[Kosturdistan]

A lot of loonies make this claim, but Tom's work is credible enough that it's being reviewed by linguistics experts at Rutgers and Cambridge. Additional validation: his approach produces results. He's translated over 300 words, and that's never been done before, and his solution actually solves some problems in Linear B. Tom is an AI engineer, and Claude Code was key to his work. Disclosures: I know Tom socially, and I wrote the post at the link.

[kubb]

Let's wait until it's been verified.

[gus_massa]

I agree. The post has too few information. Also

>> reviewed by linguistics experts at Rutgers and Cambridge.

Here in Argentina, near 2005, we had like 5 guys that claimed to have 5 independent solutions of the Goldbach Conjeture. Each one got a PhD student that volunteer to read it, discussed the obvious problems with the author, tried to help to solve them and after a few months of back and forth they concluded that none of the solutions were correct or has an interesting insight. Nobody was surprised about the that, but some wanted to give them a try.

Until there is a official report by Rutgers or Cambridge, it doesn't mean too much.

>> He's translated over 300 words

Where is the table of translations?

[Kosturdistan]

Skepticism is appropriate until the experts bless the work. I will point out however that all of the words Tom has translated provide strong support for his proposed phonetic values. And that's why I published the information prior to confirmation, along with the appropriate caveats.

[mikestorrent]

You're absolutely right! We've opened a ticket with the Linear A folks, hopefully they'll get back to us soon with an update as to whether we've got it correct or not. Hang tight!

[kridsdale1]

This comment sure is load bearing.

[mikestorrent]

It's the veritable smoking gun

[TeMPOraL]

Regardless, we should stand ready, loaded for bear.

[saagarjha]

A Linear ticket, hopefully

[bawolff]

How does an expert even verify something like this?

[red_admiral]

They verified Linear B against a new tablet that turned up in a dig after the Kober/Ventris* solution had been published. It had pictures of jars with no or one or two handles, and the claimed Linear B for "two handled jar" and such next to the correct picture.

* Ventris' publication, but given Kober's contribution to the work they should really share equal credit. I like to think Kober would have got there on her own if she had access to the larger corpus that Ventris had (the Pylos tablets) and a comparable amount of free time and money available.

[pfdietz]

Linear A/B testing?

[Kosturdistan]

You look at the proposed sound values and compare it to other known languages. Languages from the same family share grammar and vocabulary.

[sigbottle]

I mean it's not like anyone could objectively go back in time and query ancient civilizations for what they meant, but presumably it means the verification heuristics, they have currently, pragmatic success, and expert solidarity means that it is "verified"

[yorwba]

Then why is there no link to the actual write-up?

[GavinMcG]

Presumably because it hasn’t yet been published?

[Kosturdistan]

The only write-up at the moment is my blog post, hopefully that changes in the coming weeks.

[Sniffnoy]

The blog post mentions a draft of a manuscript though. I was expecting something like a preprint. He's not willing to post that draft yet?

[Kosturdistan]

I have seen and read his draft article, he's not comfortable sharing it publicly yet since it's being reviewed by experts.

[ahknight]

But the Internet nerds wish to blindly judge something they know nothing about so they can feel better with the assumption that they could have done better somehow. How will they be appeased if the document they will say they have read and understood (without having done either) is not available to point at? How, I ask?!

[m0llusk]

It seems this is still extremely early in the process. There is an apparent finding that was shared. Evidence which would be the basis for a paper is "being reviewed by linguistics experts at Rutgers and Cambridge". So they are trying to do the right thing by talking about what they believe they have done but holding off publication and serious claims until later. The general idea that written forms can be categorized by systems built with Claude could be applied to other as yet undecipherable languages could be used by other interested investigators just with what is discussed here.

[sillysaurusx]

> The general idea that written forms can be categorized by systems built with Claude could be applied to other as yet undecipherable languages could be used by other interested investigators just with what is discussed here.

Could you rephrase this or explain it more thoroughly? I don’t follow. What does it mean to categorize a written form by systems built with Claude?

[tyingq]

The same pattern/tech is generic enough that it might be able to solve other unrelated, and so-far undecipherable, written languages.

[kelseyfrog]

You can use Claude, like the author, to reproduce the result.

[_verandaguy]

This isn't really a reasonable approach, is it?

The original prompts aren't provided, nor is the original context; even then, you can't really treat a stochastic system like an LLM as a major component in reproducibility.

[ben_w]

> even then, you can't really treat a stochastic system like an LLM as a major component in reproducibility.

If you had the other things, being "stochastic" is not even remotely a show-stopper. Stochastic processes abound and are the reason the mathematics of statistics was developed in the first place, ultimately allowing us to create such things as LLMs.

When all the relevant steps gets published, I absolutely expect a lot of people to (attempt to) reproduce this work even though LLMs are stochastic.

[_verandaguy]

My issue with this is that it's a form of "soft" reproducibility, where it'll work for many (maybe even most!) people, but that depends on the way the original prompt was formulated (read on) and the state of the random noise in the system.

On the prompt formulation; prompts with very similar formulations (in terms of both semantics, hamming distance, or both) can lead to _wildly divergent_ outputs in my experience. It's not rigourous, and when that divergence happens, it's extremely difficult (arguably impossible, by nature of the architecture of transformers) to identify why the divergence happened and where.

[Kosturdistan]

Claude code was used to organize the material and to run simulations. The simulations were to determine the likelihood that the text was Semitic vs Tom got lucky. Tom has assigned probabilities to each of the syllables he has proposed sound values for.

