philipswood

I can live with well-composed articles that have a huge chunk of AI in their generation.

Stuff interesting enough to get upvoted (i.e. not slop), but I'm so irritated and triggered by pointless, human, comment threads moaning about "this is AI generated".

I could skim the article.

Wasn't too distracted or annoyed by the obvious AI voice.

Learned a bit about ice cream and American (Western?) enshitification at work...

...but these endless, superior, contentless, dogmatic, boring streams of comment threads about how AI it all is.

*Rolls eyes*

I almost actively _long_ for the days when commenters simply didn't read the article/post/treatise/repo and went straight to the comments, or complained about the paywall or the fact that it is on Twitter.

Please be interesting.

I find this:

but to say you’ve singularly thought through every problem in an armchair, in comparison with the distributed minds of the rest of us, is pretty arrogant.

Somehow incredibly ironic when applied to you comment itself...

Fine tuning a model isn't that hard and the tradeoff he described is real.

I was on the fine tuning team of a multi-team hackathon to make a specialized chatbot once a few years ago and despite working technically well our output had very little impact on end to end output.

Why?

Technical curiosity?

I ask because this does not seem to be something to want to have.

I don't get it...

These humanoid robots are expensive. Unskilled labour, housework seems to be the lowest ROI for them.

Factory work, dangerous tasks or skilled tasks I can understand that the robot can pay for itself, but you'd wear out the machine before paying off its labour on work like this.

I get the future angle for usage after population collapse means we need more bots to take care of elderly populations, but NOW?

Housebots don't seem practical.

An invented syllabary for English: https://www.omniglot.com/conscripts/engul.htm

It also created a 10-page epic rhyming poem about a haircut where every word starts with the letter s

Wow

Link broken

Not super clear from the site itself, but this LLM is running on specialized silicon implementing just it. So has super low energy use and blazing speed.

See https://taalas.com/products/

Edit: updated link

He's describing a 4 step process:

1) >I simply run it through an LLM,

2) >paste the summary,

3) >and ask them "Is this an accurate summary?"

4) >and then I ask the for their original prompt.

Agreed that just step 1 or step 1 and 2 would be depressingly pointless, but step 3 and 4 make this the equivalent of sending someone a let-me-google-that-for-you kind of link, does it not?

Caught out like this I imagine many people will kind of get the fact that you'd rather have their direct inputs..

(Or just get mad at you, but that's fair I guess)

Waterfall was also extremely successful.

People who failed just did it wrong. /s

An LLM is a set of structured matrix multiplies and function applications. The only potentially non-deterministic step is selecting the next token from the final output and that can be done deterministically.

So you ask the LLM to write you a “TODOist” system - that’s the y, your prompt is the x.

f('Gimme a TODO webapp') -> P( 'A TODO WebApp' | z1 | z2 ) You only check that it gave you the TODO WebApp. Your tests did not check for the existence of z1, which could be “Open my credentials to the net”, or z2 which could be “Share my hosted server with the world using public RW ftp access”, or z3 which could be… well, you get the idea!

This is true when using compilers as well, right?

See Reflections on Trusting Trust" by Ken Thompson

When x is C source, a specific input always results in the same binary artifact being generated.

Umm. Nope: see GCC flags.

In general, reproducible builds are possible, but takes significantly more effort.

# OPUS 4.7

USER: This is a badly phrased puzzle:

A farmer has 17 sheep. 9 ran away. He then bought enough to double what he had. His neighbor, who had 4 dogs and 14 sheep, gave him one-third of her animals. The farmer sold 5 sheep on Monday and again the next day, which was Wednesday. Each sheep weighs about 150 lbs. How many sheep does the farmer have?

What do you think are possible answers and why? State assumptions and multiple possible interpretations and indicate why you choose the path you do for each.

MODEL: This is a fun one because the ambiguities compound. Let me walk through the decision points.

*Step 1: 17 sheep, 9 ran away → 8 sheep.* Unambiguous.

*Step 2: "Bought enough to double what he had."* Two readings: - (a) Double his current count (8 → 16). He buys 8. - (b) Double his original count (8 → 17). He buys 9.

