barfbagginus

As a full-blown queer autonomous neo Marxist, I own my own f*** company.

Have done so for 5 years now.

We pay all workers their fair share of the socially necessary production time. If the organization needs surplus to invest in equipment, we requisition it from the production with consensus from the workers.

I had to kick a number of capitalists to the curb to achieve this. Why? Cuz they wanted to exploit workers, and they wanted me to do it for them.

Get lost!

You can get paid the full value of your work, if you seize control of the business. Just don't be a wimp, and out maneuver the pudding headed capitalists who know no actual economics or category theory, lol. It's f*** easy.

Well it's a shell game, right? On the one hand here's this huge valuation and future profit. On the other hand, the workers get just a small fraction of that value and still live in squalor.

Is that where your billions come from? You comfortable with that?

That's not how I run my company.

You don't pay people less because you have them by the short and curlies of their own poverty. That is exploitive. You pay them a fair share of the value they're producing.

The problem is that if corporations did this, if they stopped exploiting, if they paid a fair share, then those billions in valuation magically vanish. It's because the valuation was predicated on exploitation to begin with.

God forbid that labor would control itself

The scary part is that sci hub is in danger!

It's been under massive legal attack in a few jurisdictions, and as far as I know, has stopped scraping.

Many, many papers from after 2021 that are not available. It has become a major issue, cutting me off at times from new results important to my work or study.

It's kind of a scary time now, and I feel the strong need to regroup and find a way to get sci-hub back to scraping.

We have to imagine there are millions of people being cut off from millions of people-years of research for every year this embargo continues. It's crucial that we bring it back up!

Anyhow, thanks for being a sci-hub user and endorser! I love it when my peers in academia use it!

That reminds me I've been meaning to offer a contribution to your farming robot!

For ease of interoperability, I'd like to build a one way automation that occasionally exports your OnShape design into your GitHub. That makes it easy for users to throw parts into a CAM editor or slicer, without having any on-shape experience.

Have you thought about the automation, and would you like to discuss your concerns, needs, and like-to-haves?

I think that dirac delta solutions let us model point particles. We can lift differential operators onto the space of D' of distributions. This lets us extend the pde solution set with distributions.

Conceptually this is how we would model a field with interacting particles using PDEs, and is as far as I know the reason we solve with distributions and consider them physical in the first place.

The question of whether dirac is homotopic to vacuum would then mean exhibiting a homotopy in D', which is the dual space to the underlying function space, and represents both functions and distributions as functionals on a space of test functions. The topology of D' is the weak-* topology. Given a net N_t of distributions, N_t converges to N if N_t(g) converges to N(g) for all test functions g.

So to make your proof tight, I propose we should lift it into D', where our family u_t of shrinking deltas is already a net. We'd then prove that u_t -> 0 converges for each test function.

I think the result would be much more satisfying and watertight, since your current proof does not really make up its mind about what it's conclusion is.

Now, I am not sure that the distributions need to be a hilbert spaces to solve PDEs in this sense. This is because distributions have derivatives and can be acted on by our lifted differential operators. These solutions are weak in the sense that they are solutions with respect to the test functions. But I also believe that the solutions are faithful, in the sense that any solution corresponding to a function in the base space will also be a solution for the original operator.

Also, if you need our space of distributions to be a Hilbert space, we can pass to distributions over sobolev n space, the space of functions with square integrable nth derivatives. The distributions here end up being hilbert spaces! Not only that but the square integrability condition is pretty mild and still leaves us a lot of very useful functions for physics. I think that some quantum theories use these kinds of Hilbert spaces. Lol, take that with a grain of salt - this is lore to me, not terra firma.

Anyhow, if you want to try to persuade yourself that we can't just throw away delta functions and other distributions, and then try to lift your proof into space of distributions, I think that would be very fun!

I'll try to do that as well!

Ps. Forgive me if I'm Tom Sawyering you or nerd sniping! Please also forgive me for any mistakes I'm making!

Eksevier must die, and we must spare no effort to kill it. Work to kill elsevier.