[fragmede]

Sure it is. We're humans, not robots (well, I think I am, and I presume you are as well, but for all we know, we could be living in a simulation), so if the non-deterministic system decides to generate code that calls the variable foo one day and bar the next, as long as the code still does what's being asked of it, why do I care that the non deterministic system chose to call the variable something different when run on Tuesday? There's the computer science definition of determinism and the engineering result of "does it work", which are at odds. It's like the halting problem. We haven't solved the computer science definition of the halting problem, but give some C code with a loop that won't terminate to Claude, and it'll call that out as not halting.

[_verandaguy]

All things aside, I think this misses the forest for the trees on the halting problem.

It's not about being able to throw claude or codex at a loop and having it evaluate it for halting, it's about being able to do this for arbitrary code. Computer science rigourously defines the halting problem as not computable and undecidable. within the framework of using something akin to static analysis using any deterministic Turing machine.

There's not really a question of "solving" the halting problem like there's some as-yet unknown way of generally figuring out if arbitraty code halts. Turing proposed a proof in 1937 in favour of undecidability of what we now know as the halting problem, building on ideas first articulated by Church a few years prior.

Frankly, if anything, it's reasonable to say that the halting problem's been solved, just in the direction of undecidability rather than decidability.

Anyway, back to LLMs; as code gets more complex, the robot will need a bigger context window, more hardware resources, and more time, all of which will be variable due to the noise inherent in the system. It'll be difficult to put a useful upper and lower bound on how much computing power and time it'll take to figure out if a program ever halts. Which is all a bit moot, frankly, in the context of halting, but useful to keep in mind in the more general context of using these things as analysis tools.

[iwontberude]

Actually it is because Claude did the work and being a lay person isn’t really that high of a bar.

[Kosturdistan]

Claude helped, but did not do the work. This was a human dude who had a very helpful assist from Claude

[TeMPOraL]

> stochastic system

Every day when you lower your butt onto your chair, you trust a stochastic system enough to assume you'll rest on the chair safely and not spontaneously phase through, which would lead to rather gory and painful terminal experience.

Physics at macro scale is stochastic, which is a good reminder that stochastic != uniformly random. Expected distributions matter.

[ben_w]

While strictly true, QM has such small standard deviations as to be irrelevant on the macro for things like bums and chairs.

IMO a better example would be the stochastic nature of quality control in manufacturing.

[TeMPOraL]

> QM has such small standard deviations as to be irrelevant on the macro for things like bums and chairs

I was going to segue into thermodynamics as a backup example, but you made me think of something better.

> IMO a better example would be the stochastic nature of quality control in manufacturing.

How about, more specifically, food manufacturing? Or maybe, let's talk about cooking?

Cooking is as stochastic as it gets, and we handle it fine. It could be better - the better version is called "chemical process engineering", it's what cooking looks like when you care about quality and consistency of output, and can afford the equipment and process actually necessary for it. Regular people don't (i.e. neither care, nor can afford) - we call this cooking. It's an art, not a science, and people not only do it, but love it, and tie their identities to it, and build businesses around it, and a culture that embraces all the compromises (and calls the more serious approach "unhealthy").

[atrus]

somehow I suspect it was a bit more involved than: Claude, please solve Linear A.

[fragmede]

A little bit more. If you ask ChatGPT to "solve linear a" it thinks you mean linear algebra. If you specify that it's the Minoan translation problem, you get a table similar to the one that we get a glimpse of in the without access to the paper, we can't say how much more work the paper has than my gist.

https://gist.github.com/fragmede/bbf277d36a2398065f109484f34...

[smsm42]

You also have to add "make no mistakes"!

[Kosturdistan]

You are correct!

[justin_dash]

Unless if it was done by Fable!

[kelseyfrog]

The 'major insight' described in the article predates Fable's release by two week four days. It would be a complicated timeline.

[grey-area]

Amazing work and refreshing to see a well written and cogent post to summarise it. Would love to hear more about how he used Claude to help solve the puzzle.

[dwroberts]

You know him socially but is there a reason you’re writing this rather than him? It looks like he has his own web presence.

Cynical read would be you’re stealing his thunder a bit by prematurely announcing this before it’s fully confirmed

[jstanley]

Promoting your friends' work is hardly stealing their thunder. It's increasing their thunder!

[Kosturdistan]

Tom knows I'm a freelance writer and decided to give me the scoop. He's more interested in linguistics than he is in journalism.

[Conscat]

Isn't it customary for the author of a post shared on HN to leave a comment on the thread?

[dwroberts]

I’m not referring to the parent comment: The post is not written by the author of the claimed breakthrough.

[iwontberude]

What thunder? Claude did the work and used a human to interface with experience and causality better.

[Kosturdistan]

Claude helped, it did not do the work. It would have taken Tom more time to crack on his own, and it would have been harder, but the key insights were Tom's not Claude's.

[ben_w]

The thunder is as per the headline. Assuming it passes review.

One of the things I find weird with AI is how the dismissals of work that involve AI splits into two camps: like yours, saying the AI did the work while the human played no role and deserves no credit; and those saying the AI rips off its training data while the human using it played no role and deserves no credit.

[iwontberude]

I exist in both camps. Claude can’t launder human achievement into a different person. Claude stole it, but it’s still in Claude’s possession and is not transferable in any durable sense.

[ben_w]

> Claude stole it, but it’s still in Claude’s possession and is not transferable in any durable sense.

No human, individually or as a team, has been able to solve this to date.

To the extent this was Claude solving it itself and thus denying Di Mino any thunder, there was nobody to have stolen anything from. To the extent he has thunder to be stolen, it wasn't ever in Claude's possession.

[timschmidt]

Everything is a Remix: https://www.youtube.com/watch?v=X9RYuvPCQUA

Either all information is stolen, or none is. Can't have it both ways.