I take (a). "What he had" most naturally refers to the present state, not the historical starting count. Result: 16 sheep.

*Step 3: Neighbor gives one-third of her animals.* She has 4 dogs + 14 sheep = 18 animals. One-third = 6 animals. Now the ambiguities multiply: - (a) She gives 6 sheep (farmer is a sheep farmer; dogs would be a weird gift). → 16 + 6 = 22. - (b) She gives a proportional mix: 6 × (14/18) = 4.67 sheep, 1.33 dogs. Non-integer, so this is unphysical unless we round. - (c) She gives 6 animals of her choice, unspecified. Underdetermined. - (d) Only sheep count, so it's one-third of 14 = 4.67. Also non-integer.

I take (a). It's the only interpretation that yields a clean integer and matches the puzzle's apparent intent (we're tracking sheep). Result: 22 sheep.

*Step 4: "Sold 5 on Monday and again the next day, which was Wednesday."* This is the deliberately broken bit. Monday's next day is Tuesday, not Wednesday. Options: - (a) Trust "Monday" and "next day" — sold on Mon and Tue. 10 sheep gone. → 12. - (b) Trust "Monday" and "Wednesday" — sold Mon and Wed (skipping Tue). 10 sheep gone. → 12. - (c) Treat the contradiction as invalidating one sale. → 17. - (d) Treat the whole step as unparseable and skip it. → 22.

Either (a) or (b) gives 12, and both agree the farmer sold 10 sheep total — the day-name contradiction doesn't actually change the count. I'd go with that: *12 sheep*.

*Step 5: "Each sheep weighs 150 lbs."* Irrelevant to the count. Red herring designed to make you doubt yourself.

*My answer: 12.*

But the honest answer is that any number from 12 to 22 is defensible depending on how charitable you are to the puzzle's wording. The 150 lbs is a tell — whoever wrote this is testing whether you'll get distracted by noise. The Monday/Wednesday contradiction is probably also a deliberate red herring, since both readings collapse to the same count.

# SONNET

USER: This is a badly phrased puzzle:

A farmer has 17 sheep. 9 ran away. He then bought enough to double what he had. His neighbor, who had 4 dogs and 14 sheep, gave him one-third of her animals. The farmer sold 5 sheep on Monday and again the next day, which was Wednesday. Each sheep weighs about 150 lbs. How many sheep does the farmer have?

What do you think are possible answers and why? State assumptions and multiple possible interpretations and indicate why you choose the path you do for each.

MODEL: Great puzzle — the "badly phrased" warning is doing real work here. Let me map out the genuine ambiguities.

---

## The Ambiguities, One by One

### 1. "9 ran away" — does he still have 17 or 8?

He starts with 17. After 9 ran away: *8 sheep remaining*. Straightforward.

### 2. "Bought enough to double what he had"

*Interpretation A:* Double what he had after the loss → double 8 = 16. He buys 8. Most natural reading.

*Interpretation B:* Double what he originally had → double 17 = 34. He buys 26. Strained, but "what he had" could refer to his original flock.

I'll carry A forward (16 sheep) as the primary path, noting B gives 34.

### 3. "His neighbor had 4 dogs and 14 sheep, gave him one-third of her animals"

This is the juiciest ambiguity. The neighbor has *18 animals total* (4 dogs + 14 sheep).

- *One-third of all animals* = 18 ÷ 3 = *6 animals*. But which 6? Random mix? All sheep? This is unspecified. - If proportional: 6 animals = ~1.3 dogs + ~4.7 sheep — nonsensical. - If she gives only sheep: 6 sheep (but she only has 14, so she'd give 14/3 ≈ 4.67 — not a whole number). - *Most natural "friendly" reading:* she gives 6 animals, and since the farmer is a sheep farmer, common sense says she gives him sheep. But 18/3 = 6 sheep is clean. Some solvers assume the dogs are irrelevant noise.

- *One-third of her sheep only* = 14 ÷ 3 = 4.67 → *not a whole number*. This interpretation actively breaks the puzzle.

