FWIW, in a typical apartment or single-family home, refrigeration uses a fraction of the energy that space cooling (also via a refrigeration/vapor compression cycle) requires on a warm day (and probably year round too unless in very mild climates). The psychrometric chart path is different so there are of course differences in the amount of energy required for the sensible and latent components, but the real difference is just the volume of air that needs to be dealt with.

My point being that at least from an energy and carbon perspective, lowering the space cooling demand via more effective building envelopes or increasing the space cooling supply efficiency - eg via membrane or dessicant dehumidification, better heat pumps etc) is far more impactful on a macro scale than better refrigeration.

Granted refrigeration in a warehouse eg is really also space cooling, but I’m just making the distinction between the dT=0-25F context and the dT>25F context. If I could only choose one technology to arrive at scale to improve the efficiency, it would be for the former context.

Definitely not the volume of air. The thermal mass of air is tiny.

The difference is in the thermal mass of the building and the surface area exposed to the sun.

Also the area that needs insulating, and in the extremes the amount of air that needs to be exchanged with the outside to make the house livable, and the heat generated by the people living in it (stick a 100W lightbulb in a fridge and see how cold it can get).

The insulation is actually solvable, and for heating can basically remove the power requirements: a house heated and using heat exchange on air leaving vs entering can be heated a lot just by having people inside it, let alone the other energy they use for other purposes. It's just more expensive to build this way, and with cheap energy it can a long time to pay back. Cooling you can't push down past the heat generated inside the house divided by the COP of your cooler, though.

Yep, PassiveHouse standards which typically include an extremely tight envelope which forces installation of outdoor air supply famously can get away with just a few hundred watts of heating capacity because of heat exchange on the incoming and outgoing airstreams!

Sure I was playing a little fast and loose there, but (a) the large surface area of the home (and resulting conductive transfer through the walls + convection transfer via infiltration through gaps) is directly a result of the fact that you need a significantly larger volume for humans to move around in and live in than you do to store food and (b) even if we do look directly at the volume of air, the difference is significant since at the end of the day, since for any given constant deltaT, your energy spent is still linear with mass or volume. And we are talking about roughly 2-3 orders of magnitude difference in air volume between a house and a refrigerator.

Anyways, if you write out all of the heat balance equations, you get a few W/m2 of flux on the inside wall of the home and a few W/m2 of flux on the inside faces of the fridge, assuming a typical wood frame construction in summer time and steady states all around.

So yes, of course multiplying the flux through the home’s wall by the surface area of the home results in a massive heat gain value compared to the heat gain conducted through the surface of the refrigerator, but that’s arguably precisely because of the two different volume requirements.

> I'm particularly annoyed by using LLMs to evaluate the output of LLMs.

Even though I largely agree with parts of what you wrote, if you squint your eyes enough you can kind of see an argument along the lines of “difficult to solve but easy to verify.”

> That "problem" remains unsolved because it's actually a fundamental aspect of reality. There is no natural separation between code and data. They are the same thing.

Sorry to perhaps diverge into looser analogy from your excellent, focused technical unpacking of that statement, but I think another potentially interesting thread of it would be the proof of Godel’s Incompleteness Theorem, in as much as the Godel Sentence can be - kind of - thought of as an injection attack by blurring the boundaries between expressive instruction sets (code) and the medium which carries them (which can itself become data). In other words, an escape sequence attack leverages the fact that the malicious text is operated on by a program (and hijacks the program) which is itself also encoded in the same syntactic form as the attacking text, and similarly, the Godel sentence leverages the fact that the thing which it operates on and speaks about is itself also something which can operate and speak… so to speak. Or in other words, when the data becomes code, you have a problem (or if the code can be data, you have a problem), and in the Godel Sentence, that is exactly what happens.

Hopefully that made some sense… it’s been 10 years since undergrad model theory and logic proofs…

Oh, and I guess my point in raising this was just to illustrate that it really is a pretty fundamental, deep problem of formal systems more generally that you are highlighting.

Never thought of this before, despite having read multiple books on godel and his first theorem. But I think you’re absolutely right - that a whole class of code injection attacks are variations of the liars paradox.

