HNHacker News
TopNewBestAskShowJobs

chermi

653 karma · joined April 15, 2019

submissionscomments
chermi··on Aging brains blend memories together instead of just forgetting them
I mean with respect to your other thread too, where you seem to argue/believe academia has no idea about your brand new dynamical distributed view. I pointed to a really old paper with a self-admitted extremely simple model to point out that one of your specific insights over the field re. memory not being a memory stick is ancient and that your arrogance wrt being certain that your notion of memory is so novel that you weren't even sure there was a definition of capacity for distributed memory.

I intentionally gave older books and papers. I very clearly understand as does everyone in the field that they are not current or complete. But that awareness is precisely my point.

"Are you saying the brain is a Hopfield network, and that’s it? Because later you indicate otherwise. Kinda confused what I’m to make of it." You being confused by that is kind of my point, it shows the projection you're applying to the entire field. I understand models are wrong and iterative. I also understand the parts that survive, survive in the literature and the field at large. For example, the PNAS paper that showed "hey we think memory is distributed, right? Here's a simple model showing how memory might roughly be stored and recalled via these very simple dynamics over this very simple model. That's cool. Let's build on it." Did I make it clear enough yet that it was simple and wrong? Do you thus believe it wasn't valuable? I guess that would match your confident opinion that the whole field is dummies and of course those dummies cited an incorrect model so much.

If you stopped with saying "we know memory isn't a memory stick and here's the well-established evidence", then I'd fully agree with "Ok, but my argument wasn’t with the OP mentioning capacity, but with their analogy." and would've said nothing. Instead your posture was "I know some specifics based on recent findings that aren't obviously incorporated as of now into the current understanding -- which I refuse to read because it doesn't address precisely this thing only I understand(1) -- , so here is my theory that is certainly novel don't dare try to point me to preexisting literature that might help me refine my theory."

To be clear, I want everyone to explore their ideas and think it's great that you doubt things. But I don't like someone claiming superiority over a field they refuse to understand. And if your doubt something maybe you should check first if there's anyone else doubting it. (Hint: not only did everyone doubt it, they knew it was wrong and they're all actively working on using that doubt to improve it.) Of course a theory that predates a discovery doesn't cover that discover. Of course a simple model of memory isn't a complete model of memory, that was in fact the point! The notion of distributed memory and how it could feasibly work according to dynamics is made more precise than ever before to that time in that paper (among others). And of course it's outdated. But it contains a more precise statement and falsifiable statement of "memory is complicated dynamics and those dummies don't know and also that book is stupid because it doesn't cover X(2,3)" your theory espouses. My issue is with the arrogance and certainty you know best while refusing to actual understand what others think. And thus give no credit to. Science is iterative. Someone has an idea, builds a simple model as proof of concept. Someone else builds on that model precisely where it has holes (whose precise shape only present itself after new experiments motivated in part by the obviously wrong and simplified model and other attempts to break the model). Repeat. Built on the shoulders.. etc.

"Why’s it non-falsifiable? If you do find a single computational paradigm that explains all brain dynamics we can measure, then you have falsified the hypothesis that it’s an integration of multiple computational types.". I said non-falsifiable AND well-accepted. Non-falsifiable was the wrong word. What if I instead said trivially falsified and also adds nothing not known except to the layman who posits an analogy they clearly aren't sure of? Its non-falsifiable relative to the current state of the field, which I now know you're not interested in, because it says nothing not already stated more precisely, as of like 40 years ago.

(1) "As a rule I’ve stopped giving too much time to models that predate the last decades actual mechanistic facts. They’re fun curiosities, but hard to take seriously anymore. Here especially, the absence of astrocytes in the picture makes it hard to buy they have anything real to say about the mechanics at play. " In response to a book that has a lot to say about your new and novel dynamical theory.

(2) A hole it couldn't cover because the experiments and techniques didn't exist at that time. But they knew their theory wasn't complete. They were sharing what was learned up to that point especially the parts that seemed to generalize and survive specifics. That is, an incomplete model they knew was incomplete. But which I promise would help advance your journey.

(3) "Instead, memory, over time, is driven by a dynamical regime that adjusts its dynamics to account for the temporal pattern in the salient stimulus." Which is true and everyone in the field agrees with. And everyone agrees current models don't fully account for everything. Hence their continuing efforts to improve them by accounting for new findings.

chermi··on Uber shuts operations in Nigeria and Uganda with immediate effect
Does no one have any memory of what taxis were like and how much they cost? The taxis are gone because they weren't competitive.
chermi··on Aging brains blend memories together instead of just forgetting them
https://www.pnas.org/doi/pdf/10.1073/pnas.79.8.2554 in a very precise sense going back a long time. I don't really know who you're arguing against or refuting? You seem to be touching on some pretty well accepted ideas.