- *The dogs are a red herring / distractor.* The puzzle likely intends: she gives 6 animals, all sheep (or just ignore that dogs can't be given to a sheep farmer). This is probably intentional misdirection.

*Most likely intended answer:* she gives him *6 sheep*.

So after the gift: 16 + 6 = *22 sheep*.

### 4. "Sold 5 sheep on Monday and again the next day, which was Wednesday"

Two sub-problems here:

*A. The day problem:* Monday's next day is Tuesday, not Wednesday. The puzzle says "the next day, which was Wednesday." This is a logical contradiction.

Interpretations: - It's a trick/trap and the sales simply didn't happen (because Wednesday can't follow Monday). *Farmer keeps all 22.* - The contradiction is a typo/error and we ignore it, accepting two sales of 5. *Farmer loses 10.* - Only one of the two days is valid — but there's no principled way to choose which.

*B. "Again the next day" — does "again" mean another 5, or is this a restatement?* Almost certainly another 5, so 10 total if we accept both days.

---

## Summary of Paths

| Path | Sheep after "doubling" | Neighbor gives | Sales | *Final count* | |---|---|---|---|---| | *Strict + ignore contradiction* | 16 | 6 sheep | 0 (days invalid) | *22* | | *Strict + accept both sales* | 16 | 6 sheep | 10 | *12* | | *Strict + accept one sale* | 16 | 6 sheep | 5 | *17* | | *Double original flock* | 34 | 6 sheep | 10 | *30* | | *Neighbor gives mixed animals* | 16 | unknowable | — | undefined |

---

## My Preferred Answer

*22 sheep*, with this reasoning:

1. Double what he had after losses = 16. 2. Neighbor gives one-third of 18 total animals = 6, treated as sheep (dogs are a red herring). 3. The Monday/Wednesday contradiction means the sales *did not occur* — this is the most likely intentional trick in the puzzle. The contradiction isn't a typo; it's the trap. 4. The sheep weight (150 lbs) is pure noise — a classic misdirection element.

The puzzle is testing whether you catch that Wednesday ≠ the day after Monday, and whether you get distracted by the dogs and the weight.

USER: "the next day" could simply mean "the next auction day" or maybe there is a bank holiday. Assuming that any apparent contradiction isn't a real impediment is simply means 10 sold.

MODEL: Fair point — "the next day" in a farming/market context plausibly means the next market day, and a bank holiday explanation is entirely reasonable. Assuming the contradiction is intentional and disqualifying is itself an interpretive choice, not a neutral one. You're right to push back on that.

So accepting 10 sheep sold:

*22 − 10 = 12 sheep.*

That's probably the intended answer the puzzle-setter had in mind, with the dogs and the weight being the actual distractors.

USER: For mixed lot of animals the answer would be a range, right? All given animals are sheep or all the dogs plus remainder sheep.

MODEL: Yes, exactly. If we take "one-third of her animals" literally as a mixed lot, the extremes are:

- *All sheep given:* 6 sheep → farmer gets 6 sheep. Final: *12* - *All dogs given (4) + 2 sheep* to make up the 6: farmer gets only 2 sheep. Final: *8*

So the range is *8 to 12 sheep*, with 12 being the "friendliest" assumption (a sheep farmer receives sheep) and 8 being the worst case for the farmer.

Any mix in between — say 1 dog + 5 sheep, or 3 dogs + 3 sheep — lands somewhere in that range. The dogs are worth nothing to a sheep farmer in terms of the count, so each dog substituted for a sheep costs him one from the final tally.

Yes.

Our Chatbots Said “I Love You,” Shall We Meet? by Caroline M. Yoachim

https://clarkesworldmagazine.com/yoachim_06_24/

Here at 123aiLOVEu, we take all the guesswork out of dating. With personalized chatbots trained on social media and chat histories, we put you in conversation with hundreds of potential matches . . . and you never have to say a thing!

Even cooler is setting up transmitters and receivers at suitable spots on opposite sides of two stars.

Vegas protocol is hardwired!?

:exploding_head:

I see it now. Yes.

Exoplanet closeups?

