measles-free status, with US on track to follow
It's easy to blame Trump/MAGA if you refuse to recognize that many of the same companies producing these vaccines also played a major role in fueling the current opioid crisis which is coincidentally disproportionately impacting these very communities. Not to mention the reproducibility crisis which makes "trust the science" claims as naive as their inverse.

The core issue isn't with "antivaxxers" but with the continual erosion of trust that created the sentiment in the first place. The foundation of being willing to inject yourself with something that you personally can't verify the effectiveness or safety of is trust. At every level our social institutions: the government, large corporations, and academia, have continually chipped away at the foundations of social trust necessary for these things.

roadside_picnic··on Venn Diagram for 7 Sets
Technically a Venn diagram's entire point is to visualize all possible set relations between N sets. Their "practical" use is explicitly visualizing this.

In popular terminology they are very often confused with Euler Diagrams [0] which represent meaningful relations in sets but not all possible. You shouldn't create Euler Diagrams this complex, but the raison d'etre of Venn diagrams is to visualize the complex nature of set relations.

0. https://en.wikipedia.org/wiki/Euler_diagram

roadside_picnic··on You should write an agent
For a few months of news this works, but over the span of years even the statistical nature of language drifts a bit. Have you shipped natural language models to production? Even simple classifiers need to be updated periodically because of drift. There is no world where you lead the industry serving LLMs and don't train them as well.
roadside_picnic··on You should write an agent
> and hardware just vanish.

Okay, this tells me you really don't understand model serving or any of the details of infrastructure. The hardware is incredibly ephemeral. Your home GPU might last a few years (and I'm starting to doubt that you've even trained a model at home), but these GPUs have incredibly short lifespans under load for production use.

Even if you're not working on the back end of these models, you should be well aware that one of the biggest concerns about all this investment is how limited the lifetime of GPUs is. It's not just about being "outdated" by superior technology, GPUs are relatively fragile hardware and don't last too long under constant load.

As far as models go, I have a hard time imagining a world in 2030 where the model replies "sorry, my cutoff date was 2026" and people have no problem with this.

Also, you still didn't address my point that startups doing inference only model serving are burning cash. Production inference is not the same as running inference locally where you can wait a few minutes for the result. I'm starting to wonder if you've ever even deployed a model of any size to production.

roadside_picnic··on You should write an agent
So you're assuming there's a world where these companies exist solely by providing inference?

The first obvious limitation of this would be that all models would be frozen in time. These companies are operating at an insane loss and a major part of that loss is required to continue existing. It's not realistic to imagine that there is an "inference" only future for these large AI companies.

And again, there are many inference only startups right now, and I know plenty of them are burning cash providing inference. I've done a lot of work fairly close to the inference layer and getting model serving happening with the requirements for regular business use is fairly tricky business and not as cheap as you seem to think.

← PreviousPage 4 of 7Next →