"All I’m trying to say is that analogy to current typical memory storage systems to explain the brains memory processes is not helpful." Again, my impression is that the original author's idea was about capacity in general, then he gave an analogy. The analogy was wrong, but your reply went way beyond his specific analogy to the extreme of discarding memory itself as useful concept. To be clear, I agree that it is distributed and lossy and time-dependent. I agree it is not just a simple read-off of a static chunk with a fixed address.

"Which points to a possibility: maybe the brain is multiple types of computers interacting. And the physical realization of these computing architectures aren’t spatially separated but thread through each other in the biochemistry and physical dynamics of cells."

I would say that is both non-falsifiable and well-accepted.

https://mitpress.mit.edu/9780262041997/theoretical-neuroscie... might interest you and on the other end bialek's spikes shows quite a bit actual does happen near the single neuron limit. bialek also does/did a lot of other stuff on capacities and processing you might be interested in. https://mitpress.mit.edu/9780262181747/spikes/

chermi··on Aging brains blend memories together instead of just forgetting them
None of that changes whether there is a physical capacity, which I think was the larger point? There is no reason to believe distributed memory doesn't suffer from the capacity component of the bottleneck. In fact iirc there was some late 80s/early 90s papers on the memory capacity of NN. Btw I would advise against the absolute statement that there's absolute segregation of memory and processing.

I guess my point is the brain not being "von Neumann" in architecture or digital is not proof that isn't a "computer" of some sort.

chermi··on True Rate of Unemployment
But the second order effect is that if real estate is no longer treated as an investment, then it becomes less necessary to own a house.
chermi··on True Rate of Unemployment
So instead... What? The high price right now to live near work or relocate is regressive. I don't know what proof there is that "money not going back in to the economy"* is a larger effect than increased housing affordability?

*btw, if the money went toward interest on debt, all else being equal, that is money back in to the future economy via reduced tax demand. Of course this won't actually happen and the government will spend every cent it gets and more. So maybe your opposition should be with the government and not LVT?

chermi··on True Rate of Unemployment
Can you explain what, if any, difference there is between LVT and georgism? How do we know simply building more wouldn't lower the returns on real estate and naturally disincentive using real estate as in investment vehicle? I guess maybe federal level taxes are basically un-repealable whereas construction has a strong local component? I understand there's complicated feedback effects here. For example, buildings be effectively illegal precisely where it's needed most.

Would decoupling education funding from local taxes do anything? I'm thinking maybe it decreases the incentive to hold on to a house thus increasing effective mobility/geographic diffusion? Probably a small effect but it seems possible it could have second order effects.

chermi··on True Rate of Unemployment
I think even when renters outnumber they're still less likely to be involved in local government. Current owners have a vested interest. Only nimbys show up
chermi··on True Rate of Unemployment
I thought in Singapore it was all termed leases? That the government always owns the land?
chermi··on Three sites made 215,128 “best software” pages for AI. Perplexity cites them
If you let a plain llm search the internet with no guidance, it's basically a string matcher with no concept of quality. I thought perplexity's whole point was being good at search?
chermi··on Evidence of Fraud in an Influential Study About Procrastination
I mean more the Marxist school of thought, not just the economics. Things like critical theory and post modernism.

But also, he made predictions of the near future in his time. They were incorrect. No updates were made. Those who did try to refine and update his ideas are far less influential. His original theory survives like gospel. I wish I could say only by non-academics.

Edit - back to economics. It's funny. I would say in economics the closest thing we have to double blind wrt economic systems is socialism/communism vs. capitalism. It's clearly not a true double blind. But also clearly some people just don't care to feed that data back in to their models. Because they aren't scientific.

chermi··on Why are rivers so mathematical?
That questions makes no sense? What is "mathematical" as a property? It has it backwards. Accepting the property of being mathematical, isn't it caused effectively by how much we study and model it mathematically?
chermi··on NASA: Fill in a name, and you can have an exclusive coordinate in the universe
It's a solid angle basically, right ? That's a coordinate
chermi··on Google Antigravity introduces Boost deep reasoning (/boost)
Invented, failed to make a good product out of it.
chermi··on Google Antigravity introduces Boost deep reasoning (/boost)
There is almost no upside in adopting any new Google product. They kill or abandon almost everything, for decades now. Try it maybe. But don't build any reliance on it.
chermi··on Evidence of Fraud in an Influential Study About Procrastination
That's covered by trying and seeing what sticks. That's really all there is to it. I already got downvoted for saying this in other words - feedback against observed phenomena is what matters. Everywhere you see shitty "science" you see either missing or ignored feedback.
chermi··on Evidence of Fraud in an Influential Study About Procrastination
Science that is science is self-correcting via prediction and experiment. The key piece is the feedback. The "scientific method" as taught in grade school hardly matters and is basically cargo cult. The "is this science" lint is "are you working toward more accurate and especially predictive models of observed phenomena? Are you actively trying to destroy your theory?" If the feedback is missing or ignored, it's just opinion pieces.