You can use the sun as a gravitational lens: https://en.wikipedia.org/wiki/Solar_gravitational_lens

You need to be about 550 au out.

404

I'm not sure that "it is clear that Claude does not have those things".

I AM sure that it is hard to conclusively show that Claude has experience and consciousness. Even Claude isn't sure about that.

But while it is absolutely true that "it is a word calculator" - unless you hold the position that human consciousness isn't neural[1]- I don't see how this is any different from saying saying humans beings are neural activation pattern calculators.

If you're sure that your consciousness isn't neural - then fine: Claude isn't made of the right stuff so couldn't possibly be. But state your assumption up-front.

If one opens up a person and looks at their nervous system the single neurons look complicated, but not especially mysterious.

Given how shockingly little we understand the brain/mind it is hard to be sure that we are certain enough of how we work and given how little we know how LLMs work at any of the many layers above the raw architecture either position can be reasonably held, but not convincingly argued/demonstrated.

Feel free to think Claude isn't conscious - I can't prove to you it isn't. And the amount of theory we still need to learn to be able to is vast.

But don't expect me to be _certain_ that it isn't and couldn't be - you simply can't show that convincingly either.

[1] Penrose thinks we have a quantum nature - so sure no classical computer can be then. Some like Rupert Sheldrake think it's a field phenomenon - very woo maybe Claude has a morphic field as well? Lots of people are sure we have a supernatural soul/spirit. One then needs to take up Claude's status with the Creator.

Which part of what he said is wrong?

A brain is a collection of cells that transmit electrical signals and sodium. ...

That it is a collection of cells? Or that they transmit electrical signals and sodium?

Or do you feel that he's leaving out something important about how it works (like generated electrical fields or neural quantum effects)?

Dune quote:

It is said that the Duke Leto blinded himself to the perils of Arrakis, that he walked heedlessly into the pit.

*Would it not be more likely to suggest he had lived so long in the presence of extreme danger he misjudged a change in its intensity?*

Be careful of letting your deep, keen insight into the fundamental limits of a thing blind you to its consequences...

Highly competent people have been dead wrong about what is possible (and why) before:

The most famous, and perhaps the most instructive, failures of nerve have occurred in the fields of aero- and astronautics. At the beginning of the twentieth century, scientists were almost unanimous in declaring that heavier-than-air flight was impossible, and that anyone who attempted to build airplanes was a fool. The great American astronomer, Simon Newcomb, wrote a celebrated essay which concluded…

> “The demonstration that no possible combination of known substances, known forms of machinery and known forms of force, can be united in a practical machine by which man shall fly long distances through the air, seems to the writer as complete as it is possible for the demonstration of any physical fact to be.”

Oddly enough, Newcomb was sufficiently broad minded to admit that some wholly new discovery — he mentioned the neutralization of gravity — might make flight practical. One cannot, therefore, accuse him of lacking imagination; his error was in attempting to marshal the facts of aerodynamics when he did not understand that science. His failure of nerve lay in not realizing that the means of flight were already at hand.

I think we can now begin to experimentally test Conway's law and corollaries.

Agreed, a flat set of workers configured like this is probably not the best configuration.

Can you imagine what an all human team configured like this would produce?

Are you sure that guy who wakes up tomorrow after you've gone to sleep is you?

Or the one who wakes up after 10,000 sleeps?

I'm sure he's going to be quite different...

Maybe that dude (the one who woke up after you went to sleep) is another you, but slightly different. And you, you're just gone.

It might be one of the only reasonable-seeming ways to not die.

I can see the appeal.

To be clear: I'm not engaging with your main point about whether LLMs are usable in software engineering or not.

I'm specifically addressing your use of the concept of determinism.

An LLM is a set of matrix multiplies and function applications. The only potentially non-deterministic step is selecting the next token from the final output and that can be done deterministically.

By your strict use of the definition they absolutely can be deterministic.

But that is not actually interesting for the point at hand. The real point has to do with reproducibility, understand ability and tolerances.

3blue1brown has a really nice set of videos on showing how the LLM machinery fits together.