It's been a while since I thought about the Incompleteness Theorem at the mathematical level, so I didn't make this connection. Thanks!

Yeah, I think I’m with you if you ultimately mean to say something like this:

“the labels are meaningless… we just have collections of complex systems that demonstrate various behaviors and properties, some in common with other systems, some behaviors that are unique to that system, sometimes through common mechanistic explanations with other systems, sometimes through wildly different mechanistic explanations, but regardless they seem to demonstrate x/y/z, and it’s useful to ask, why, how, and what the implications are of it appearing to demonstrating those properties, with both an eye towards viewing it independently of its mechanism and in light of its mechanism.”

So the author’s core view is ultimately a Searle-like view: a computational, functional, syntactic rules based system cannot reproduce a mind. Plenty of people will agree, plenty of people will disagree, and the answer is probably unknowable and just comes down to whatever axioms you subscribe to in re: consciousness.

The author largely takes the view that it is more productive for us to ignore any anthropomorphic representations and focus on the more concrete, material, technical systems - I’m with them there… but only to a point. The flip side of all this is of course the idea that there is still something emergent, unplanned, and mind-like. So even if it is a stochastic system following rules, clearly the rules are complex enough (to the tune of billions of operations, with signals propagating through some sort of resonant structure, if you take a more filter impulse response like view of a sequential matmuls) to result in emergent properties. Even if we (people interested in LLMs with at least some level of knowledge of ML mathematics and systems) “know better” than to believe these systems to possess morals, ethics, feelings, personalities, etc, the vast majority of people do not have any access to meaningful understanding of the mathematical, functional representation of an LLM and will not take that view, and for all intents and purposes the systems will at least seem to have those anthropomorphic properties, and so it seems like it is in fact useful to ask questions from that lens as well.

In other words, just as it’s useful to analyze and study these things as the purely technical systems they ultimately are, it is also, probably, useful to analyze them from the qualitative, ephemeral, experiential perspective that most people engage with them from, no?

> The flip side of all this is of course the idea that there is still something emergent, unplanned, and mind-like.

For people who have only a surface-level understanding of how they work, yes. A nuance of Clarke's law that "any sufficiently advanced technology is indistinguishable from magic" is that the bar is different for everybody and the depth of their understanding of the technology in question. That bar is so low for our largely technologically-illiterate public that a bothersome percentage of us have started to augment and even replace religious/mystical systems with AI powered godbots (LLMs fed "God Mode"/divination/manifestation prompts).

(1) https://www.spectator.co.uk/article/deus-ex-machina-the-dang... (2) https://arxiv.org/html/2411.13223v1 (3) https://www.theguardian.com/world/2025/jun/05/in-thailand-wh...

> For people who have only a surface-level understanding of how they work, yes.

This is too dismissive because it's based on an assumption that we have a sufficiently accurate mechanistic model of the brain that we can know when something is or is not mind-like. This just isn't the case.

Nah, as a person that knows in detail how LLMs work with probably unique alternative perspective in addition to the commonplace one, I found any claims of them not having emergent behaviors to be of the same fallacy as claiming that crows can't be black because they have DNA of a bird.

> the same fallacy as claiming that crows can't be black because they have DNA of a bird.

What fallacy is that? I’m a fan of logical fallacies and never heard that claim before nor am I finding any reference with a quick search.

(Not the parent)

It doesn't have a name, but I have repeatedly noticed arguments of the form "X cannot have Y, because <explains in detail the mechanism that makes X have Y>". I wanna call it "fallacy of reduction" maybe: the idea that because a trait can be explained with a process, that this proves the trait absent.

(Ie. in this case, "LLMs cannot think, because they just predict tokens." Yes, inasmuch as they think, they do so by predicting tokens. You have to actually show why predicting tokens is insufficient to produce thought.)

It's much simpler than that. X is in B therefore X is not in A is what being said, and this statement simply doesn't make sense unless you have a separate proof that A and B don't intersect.

Good catch. No such fallacy exists. Contextually, the implied reasoning (though faulty) relies on the fallacy of denying the antecedent. The mons ponus - if A then B - does NOT imply not A then not B. So if you see B, that doesn't mean A any more than not seeing A means not B. It's the difference between a necessary and sufficient condition - A is a sufficient condition for B, but the mons ponus alone is not sufficient for determining whether either A or B is a necessary condition of the other.