Use the library Genesis

Use the QubesOS PDF sanitizer or similar

Integrate a sanitizer into the Zotero ref manager

Contribute to libgen software projects

Use the sci-hub

Tell your peeps, tell your peers, tell your org

Sabotage peeps peers and orgs that won't help

Okay, fair point, I was not understanding that topological defects are points in solution space that are not path connected to the vacuum solution. I'm learning as I'm going!

Now I also realize that the paper seems to say that the both the ordinary dirac shell solution and their modified shell are TDs, without proving it.

I'd like to work up to proving whether the collapsing modified shell really does homotopy into into a point and then fade away into the vacuum.

But first, I'm struggling with

> nobody would argue that a point charge is a topological defect

It's actually not clear to me that the point mass is not a TD!

Let's try to write the homotopy sending a point mass Mδ₀ to the vacuum solution 0. Let t vary from 0 to 1. Then a possible homotopy is

h(t) = (1-t)Mδ₀

This gives us

h(0) = Mδ₀

h(1) = 0

The problem is that I don't know how to prove continuity of h.

First, I don't know how to compute even the continuity of neighboring Delta functions for t < 1. But that feels intuitively like it should be continuous.

On the other hand, I REALLY don't know how to prove continuity at t = 0, since the function seems to spontaneously collapse from a distribution with a kind of pseudoinfinite value at the origin, into a regular function with the value 0 at the origin.

Using the notation of inner products and test functions, can we prove that it's continuous both for t < 1 and t = 1?

I know that's a bit more technical than we usually get here on HN! I truly appreciate the help!

Again I consider the terms synonyms.

The solitons/defects belong to the poisson equation for newtonian gravity. I called that a gravitational field equation, since it models gravity as a field.

You called it a scalar field model because it's a scalar field. Your contention is that the gravitational field equations typically refer to Einstein's field equations. I grant that, but also consider poisson's equation as a gravitational field equation. They're both classical field theories for gravity.

Again it's seeming like a bit of a case of potato/potato, rather than advancing the conversation. I'd like you to accommodate me a little better please, and hold off on nitpicking unless it leads to constructive synthesis.

Are you okay with the clarified terminology and my request?

If so, see my other response one layer up, where I asked about problem with the s term in their modified shell solution. We actually have some nits to pick there! I hope to see you there in the other thread!

Please forgive me for double dipping, and read my other response on the soliton/topological defect matter.

Re: stability / realism

My concern comes at a point even before yours - the solution given is in a spherically symmetric and perfectly flat space time without any actual matter to bend the gravitational tensor. If it were moving through space with an undulating and assymmetric mass/gravity field, we might expect the shell itself to deform in random ways as it encounters ripples in space. There's no force that would restore it to spherical symmetry, so it might drift and disperse into some weird blob. The extra degrees of freedom of the curved space may even let the solution wiggle and diffuse into an ordinary solution, so that it no longer represents a topological defect in the solution space.

The shells certainly wouldn't remain concentric in my thinking, and might not need to. In my thinking I don't see that the shells necessarily need to impose forces against each other when they pass through each other. Imagine an onion shell where the onion layers are free to phase through each other and become a bit wibbly wobbly.

Eventually you have a space with many topological singularity walls of this kind of passing through each other. This random "space lasagna" is actually much more realistic than concentric shells. A problem I see though is that we might need concentricity to induce a centrally directed attractive force keeping galaxies together.

Ie, why does random isotropic space lasagna bind together spherical galaxies and clusters?

Actually modeling this more realistic and chaotic version of the topological defect solution could be very exciting! Of course it's no more satisfying than dark matter, because we don't know what would actually cause the singularity lasagna solutions to form in the first place, instead of just ordinary solutions.

But wait there might even be a problem before this!

Re: the dirac delta s term.

In the definition of the modified shell model on the first page, we have a constant term s such that sδ(R-r)=1 if r=R. This might be nonsense.

First let's point out that R-r=0, so this implies sδ(0) = 1. But there is no real number s such that sδ(0) = 1 is finite, since d(0) is not a real number.

But okay, maybe s is a distribution that transforms the dirac delta distribution into the kronecker delta function. If so which distribution is it?

The Fourier transform gives us a clue. Let k be the kronecker delta at 0

k(x) = 1 if x=0, else 0.

s now scales δ into k:

sδ = k

Applying the fourier transform we get

F(sδ) = F(k)

F(s) * 1 = 0

where * is convolution and 1 and 0 are constant functions.

