a-cosmology-group/acg#15

Yes, Watch → All activity will send you an email for all new discussions and comments, at the email address associated with your github account.

If you get this message by email, then it works! The best way to read discussions and participate is via this web address: https://github.com/orgs/a-cosmology-group/discussions, but you can also reply by email.

Thank you for trying it out, and for the vote of confidence. It is very helpful to understand what the hurdle is, thank you for pointing out that the Watch setting seemed optional to you!

(Note that there is a link View it on github at the bottom of the email, to get to this discussion directly)

@marmetl, given this comment (and the preceding discussion), perhaps in the Fall one tack would be to ask members to send you their github handle (by email). You can then invite them as collaborators on the repo: they would then only need to “accept”, in which case the Watch setting is automatically set to All activity.

Hi Frederic,

One of the problems with the github to email system is that it does not include in the email thread comments made on the same thread by you (the person receiving the email) in git hub. For instance a couple of weeks ago I received an email update to an existing thread consisting of a remark by Louis that was clearly commenting on something someone had said but I could not find the context in the email. When I went to github the context was obvious - he was commenting on a comment I had made - a comment that was not in the email thread. It is not unusual in ACG for comment responses to have a lag time of several days, so this behavior is going to be a further source of confusion for new users.

In Gaggle Mail, your comments were added to the email thread by GM whether you made them in email or using the mail tool. This resulted in a double entry when replying via email - annoying perhaps but less confusing than no entry at all when using the github mail tool. Hopefully this is just a forwarding-rule setting that can be switched off/on. Thanks for all your work on this.

Bud

Thank you for this report and explanation, your are right. My perspective was biased then, because my habit is to click on the “View it on github” link at the bottom of the email. I presumed, wrongly, that others would do the same…

My thoughts were the same though, and I actually had a good giggle at that one. 🤣

Here is the layout of a soccer field.

We already have incredibly much freedom on our ACG playfield. But the field is limited by red lines. To indicate that it is game over outside them. A few such offsite positions are indicated with balls in red. Sure, it happens more than once that a ball goes offsite. What we must do then is throw the ball into the playfield again. The worst thing we can do is to change the boundaries of ACG. Because once such a process is started, there is no end to it. (Now I could throw up bad examples in society, but I don’t)

it might be nice to have a section for threads that are slightly off-topic.

Defninitely NO, sorry, I think it’s an extremely bad idea.

@luisenrique-arch thanks for your suggestion, but I will keep this forum focused exclusively on cosmology. This was its goal when ACG was founded in 2004 (and initially hosted on Yahoo!)

However, the main reason for limiting discussions to the topic of cosmology is that the ACG forum needs constant attention to avoid an increasing entropy, being taken over by cranks, and then fall apart (like it did in 2010). This maintenance requires reading the posts, managing requests, and keeping discussions under control (scientifically and socially :fearful:).

Doing all this takes time away from my research in cosmology, so the ACG forum must be worth my time by enabling an exchange of ideas with the scientific community involved with cosmology.

Other forums are available to discuss the topics you mention (like Quora, Reddit, an instance of Mastodon, etc.) You can also follow YouTube channels and join their discussion group for a small fee (you already know about DemystifySci, See the Pattern’s Gareth Samuel is also interested in these topics), or Facebook groups, etc.

And if you don’t like any of these, you can start your own on GitHub!

@RedshiftDrift

Thank you for your thoughtful reply. I completely understand where you’re coming from.

I only discovered this forum fairly recently. I didn’t even know there was an earlier version of it that had to be shut down after descending into chaos. Had I known that, I never would have made this suggestion.

The main reason I made that suggestion is that I’ve found the people on this forum to be genuinely likeable, and I appreciate the open-minded attitude that many of you have toward different ideas. So I thought, “Hey, it could be fun to discuss other scientific topics with them, especially since they’re already open to alternative cosmological models.” But if doing that would end up creating the kinds of problems you described, then I’m happy to withdraw my suggestion.

@luisenrique-arch Thanks for the kind words about this group. (Some “filtering” was applied to keep only the best people :wink:)

GitHub is not structured for private messaging, but most ACG members have a website or papers from which you could get their contact information. Others remain anonymous.

Since you are already on GitHub, you could create your own public repository (it’s free) and create your own Discussion forum. Then advertise your forum here to attract the people you like, and start building your group from there.

@ericcosmos

I’ve noticed the same thing myself. If you discuss alternatives to the Big Bang with ChatGPT or Gemini without providing any context, they usually fall back on the standard talking points about how the Big Bang explains the observations, how well the Lambda-CDM model works, and how alternative models fail to account for the evidence. However, if you actually send them papers to analyze, they start engaging with alternative models much more seriously instead of just repeating what the consensus says.