I think s/he meant swans instead (in ref. to Popperian epistemology).

Not sure though, the point s/he is making isn't really clear to me as well

I was thinking of the black swan fallacy as well. But it doesn’t really support their argument, so I remained confused.

I've seen some of the world's top AI researchers talk about the emergent behaviors of LLMs. It's been a major topic over the past couple years, ever since Microsoft's famous paper on the unexpected capabilities of GPT4. And they still have little understanding of how it happens.

Thank you for a well thought out and nuanced view in a discussion where so many are clearly fitting arguments to foregone, largely absolutist, conclusions.

It’s astounding to me that so much of HN reacts so emotionally to LLMs, to the point of denying there is anything at all interesting or useful about them. And don’t get me started on the “I am choosing to believe falsehoods as a way to spite overzealous marketing” crowd.

No.

Why would you ever want to amplify a false understanding that has the potential to affect serious decisions across various topics?

LLMs reflect (and badly I may add) aspects of the human thought process. If you take a leap and say they are anything more than that, you might as well start considering the person appearing in your mirror as a living being.

Literally (and I literally mean it) there is no difference. The fact that a human image comes out of a mirror has no relation what so ever with the mirror's physical attributes and functional properties. It has to do just with the fact that a man is standing in front of it. Stop feeding the LLM with data artifacts of human thought and will imediatelly stop reflecting back anything resembling a human.

> Why would you ever want to amplify a false understanding that has the potential to affect serious decisions across various topics?

We know that Newton's laws are wrong, and that you have to take special and general relativity into account. Why would we ever teach anyone Newton's laws any more?

Newton's laws are a good enough approximation for many tasks so it's not a "false understanding" as long as their limits are taken into account.

I don’t mean to amplify a false understanding at all. I probably did not articulate myself well enough, so I’ll try again.

I think it is inevitable that some - many - people will come to the conclusion that these systems have “ethics”, “morals,” etc, even if I or you personally do not think they do. Given that many people may come to that conclusion though, regardless of if the systems do or do not “actually” have such properties, I think it is useful and even necessary to ask questions like the following: “if someone engages with this system, and comes to the conclusion that it has ethics, what sort of ethics will they be likely to believe the system has? If they come to the conclusion that it has ‘world views,’ what ‘world views’ are they likely to conclude the system has, even if other people think it’s nonsensical to say it has world views?”

> The fact that a human image comes out of a mirror has no relation what so ever with the mirror's physical attributes and functional properties. It has to do just with the fact that a man is standing in front of it.

Surely this is not quite accurate - the material properties - surface roughness, reflectivity, geometry, etc - all influence the appearance of a perceptible image of a person. Look at yourself in a dirty mirror, a new mirror, a shattered mirror, a funhouse distortion mirror, a puddle of water, a window… all of these produce different images of a person with different attendant phenomenological experiences of the person seeing their reflection. To take that a step further - the entire practice of portrait photography is predicated on the idea that the collision of different technical systems with the real world can produce different semantic experiences, and it’s the photographer’s role to tune and guide the system to produce some sort of contingent affect on the person viewing the photograph at some point in the future. No, there is no “real” person in the photograph, and yet, that photograph can still convey something of person-ness, emotion, memory, etc etc. This contingent intersection of optics, chemical reactions, lighting, posture, etc all have the capacity to transmit something through time and space to another person. It’s not just a meaningless arrangement of chemical structures on paper.

> Stop feeding the LLM with data artifacts of human thought and will imediatelly stop reflecting back anything resembling a human.

But, we are feeding it with such data artifacts and will likely continue to do so for a while, and so it seems reasonable to ask what it is “reflecting” back…

> I think it is useful and even necessary to ask questions like the following: “if someone engages with this system, and comes to the conclusion that it has ethics, what sort of ethics will they be likely to believe the system has? If they come to the conclusion that it has ‘world views,’ what ‘world views’ are they likely to conclude the system has, even if other people think it’s nonsensical to say it has world views?”

Maybe there is some scientific aspect of interest here that i do not grasp, i would assume it can make sense in some context of psychological study. My point is that if you go that route you accept the premise that "something human-like is there", which, by that person's understanding, will have tremendous consequences. Them seeing you accepting their premise (even for study) amplifies their wrong conclusions, that's all I'm saying.

> Surely this is not quite accurate - the material properties - surface roughness, reflectivity, geometry, etc - all influence the appearance of a perceptible image of a person.

These properties are completely irrelevant to the image of the person. They will reflect a rock, a star, a chair, a goose, a human. Similar is my point of LLM, they reflect what you put in there.

It is like puting vegies in the fridge and then opening it up the next day and saying "Woah! There are vegies in my fridge, just like my farm! My friege is farm-like because vegies come out of it."

On the other hand, IP addresses have crossed into the popular lexicon in exactly this manner… it’s common enough to hear people say “what’s my “ip?” or “are there any free ips?” or what are the IPs for x/y/z”.

I agree that it sounds stupid and incorrect, but that doesn’t necessarily mean using MCP as a metonym for MCP server.

Good point. Other examples are Wi-Fi (e.g. "What's your Wi-Fi?"), DNS (e.g. "You should change your DNS") and USB (e.g. "I only have 2 USBs on my laptop"). So who knows, maybe "MCPs" will catch-on.

I always say "The Google" though, so maybe I'm guilty as well of playing fast and loose with the Engrish Rangurage.

As someone who wrote my first line of code in approx 2010 and used git & GH for the first time in… 2013? it kind of amazes me to remember that Git is only 20 years old. GitHub for instance doesn’t seem surprising to me that is <20 years old, but `git` not existing before 2005 somehow always feels shocking to me. Obviously there were other alternatives (to some extent) for version control, but git just has the feeling of a tool that is timeless and so ingrained in the culture that it is hard to imagine (for me) the idea of people being software developers in the post-mainframe age without it. It feels like something that would have been born in the same era as Vim, SSH, etc (ie early 90s). This is obviously just because from the perspective of my programming consciousness beginning, it was so mature and entrenched already, but still.

I’ve never used other source control options besides git, and I sometimes wonder if I ever will!

What surprises me more is how young Subversion is in comparison to git, it's barely older.

I guess I started software dev at a magic moment pre-git but after SVN was basically everywhere, but it felt even more like it had been around forever vs the upstart git.

I'm old enough to have used RCS. Very primitive and CVS was soon in use. Git is a breath of fresh air compared to these ones.

Any version control where you had to manually (and globally) "check out" (lock) files for editing was terrible and near unusable above about 3 people.

Version control systems where you didn't have shallow branches ( and thus each "branch" took a full copy / disk space of files) were awful.

version control systems which would have corruption data-bases (Here's to you Visual source safe) were awful.

Subversion managed to do better on all those issues, but it still didn't adequately solve distributed working issues.

It also didn't help that people often configured SVN to run with the option to add global locks back in, because they didn't understand the benefit of letting two people edit the same file at the same time.

I have a soft-spot for SVN. It was a lot better than it got credit for, but git very much stole the wind from under its sails by solving distributed (and critically, disconnected/offline) workflows just a bit better that developers could overlook the much worse UX, which remains bad to this day.

>It also didn't help that people often configured SVN to run with the option to add global locks back in, because they didn't understand the benefit of letting two people edit the same file at the same time.

I think it was more that they were afraid that a merge might some day be non-trivial. Amazing how that fear goes away once you've actually had the experience.

(I had to check because of this thread. SVN and Git initial releases were apparently about 4 and a half years apart. I think it was probably about 6 years between the time I first used SVN and the time I first used Git.)

I still use RCS, typically for admin files like fstab or other config files in /etc.

Doing `ci -l` on a file is better and faster than `cp fstab fstab.$(date +%Y%m%d.%H%M%S)`

It's always hard to describe the minutiae of things happening in the span of just a couple of years, but I think you're overly broad here.

Wikipedia tells me the initial release of Subversion was in late 2000, and for git it was 2005 - but although those were kinda just smack in the middle of my first years online, learning to code, starting with FLOSS work, and so on - I think those years were pretty important with the shift to the WWW and then web 2.0.

I basically don't remember a world without SVN, but that's probably because I just missed the cutoff and projects and companies were migrating from CVS from 2002 on or so, because the model was very similar and while it wasn't drop in, it made sense.

For git I want to say it took just a little longer, and the decentralized model was so different that people were hesitant, and before github in 2009 (I know it was founded in 2008, but my user id is below 50000 and it felt very much new and not at all widespread in non-rails circles before that) I would have called it a bit niche, actually - so it's more like a 7year span. But of course I was living in my bubble of university, and working for 2 small companies and as a freelancer in that time. I think bigger FLOSS projects only started migrating in droves after 2010/2011. But of course my timeline could be just as wrong :D

Yeah, odd to learn. I remember dipping my toes into source control, playing around with CVS and SVN right around when git was originally announced and it felt so "modern" and "fresh" compared to these legacy systems I was learning.

> What surprises me more is how young Subversion is in comparison to git, it's barely older.

Subversion was so awful that it had to be replaced ASAP.

True. Also, Subversion was so great that it very quickly replaced the alternatives that predated it.

Not true. CVS stuck around a while longer.

There were far, far worse things out there than Subversion. VSS, ClearCase, an obscure commercial one written in Java whose name escapes me now..

Subversion was basically a better CVS. My recollection is that plenty of people were more than happy to switch to CVS or Subversion (even on Windows) if it meant they could escape from something as legitimately awful as VSS. Whereas the switch from Subversion to Git or Mercurial had more to do with the additional powers of the newer tools than the problems of the older ones.

I’ve been using Hatchet since the summer, and really do love it over celery. I’ve been using Hatchet for academic research experiments with embarrassingly parallel tasks - ie thousands of simultaneous tasks just with different inputs, each CPU bound and on the order of 10s-2min, totaling in the millions of tasks per experiment - and it’s been going great. I think the team is putting together a very promising product. Switching from a roll-my-own SQS+AWS batch system to Hatchet has made my research life so much better. Though part of that also probably comes from the forced improvements you get when re-designing a system a second time.

Although there was support for pydantic validation in v0, now that the v1 SDK has arrived, I would definitely say that the #1 distinguishing feature (at least from a dx perspective) for anyone thinking of switching from Celery or working on a greenfield project is the type safety that comes with the first class pydantic support in v1. That is a huge boon in my opinion.

Another big boon for me was that the combo of both Python and Typescript SDKs - being able to integrate things into frontend demos without having to set up a separate Python api is great.

There are a couple rough edges around asyncio/single worker concurrency IMO - for instance, choosing between 100 workers each with capacity for 8 concurrent task runs vs 800 workers each with capacity for 1 concurrent task run. In Celery it’s a little bit easier to launch a worker node which uses separate processes to handle its concurrent tasks, whereas right now with Hatchet, that’s not possible as far as I am aware, due to how asyncio is used to handle the concurrent task runs which a single worker may be processing. If most of your work is IO bound or already asyncio friendly, this does not really affect you and you can safely use eg a worker with 8x task run capacity, but if you are CPU bound there might be some cases where you would prefer the full process isolation and feel more assured that you are maximally utilizing all your compute in a given node, and right now the best way to do that is only through horizontal scaling or 1x task workers I think. Generally, if you do not have a great mental model already of how Python handles asyncio, threads, pools, etc, the right way to think about this stuff can be a little confusing IMO, but the docs on this from Hatchet have improved. In the future though, I’d love to see an option to launch a Python worker with capacity for multiple simultaneous task runs in separate processes, even if it’s just a thin wrapper around launching separate workers under the hood.

There are also a couple of rough edges in the dashboard right now, but the team has been fixing them, and coming from celery/flower or SQS, it’s already such an improved dashboard/monitoring experience that I can’t complain!

It’s hard to describe, but there is just something fun about working with Hatchet for me, compared to Celery or my previous SQS system. Almost all of the design decision just align with what I would desire, and feel natural.

Yep - this is also the official recommended method by Hatchet, also sometimes called payload thinning.

There are discussions about lowering the seams (harder to generate spin and makes the same spin rates less aerodynamically effective) as well as lowering the mound.