Note that F(k) = 0 because F(0) = 0 and k = 0 almost everywhere.

Now

[F(s)*1](x) = 0 for all x

Which implies

∫F(s) = 0.

So we know that s can't be a constant, and its Fourier transform has a vanishing integral.

This isn't enough for me to reject s, but I'm puzzling as to what it actually is. What do you think about this?

In my usage, a soliton and a topological defect are synonyms for certain isolated solutions to differential equations. Please replace soliton with topological defect in my statement above and see if it makes more sense.

The idea is there are families of solutions to partial differential equations that are stable, but cannot not deform continuously into any of the "ordinary" solutions without leaving the set of solutions at some intermediate point of deformation. These outlying solutions are topologically isolated Islands in the solution space.

We can think of them as disconnected families of non-standard solutions - solitons.

We can also think about them as topological defects in the set of solutions. We ordinarily expect the solution set of a differential equation to form a path connected set including the trivial solution. If there is some region of solutions that is topologically cut off from the ordinary solutions, we call its solutions topological defects. But these are nothing more than the solutions that can't be smoothly deformed into the zero solution.

Note that many icebergs have more than one equilibrium, with various final heights above the water.

For example this iceberg towers above the water in one equilibrium, but barely rises above the water in its other equilibrium.

https://ibb.co/SdcKMBV

I was about to post something similar. But you nailed it before I could swing my own hammer, to use a figure of speech.

I was about to post something similar. But you nailed it before I could swing my own hammer, to use a figure of speech.

Many collaborators I've talked to have shown confusion about why they should do things to cause some far away revolutionary improvement to their own lives, or even less the lives of others, when it would involve putting themselves and loved ones at risk.

Like not only do they feel the risk is too great and the payoff too little. But they also can't comprehend what would motivate anyone to act that way. Sometimes they even insist that revolutionaries must be insane or mentally deficient.

But I don't know if it's the case that people make a decision to be this way. I think it's just the default way to be, given a relatively stable life.

People won't really be able to account for why they became this way, or think of a time when they weren't this way. To them it will just be common sense, while revolutionary thought will just seem dangerous, anathema, and basically unthinkable and wasteful.

I think it's like a mental block that evolved to let us form stable social groups. And I think that certain circumstances can unblock it.

In particular I think that a person can get so traumatized by the state of affairs and things that happen to them that they have a strong reaction that causes them to give up hope for getting rewarded by the current regime. Then a person begins investing significant portions of their effort not in just surviving the current regime, but in overthrowing and changing it in a positive way. They begin to live on the hope of justice they can bring for not only themselves, but also others. They see their own lives as irrevocably altered by the trauma, and become agents for the cause of breaking that cycle. To them, the dream of a future generation not having to experience that trauma can be more motivating than thoughts of their own survival.

I believe that just like the first mode evolved to keep societal groups stable, the revolutionary mode evolved to let groups compete with each other and achieve beneficial new power structures.

Plus, humans bullshit all the time, even well paid and highly trained humans like Doctors and Lawyers. They will bullshit while charging you 400 an hour. Then they'll gaslight you if you try to correct their bullshit.

AI will bullshit sometimes, but you can generally call it on the bullshit and correct it.

For the tasks that it helps me with, I could work with a human. But the human I could afford would be a junior programmer. Not only do they bullshit more than a well prompted AI, but I also have to pay them 30 an hour, and they can't properly write specs or analyze requirements. GPT 4 can analyze requirements. Much better then a junior, and I'm many ways better than me. For pennies.

I do use it in the real world, to maintain and develope software for the industrial design company I own. It would be foolish if I didn't. I've been able to modernize all our legacy codes and build features that used to stump me.

Maybe the fact is that I'm an incompetent programmer, and that's why I find it so helpful.

If that's the case so be it! It's still a significant help that is accessible to me. That matters!

Topological defects - also known as soliton solutions - are isolated solutions of PDEs that cannot be continuously deformed into the ordinary solutions.

This is saying that the observed effects of DM could be explained as a bunch of enormous solitons in the gravitational field equations. Each soliton has no mass, but together they explain galaxy rotation curves and DM lensing.

But there is a catch. The authors found a massless soliton in the poisson field equation for newtonian gravity. General relatively transforms into this equation in the limit when space is flat. But we don't know if that solution is still stable in spacetimes - like ours - that are curved by massive objects.

In a curved space time being hit by relativistic particles and gravity waves, these spherical shells could wobble and collapse, fade away, or break apart. Proving this one way or another will be interesting work!

So your points are

I don't really pick a definitiom intelligence

llms could be regurgitating training data

they take more data to train than humans

They can't do some tasks, like driving

However, in my experience, LLM are more empathetic than humans, more able to help me reason about my feelings and communication problems than humans, less likely to perform microagressions or be racist or ableist than humans, and better at math and science than most humans. These are just my personal feelings as an autistic person, which I can back up only loosely with benchmark data, but which I will expect to see the world coming to realize over the next years.

So in terms of being able to constructively interact with me in an intelligent and helpful way, LLMs are often more useful better than humans that I have access to. I say they are smarter than these people as well, because AI will give me solutions that are useful, and which other humans could not give me.

The fact that it cannot drive doesn't bother me since I don't consider driving a general skill but a specialized skill. It can still have general intelligence without being able to do some specific things. Going back to my original post, I specifically reject AGI definitions where to be generally intelligent the AI has to out perform humans in every possible skill. I would consider that a super intelligent AGI.

As for the information problem and data issue, AIs so far have been black boxes isolated from reality and we haven't solved the online continuous learning problem. I believe that as we turn AIs into agents which are constantly interacting with reality via high bandwidth token streams, we will have a lot more data to train with. I also believe that we'll start being able to train continuously on that data. Then even assuming that training is no more efficient than it is today, I think the extra data could make the difference.

I'm also not convinced that AI won't eventually be able to learn from as little data as humans do. I don't think it has to be the case, and I also don't discount the possibility of an AI winter that leaves AI less than efficient than humans are for a long long time maybe even forever. However I also feel like we may come to understand why humans learn so fast, and might be able to transfer some insights into artificial systems. I also know that people will be trying very hard to solve the AI energy and data usage problems, since their major threats against large-scale AI adoption. So we'll be trying really hard to do it and we'll have a blueprint for how to do it - our brains. That means there's a chance we'll crack that problem.

Finally the regurgitation issue is irrelevant to intelligence - just like it would be irrelevant if the brain is secretly just regurgitating stuff it learned. Because the brain can also do novel things.

Furthermore we know that llms can learn and usefully reason about context information outside of their training distributions. This is called in context learning.

For example if I come from a culture that the AI was not really well trained on, I can give it four or five examples of values that are important to me in that culture, and then it will be able to extrapolate how to apply or respect those values in situations that I present.

And again here's the kicker- it'll do this more faithfully than the average person. Remember that if you tell a person five values from a culture outside of their own, and ask them to uphold those values... Perhaps half will just get angry and give you some kind of racist slur, and then 80% of the remainder will lack the empathy and mental flexibility to do a good job.

Finally I need to point out that I have studied AI for over two decades out of books starting from the '80s, then the '90s, then the 00s and 10s. And the change in the literature and capabilities has been unreal.

Perhaps you are forgetting how feeble AI was before, or simply not putting it to use. There are many many tasks that no AI from over 3 years ago could have touched, and now suddenly you can do it for just a $20 a month subscription.

The change in capabilities is so drastic that I wonder if you're simply discounting that change because you're not using AI, comparing it to old AI, or seeing it enable things that no AI before could have possibly done, no matter how hard you tried.

So to conclude, the change has been too great, enabled too many new things, and taking such a big departure from old AI, and consistently outperforms humans on so many tasks that I find important, that I feel it would be not only senseless to say that there isn't some intelligence there - some useful information processing capability that I can depend on and rely on more than a human, in many tasks and settings where humans are consistently bad. In fact it would be harmful for me if I didn't realize that these things have changed, because I would not be benefiting from them.

It was a speech recognition issue! Yes, I'm talking about the dirac delta function.