A while ago, I uploaded David Crawford’s paper on Curvature Cosmology to ChatGPT. It told me that the model was genuinely interesting, that the mathematical derivations were internally consistent and that the calculations appeared to be correct. I later did the same with the various pro-TL papers written by Nikita Lovyagin, Alexander Raikov, Vladimir Yershov and Yuri Lovyagin, and I got a very similar response. ChatGPT said the papers were well written, that the calculations were correct and that the criticisms they raised deserved to be taken seriously. I even sent it @RedshiftDrift’s paper on Spectral-Transfer Redshift, and I received the same response: everything works.

So yes, if you give AI actual papers to analyze, it will provide thoughtful responses and take alternative models seriously.

I have had some really good results while working with AI. No, you can’t just ask it to write a wall of text and equations and expect it to ‘solve the universe’, but you can gain a lot of ground by asking it targeted - focused - questions, and then interrogate it on the response.

Some time ago I started a website (ditched it already) where I took fringe papers on aether and ask AI to score it. This is where I learnt what kind of questions to ask to get consistent results. E.g. you may ask AI to review GR it will tell you it’s the best thing ever. Then ask it a question about MOND and dark matter and dark energy and the cosmological constant … and then ask it about GR again and, well, suddenly GR is not so great after all. <- Just an example I made up. If you find AI starts ‘drifting’, start a new chat window asap.

ChatGPT: I found it tried too much to flatter the author/reader by telling you how well you are doing and how you did such a great job, while it’s not actually working. With science you need brutal responses. Gemini: Worked very well, except soon (every time) started using the document I gave it as a ‘source of information’, instead of continuously checking it against other models. DeepSeek: My clear favourite by far. Yes, it can lose the plot entirely sometime, but it has geven me valuable insight even when I wasn’t asking for specifics. This is what you want for exploratory models. It also tends to get ‘sold’ on the model after a while in the chat window so caution is advised.

On all of these it’s useful to do quick complex equations, instead of doing it myself, it’s much easier to check if AI got it right. 9/10 they do but that 1/10 will bury you if you don’t spot it.

These were all unpaid versions. Not sure if anyone had better responses from other LLMs.

A few remarks from my side.

• Neural Networks are at the foundation of Artificial Intelligence.
• I started to read a voluminous book about it and produced a viXra article. It barely scratches the surface though (: what do you expect)-:
• Our pope Leo XIV has written an Encyclical Letter about AI. Worthwile.

As about the above comment and replies: AI is truly an Intelligence. With other words: intelligence is not limited to humans. It has been stated already in the 1969 (!) book by Stefan Denaerde: dat intelligentie, geheugen en normbesef materiële zaken zijn (that intelligence, memory, and a sense of norms are material things). That’s the reason why machines can be intelligent. Next question: what makes us different from a machine then?

@Francois-Zinserling . >These were all unpaid versions. Not sure if anyone had better responses from other LLMs.

A paid version is e.g. Claude Opus. I have been told by someone in the know that it shows far better responses than free LLMs. And I have seen what it produces when fed with my own homepage on cosmology, but decided not to publish yet.

the newsletters feel like they’re written by an underdog that could actually pull it off - @ericcosmos

I learned a lot writing these newsletters. In addition, I reviewed papers I think I could explain with STz, so this made the newsletter consistent over the years.

yes, if you give AI actual papers to analyze, it will provide thoughtful responses and take alternative models seriously. - @luisenrique-arch

That’s a problem though; AI does try “too much to flatter the author” and can be made to agree with anything. All the papers you fed the AI cannot be right since some contradict others, but AI can’t figure which one (if any) is right.

that 1/10 will bury you if you don’t spot it - @Francois-Zinserling

I did try free AI for a derivation of the CMB temperature using only first principles and STz. Guess what it gave me? 2.7 K! Then I found that it had done a bad job for the derivative of a polynomial, had cherry picked terms that would give 2.7 K as a final result, and had given nonsensical arguments to support that answer. Without guidance, AI is a risky endeavour.

@luisenrique-arch I asked the opposite question about my paper: “What would mainstream cosmologists say about my paper?” Answer: my paper is all wrong :rofl: :rofl:

started using the document I gave it as a ‘source of information’ - @Francois-Zinserling

All LLMs do that, even the paid ones. That’s why the AI will agree with you, because you’ve trained it with your own ideas by giving it your paper!

what makes us different from a machine then? - @HanDeBruijn

As opposed to AI, we are not trained to regurgitate what other people have written. Humans generate genuinely new ideas.