Edit- I don't know what's controversial about this? Feedback is the self-correcting piece.

chermi··on Separating logic and language
disclaimer: I really don't nearly enough about this stuff to be taken seriously, these are just my immediate thoughts.

fMRI is one of the best tools we have, but that doesn't mean it's good. fMRI shit should always be taken with spoon of salt. source: a bunch of grad school friends in neuroscience and psychology a few years ago.

"on two patients whose language networks were destroyed by stroke, plus fMRI on healthy volunteers solving the same puzzles silently. Both lines of evidence point the same way: the language network sits out during reasoning." that doesn't mean they didn't learn the meta-ability of how to solve such things without language. You can learn something with language, move it to a separate "database" in some abstracted form then pull from that "database" from a non-language component. In fact I would think the language-isn't-thinking people would favor an explanation where the thinking ability is stored independently of language. I guess my point is that no "language" component being lit up during a specific reasoning task doesn't exclude the possibility we learned that specific reasoning algorithm (for lack of a better word) via language + other things.

TL;DR maybe this specific "logic" ability is a learned-via-language+ component of the brain. Language not being active during that puzzle could just mean language+ generated that component and stored it in a more abstract form than language.

chermi··on Tether: iMessage, SMS, etc. on Linux
You understand I'm on your side here, right?
chermi··on The growing divide between AI hype and software engineering reality
See: /\
chermi··on The growing divide between AI hype and software engineering reality
Exactly the post you'd expect at this stage in the technology adoption and hype cycle. People got over hyped not understanding how technology and technology adaption works. Then they get a bunch of like 6-12 month lagging indicators further convincing them of the worse. Right when they become most certain the technology is useless is precisely when the people that have adopted it and truly understand it leave them in the dust. There's gotta be a name for it?
chermi··on Tether: iMessage, SMS, etc. on Linux
Lol. Someone liking this change is stupid for letting it lift their outlook, right? How stupid is it then to let this comment rob your hope.
chermi··on Most Physicists Don't Agree on Most Things
"I don't know if I'm interpreting you correctly, but I for one have enough faith in people to guess that most people wouldn't interpret the poll as implying physicists disagree about the wavefunction of the ground state of hydrogen or something."

I guess I'm on the opposite end. I think many people will have any idea what a ground state is and those very people will come away most confidently believing physicists don't understand shit.

chermi··on Nvidia projects $673B in sales as AI demand widens
I'm not sure. I think they count training as R&D? Hopefully we learn just they're doing their accounting with anthropic ipo
chermi··on Humanity has the debate about AI consciousness backwards
Those are also different things. I think it should unambiguously clear that humans are special to humans. If not that's basically a mental health problem lol
chermi··on Gemini Omni 1.1 Flash
They're main edge has been multimodal. I think they're still the best overall on multimodal? If I were them I would try to be the best at at least something.

I never personally got the least bit excited about Sora or nanobanana or whatever video/audio generation thing. But I guess I'm just not their customer. I do love the read-side of it though.

chermi··on Gemini Omni 1.1 Flash
Google we ain't falling for it. #neverforget3.5prowithinamonth
chermi··on The Teaser Period: Why the AI Boom Is Hitting a Reset Wall
Individuals and families buy houses(ideally). Entities like corporations buy work, often intelligent work. If they can crank up outputs and profits via more intelligent work done by models, I think the ceiling is global demand for the entity's product. Which is still a bound, and ultimately set by individual consumers.

The question is whether the consumers will have the income to spend. So I kind of agree in the end I guess.

chermi··on Nvidia projects $673B in sales as AI demand widens
Yeah but remember 1) stuff gets more expensive the more it's "processed"/further up the chain. It's hard to think of something further up the value chain than a modern gpu server. Just look at one input, asml machines. 2) a lot of dram, labor etc is also baked into the gpus.
chermi··on Nvidia projects $673B in sales as AI demand widens
Capex covers a lot of non-gpu stuff, but yeah it doesn't sound crazy to me.
← PreviousPage 2 of 19Next →