Hey scrubs, I own my own company, mentor college kids in category theory and open source programming, and generally love life.

Have a nice day!

You have no proof that every modification of the architecture will continue to have hallucinations. How could you prove that? Even LeCunn admits that the right modification could solve the issue.

You're trying to make this point in a circular way - saying it's impossible just because you say it's impossible - for some reason other than trying to get to the bottom of the truth. You want to believe that there's some kind of guarantee that no offspring of the auto regressive architecture can never get rid of hallucinations.

I'm saying they're simply no such guarantee.

You define the density function of the shell using a direct Delta function and the derivative of a direct Delta function. That means that on the shelf the density doesn't take any real value.

But we can formally integrate Delta functions and their derivatives. When we perform the mass integral we end up with the result m = 0.

It's an artifact caused by allowing generalized functions as our density functions

A problem is that we don't know how to physically interpret a distribution as a density function. Mass distributions caused by matter all have positive real valued functions as density functions. So if we have these structures in the universe, their cause is probably not physical matter!

It would be as if god put his thumb on the scale, and his finger print indented a concentric series of shells where the mass density goes completely nonphysical.

The sense in which they are defects is that on the shell, the density function is a distribution and takes no real value. Distributions are singularities/non-functions that we get when we take certain limits of ordinary functions, or solve certain differential equations with generalized functions.

The distribution used works roughly as if we overlapped two equal positive and negative mass shells. But there are some extra details that ensure there is a net inward force for matter situated on the final shell. We'd have to actually work the math to really understand why that force appears, without hand waving.

You're off on two points; number of shells and the mass of the shells.

First, the number of shells is finite. This actually gives us one of the points where we can test this theory:

Namely, if there are discrete shells, their effect on gravitational lensing will introduce artifacts that we may be able to detect. We would observe the edge of a shell as a shearing zone in the gravitational lensing around a massive structure, where the amount of lensing jumps dramatically, significantly magnifying the cosmic background in the shear zone. These would look like rings around gravitational lensing sources.

This discrete lensing artifact - these rings that might appears - could let us reject some dark matter theories, which predict smoother lensing with no rings.

Second the shells themselves do not have positive or negative mass. They have density functions which are a sum of a delta distribution and the derivative of a delta distribution. On the shell singularity itself, the density function breaks down - from being a function into being a distribution. So we can't say it's positive or negative mass - it's a singularity without any mass.

The reason we invoke positive and negative shells is because we can think of distributions as limits of ordinary functions. We can think of the topological defect shell -as if- it was the limit of a pair of positive and negative physical mass shells of positive thickness as they merged together into a shell of zero thickness. In the limit, there is no longer any externally observable mass at any enclosed radius. There's no longer any finite mass density function that properly describes the singular shell's density - the density has become a distribution.

The corruption is systematic and automated, so the struggle and response must be systematic and automated as well as forceful. Not knee jerk. But more organized than simple individuals doing the right thing. And much more forceful.

It must become very dangerous to be an exploiter of the human condition. Must become very dangerous for capitalists, executives, marketers, and so on.

Force and systematization/automation are required to gain power. And power is required to end capitalism.

You're saying a bunch of nothing

You think the person building technologies that cause human suffering values human life? Come on now.

It seems a bit shady, with respect. I want to talk about papers and publicly available systems that are out there. I'm not going to engage with an amoral web 3 head who's got some proprietary code in his trench coat and wants me to walk into the alleyway to do a deal. I'm building an open source system, so others can train their own databases, not just me.

Are you interested in publishing your code as an open source system? Then I will pledge at least 40 man hours to the code base, doing tasks that you've been too busy to take care of, but which will make your code and clients happy.

I have background in image processing and compression , python packaging/pypi deployment, CI/CD, testing, and benchmarking. I could definitely help you knock down some of your tickets.

What do you think? A little too brusque on the approach? Or a potential colab?

Collaborators always have their excuses, and yet freedom fighters always seem to ignore those very excuses. What makes the difference in outlook?

Great, now I'm imagining GPT flexing its roided biceps while making sigma faces, as edgy incel music goes hard in the background with spiky synths and a boy's choir.

After seeing how awesome the showcase looks, I'm not even sure I'm mad about this, lol