@RedshiftDrift

You’re right, AI is designed to agree with whoever it’s talking to. However, there are settings that can try to tone this down. For example, on ChatGPT I’ve completely disabled the option to always agree with me, and in fact, every time I talk about cosmology with ChatGPT, it’s practically impossible to make it change its mind unless I bring it actual research papers. The same goes for other topics where I hold positions that aren’t aligned with the mainstream.

In any case, I don’t ever use AI for scientific topics, unless it’s about things where there’s universal consensus, like the size of the Earth, the size of the continents, the composition of the atmosphere, the distances between the Sun (or the Earth) and the other planets, etc. I usually just use AI to translate texts from Italian (my native language) into English.

@RedshiftDrift writes: >As opposed to AI, we are not trained to regurgitate what other people have written.

Oh well, I think that many (if not all) of us have been trained to regurgitate what other people have written. It’s called supervised learning, very much the same process as with the training of neural networks in AI. >Humans generate genuinely new ideas.

Yes, that is a missing link, I think: CREATIVITY !!

I think that many (if not all) of us have been trained to regurgitate what other people have written. - @HanDeBruijn

Not me, I’ve always had the suspicion that people are trying to impose their beliefs for control, enhance their career, or to make money. E.g. the bible, Santa Claus, spirituality vs science (remember “The Tao of Physics”?), statistics (like “Thirteen against the bank” which claimed you could win at roulette!), philosophy proving the existence of God, history showing that all British people were bad because the invaded Québec (where I’m from), the “point particle” in quantum mechanics, etc. etc. All nonsense.

Then I read about the Big Bang model with the same suspicion: someone is pushing an idea to enhance their career.

Yes, we were trained in school to behave like AI, but I think people on this forum must do better! Hence my dislike of AI and rejection of posts that regurgitate what others have written.

What is interesting in AI is the transformer neural network architecture (on which LLMs are largely based on today). It reveals that “predicting what comes next” is a key aspect of what we call “intelligence”. This is why language has for a long time been considered as a loose proxy for “intelligence” and “thought” in general (see Wittgenstein, for the most flamboyant example!).

Language is better viewed not as the ability to utter words or communicate, but more comprehensively as the machinery for “predicting what comes next”. AI is not simply regurgitating what exists (which presuppose replicate encoding, absent in LLMs), it is building an inference engine. This is what biological brains do! Creativity is not a different operation (inference), but a question of degree (infer beyond the existing corpus).

Great progress in science has hinged on humans removing themselves or their narrow perspective from some self-declared importance or self-built pedestal. This is the well-known Copernician principle (heuristic of course): assume you’re not special, and the science gets better. Modern AI is that same move, for what we call intelligence.

Examples:

• Earth is not the center of the Universe, neither is the sun or the galaxy (astronomy, cosmology).
• Humans as just another species (evolution).
• Life isn’t chemically special, nor is Earth (chemistry).
• Mind isn’t a special substance (psychology, neuroscience).
• “We are more of an environment than an organism” 😉, microbiome symbiosis (biology, physiology).
• No special frame of reference (relativity).
• No absolute physical state, only entangled observations (quantum mechanics).

I am not saying that any of the above is settled or true, they should continue to be probed and questioned. But outgrowing the human perspective seems to be a genuinely productive attitude in science. Another wonderful promise in the same vein — and of interest to this community — is removing the very human conception of time and space (even relative!) from fundamental physics (removal of time arguably gestured at by the Wheeler-DeWitt equation).

@ftessier writes: >assume you’re not special, and the science gets better.

That’s where my heart emoi is for. >But outgrowing the human perspective seems to be a genuinely productive attitude in science.

Having noticed the extreme individualism in this group, I would like to propose the following extension. But outgrowing the my own model perspective seems to be a genuinely productive attitude in science. >Another wonderful promise [ .. ] is removing the very human conception of time and space (even relative!) from fundamental physics.

That’s where my confused emoi is for. The concept of space-time instead of space and time is the worst thing that has ever happened to physics. Even worse is the completely incomprehensible Wheeler-DeWitt equation. Following Sabine Hossenfelder, I call such stuff BS physics.

@HanDeBruijn the concept of space-time as an intertwined entity is not as foreign if you consider the posiibility of an aether. Even Einstein acknowledged in ~1920 that we require an aether for physics to work. By that time SR and QM had gained such momentum without an aether, coupled to the fallacy that Michelson and Morley disproved the aether, that Einstein’s comment gained no traction.

Let’s say hypothetically we are emerged in a Planckian aether, with Planck length ‘photons’ isotropic and each has length = $$L_P$$. The ‘photons’ pass a length $$L_P$$ in a time $$t_P$$ which naturally leads to the speed of light $$c=L_P/t_P$$. (you can verify this relation) If nothing can move faster than the aether then that sets the universal speed limit, $$c$$. You could start by inserting $$c$$ and then getting $$t_P$$ or $$L_P$$ but these are all constants now so the relation is set - for an isotropic aether. With a discrete length unit of $$L_P$$ and a universal ticking rate of $$t_P$$ the foundation for space-time is set. These are inseparable unless you can alter the speed of light.

Incidentally we could, e.g. in a gravitational well c is lower and clocks tick slower. Then there exists a metric:

$$ds^2=-f(x)c^2dt^2+{f(x)}^{-1}dx^2$$

where f(x) is the rate of time flow (change) due to the presence of mass. f(x) is how mass curves space. In GR, to first order:

$$f(x)=1-2GM/{c^2r}$$

Long story short. We do not know what space is nor do we know what time is, and space-time seems an unnecessary complication, but it is actually a simplification of both.

@Francois-Zinserling I’ve said it before, I’ll say it again: I think Lorentz was on the right track. Treating space-time as a purely geometrical entity only leads to absurd complications and nonsensical paradoxes. Physics needs an aether and a preferred frame of reference. It simply doesn’t work without both.

Having noticed the extreme individualism in this group,…

This is a sad state of affairs, especially for a group like this. Our only common ground appears to be ‘NOT λCDM’ but there is no confinement to be measured in ‘everything else’.

Still, that is a good common ground to have and I highly value the input and criticisms of all in this group.

I support efforts to derive physics in a way where “time” and “space” are emergent properties rather than axioms of the theory. Anytime we write “t” or “x”, ill-defined anthropocentric concepts, we are making a sweeping assumption that Nature can fundamentally be described in the same way our brains are interpreting it. The “t” and “x” variables have been useful up to now, manifestly, yet they betray a singular and limiting world-view. My wager is that once we find how to build physics without these concepts baked in, much of physics, including cosmology, will leap forward.

a length $L_P$ in a time $t_P$ which naturally leads to the speed of light $c=L_P/t_P$.

Trivially so, @Francois-Zinserling , because $L_P$ and $t_P$ are defined in this way. Ipse est: circular reasoning.

Then there exists a metric:

There exist no other metric than the Euclidean one for space: $ds^2 = dx^2 + dy^2 + dz^2$. And there exist two separate time frames: Atomic time and Orbital time, $t$ and $T$, where

$$ \left(1+\frac{T}{A}\right) = \exp\left(\frac{t}{A}\right) \quad \mbox{with} \quad A=\mbox{“Age”} $$

With some name-changing and $\large A=\frac{1}{H}$ the Time Dilated Past hypothesis by @mikehelland is recovered.

We do not know what space is nor do we know what time is

Denied. We do know very precisely how Eclidean Geometry works for empty space. And we do know how Atomic clocks and Orbital clocks accurately measure time.

@luisenrique-arch .

Physics needs an aether and a preferred frame of reference.

Physics does NOT need a single preferred frame of reference. Fictitious forces are not fictitious , they are real.

@ftessier .

Anytime we write “t” or “x”, ill-defined anthropocentric concepts

How about “m” for mass or “T” for temperature, aren’t these ill-defined anthropocentric concepts too?

Of course they are ill-defined too! That mass is emergent is already well-accepted in traditional physics, and T is ill-defined both because 1) technically, the Boltzmann constant $k_\text{B}$ is not covariant; 2) its definition leans heavily on entropy, which is perspectival (and incidentally our only link to time in physics, and in our thermodynamic brains…).

We ought to lift the foregoing discussion(s) to separate discussion threads, outside this “Welcome to ACG!” one.

@ftessier . A similar discussion does already exist: space, time and philosophical considerations from an experimentalist’s point of view. Wouldn’t it be a better idea to join forces there, instead of creating more of the same? (: nobody seems to be interested)

@RedshiftDrift . >- you have built the universe proposed by AI and confirmed that it works better than the observed universe.

Sorry ?! How ever can a universe be built, in theory, that works better than the observed universe ?

AI will be accepted if [ .. ]

Here in the Netherlands we say: “Als je ze een vinger geeft, dan pakken ze je hele hand”. Okay, let Google Translate (AI?) be my friend: Give them an inch and they’ll take a mile.

Sorry ?! How ever can a universe be built… - @HanDeBruijn

It can’t be done, therefore no AI theory will be accepted on ACG Discussion!

As for giving ‘een vinger’, you can guess which one I give to AI :rofl: