HNHacker News
TopNewBestAskShowJobs

YeGoblynQueenne

24,943 karma · joined September 26, 2015

This a common question on this board:

What is reasoning?

In computer science and AI when we say "reasoning" we mean that we have a theory and we can derive the consequences of the theory by application of some inference procedure.

A theory is a set of facts and rules about some environment of interest: the real world, mathematics, language, etc. Facts are things we know (or assume) to be true: they can be direct observations, or implied, guesses. Rules are conditionally true and so most easily understood as implications: if we know some facts are true we can conclude that some other facts must also be true. An inference procedure is some system of rules, separate from the theory, that tells us how we can combine the rules and facts of the theory to squeeze out new facts, or new rules.

There are three types of reasoning, what we may call modes of inference: deduction, induction and abduction. Informally, deduction means that we start with a set of rules and derive new unobserved facts, implied by the rules; induction means that we start with a set of rules and some observations and derive new rules that imply the observations; and abduction means that we start with some rules and some observations and derive new unobserved facts that imply the observations.

It's easier to understand all this with examples.

One example of deductive reasoning is planning, or automated planning and scheduling, a field of classical AI research. Planning is the "model-based approach to autonomous behaviour", according to the textbook on planning by Geffner and Bonnet. An autonomous agent starts with a "model" that describes the environment in which the agent is to operate as a set of entities with discrete states, and a set of actions that the agent can take to change those states. The agent is given a goal, an instance of its model, and it must find a sequence of actions, that we call a "plan", to take the entities in the model from their current state to the state in the goal. This is usually achieved by casting the planning problem as pathfinding over a graph with a search algorithm like A*. Here, the agent's model is a theory, the search algorithm is the inference procedure, and the plan is a consequence of the theory. Deductive reasoning can be sound, as long as the facts and rules in the theory are correct: from correct premises we can deduce correct conclusions. We know of sound deductive inference rules, e.g. A*, and Resolution, used in automated theorem proving and SAT-Solving, are sound.

The classic example of inductive reasoning is inferring the colour of swans. Most swans are white (apparently) so if we have only seen white swans we have no reason to believe there are any other colours: we are forced to infer that all swans are white. We may only be disabused of our fallacy if we happen to observe a swan that is not white, e.g. a black swan. But who is to say when such a magnificent creature will grace us with its presence, outside of Tchaikovsky's ballets? Induction is thus revealed to be unsound: even given true premises we can still arrive at the wrong conclusions. Another example is the scientific method: imagine an idealised scientist, perfectly spherical, in a frictionless vacuum. She starts with a scientific theory, then goes out into the world and makes new observations about a phenomenon not described by her theory. She constructs a hypothesis to extend her theory so as to explain the new observations. The hypothesis is a set of rules, where the premises are the consequences of the rules in her initial theory. Then, being an idealised scientist, she goes looking for new observations to refute her hypothesis. Science only gives us the tools to know when we're wrong.

Abductive reasoning is the mode of inference exemplified by Sherlock Holmes. We can imagine Sherlock and Watson standing outside a tavern in London, watching as a gentleman of interest steps out of the tavern with egg on his lapel. "Ah, my dear Watson, what can we conclude from this observation?". "Why my dear Holmes, we can conclude that the man had eggs for breakfast". Holmes and Watson can arrive at this conclusion, about a fact that they have not directly observed, because they have a theory with a rule that says "if one eats eggs, one may get some on one's lapels". Working backwards from this rule, and their observation of egg on the man's lapels, they can guess that he had eggs even if they didn't directly observe him doing so. Abduction is also unsound: the man may have swapped coats with an accomplice, who was the one who had eggs for breakfast instead.

And now you know what "reasoning" means. So the next time someone asks: "what is reasoning?", you can let them know and turn the discussion to more interesting, more productive directions.

submissionscomments
YeGoblynQueenne··on How the Tobacco Industry Drove the Rise of Ultra-Processed Foods (2025)
So, I haven't made any yogurt in a while but from memory it takes about half an hour for milk to set, depending on what milk and what yogurt starter etc. No way you can make an emulsion of e.g. milk, sugar and vanilla that is stable for that long. So either you have to use stabilizers, or you have to accept that your flavourings are not going to be evenly distributed throughout your yogurt.

I can't know the latter since I can't buy that brand in Greece (for rather obvious reasons it's not imported here). If I look at the list of ingredients for e.g. the Vanilla Bean yogurt I see "Pasteurized Whole Milk, Pasteurized Cream..." and that's your stabilizer there.

How much is there? The "Nutrition Facts" label on top lists "Total fat 12g" per serving which is "1 container" of 140 grams. That's ~9% fat, so about 5% more than plain cow's milk (at 4g per 100g). The total fat is more than the fat in ewe's milk (about 6%). I ... guess the extra fat makes the milk and vanilla bean powder emulsion last a bit longer so it gets better mixed in during the fermentation time?

Dunno. The packaging says "Traces of vanilla seeds and marbling on top of the yogurt are the results of a natural infusion process". The other flavours don't have a similar disclaimer but they all have extra fat in the form of pasteurized cream.

So anyway you get both a stabilizer and uneven distribution at least for the vanilla. If you have a strong enough flavour I reckon that's not a problem though.

EDIT: (The fruit-at-the-bottom variants are presumably UPFs).

They don't have to. If you're just mixing some fruit in milk and then fermenting it to make yogurt, that's just processed (going by NOVA AFAIU). But looking at the ingredients list I see e.g. "Corn starch, Natural Flavor, Lemon Juice Concentrate, Pectin" etc. Ew. No, that's a UPF. You don't need to put that stuff in yogurt. Not if the point is to enjoy it.

YeGoblynQueenne··on Can AI design circuit boards yet?
I guess I walked right into that one.
YeGoblynQueenne··on Pushdown automata in statistical machine translation (2014)
Oh absolutely. I'm just curious because we never see articles like that on HN, so I wondered if something happened that made it newsworthy.
YeGoblynQueenne··on “Next-token predictor” is the wrong mental model for LLMs
Alright, I'll have to check up on that. Thanks for being nice about it.
YeGoblynQueenne··on How the Tobacco Industry Drove the Rise of Ultra-Processed Foods (2025)
Ewe's milk yogurt is more caloric than plain cow's milk yogurt (because of higher fat content). Does that make ewe's milk yogurt "worse for you than UPF yogurt"?
YeGoblynQueenne··on How the Tobacco Industry Drove the Rise of Ultra-Processed Foods (2025)
I've never had that. Is the honey on the bottom?

EDIT:

>> Super clean. No weird chemicals. Worse for you than UPF yogurt.

Ingredients list says it's got 21% sugars. Why is that "worse for you than UPF yogurt"?

I think you have a weird headcanon about what's good and bad for you and that has nothing to do with UPFs. 21g of sugar (of which 11 is honey going by the ingredients list always) is a US teaspoonful and a half. You 're not gonna die younger if that's your daily sugar intake.

YeGoblynQueenne··on How the Tobacco Industry Drove the Rise of Ultra-Processed Foods (2025)
No you couldn't. The reason is you can't just mix stuff like sweeteners into yogurt and ship it or you end up with a gloopy goop as I say above.

The thing that people don't get about yogurt is that it's a gel, and one with a very delicate structure to boot. The moment you so much as stick a spoon into yogurt you disturb the gel matrix and that causes whey to seep out, an effect called "syneresis". If you've ever tried the kind of yogurt I say above that I'm having for breakfast, you'd know that if you eat say half a tub and then leave the rest for a while and come back to it later you find it has turned into a watery mush that's not very appetising at all. A lot of good yogurt goes to waste like that because folks who want to maintain their waistline will eat some and leave the rest "for later", and it's a balancing act for producers to make yogurt tubs small enough that people will want to eat them in one go and large enough that it makes sense to sell them [1].

One traditional way to deal with syneresis is straining which produces what everyone else calls "Greek yogurt" and we, Greeks, call "strained yogurt" or "bag yogurt" (because traditionally it was strained through a linen bag). However, even strained yogurt suffers syneresis as long as it retains any moisture and as a result you still can't ship a tub of "Greek yogurt" with say fruit, or honey, mixed in just like that [2]. That's why you never get delicious FAGE yogurt with fruit already mixed in [3].

You also can't make yogurt with stuff in it from the get go, because yogurt starts as milk, therefore a liquid, and whatever you want to mix in it will sink to the bottom or float to the top before the milk has a chance to set. Some yogurts with fruit jams for example have the jams on the bottom because that way you just pour the milk on top and let it ferment and then you get jam with yogurt on top and you can scoop both in one spoonful. I've also seen home-made yogurts with fruit mixed in that stay at the top. But you can't just mix the jam or the fruit in the milk and expect it will stay mixed in when the milk sets.

In order to sell yogurt with admixtures without it turning into a goopy gloop you have to stabilise its structure, so that's where the admixtures come in that cause sweetened yogurts to become UPFs. In other words, you need stabilizers. Then you probably also need emulsifiers to "correct" the texture. I don't eat that kind of stuff but I'm gonna bet that the sweeteners also end up being industrial-style ones like high fructose corn syrup or maltodextrin and so on, just for reasons of economies of scale and whatnot. So you end up with something that's many times removed from fresh yogurt that you can make in your own kitchen with milk, a thermometer and a bit of yogurt [4] and that counts as a "processed food" in NOVA terms.

So, no, you can't just ship sweetened yogurt without it turning into a UPF. Or, perhaps you can if you're rally crafty about it, but it's not going to be easy, nor cheap, it's going to turn into a chemical monstrosity anyway, and nobody's gonna bother.

Btw none of the above pertains to Skyr or similar products that I don't know anything about. I'm talking about bovine milk yogurt in the style traditionally made in the Eastern med and the Balkans.

_______________________

[1] There's a bit of an exception in ewe's milk yogurt which is much firmer and suffers syneresis more gradually because of the higher protein content of sheep milk. Still, the largest ewe's milk yogurt I used to see was 250g and I don't see them anymore at all though possibly there's other reasons for that. I have never seen ewe's milk yogurt outside of Greece but I believe it's very common throughout the Balkans.

[2] You can strain all the whey out of yogurt. But then you end up with cream cheese, not yogurt.

[3] Btw the scene in Silicon Valley is probably the result of US teaspoons being bigger than European teaspoons. I can totally stick a teaspoon in the little jam tub in Greece.

[4] It takes yogurt to make yogurt. I know. I was endlessly confused about that as a kid.

YeGoblynQueenne··on How the Tobacco Industry Drove the Rise of Ultra-Processed Foods (2025)
The majority of people in which part of the world? Certainly not e.g. in the Eastern Med and the Balkans, I can assure you of that.
YeGoblynQueenne··on How the Tobacco Industry Drove the Rise of Ultra-Processed Foods (2025)
>> They produce lactic acid, which is a stabilizer.

Is it? The role of lactic acid in yogurt is to hydrolyze milk proteins and turn them into solids, so you can eat it with a spoon.

YeGoblynQueenne··on How the Tobacco Industry Drove the Rise of Ultra-Processed Foods (2025)
Nobody seems to have noticed the "product" next to "yogurt":

>> * Almost all yogurt products, flavored or otherwise

If we weren't on HN I'd say that's a subtle engagement hack calculated to generate maximum confusion and irate debate in the comments below. But we're on HN and good faith etc.

So, yes, "yogurt products" are all going to be ultra-processed not least because they will have to destroy the gel matrix of yogurt to incorporate whatever other ingredients they must like fruit juice or sugar, stabilisers, emulsifiers etc. Actually I guess they must include the stabilisers and emulsifiers because they mix in flavours and therefore destroy the yogurt gel matrix, which would otherwise result in an unpleasant gloopy mess.

Speaking of course in all this from the privilege of a summer in Greece where I've start every morning since June with a tub of 150g of fresh (not strained) goat's milk yogurt with the skin on and with a (European) teaspoonfull of honey.

Not healthy, you understand. Not with all the saturated fats and sugars. But, boy, is it palatable.

YeGoblynQueenne··on “Next-token predictor” is the wrong mental model for LLMs
Eh, no, that's not right. I might need to brush up on my Sutton & Barto but the RL task is traditionally defined as, informally, "given a current state observation predict the next action, state and reward". A policy is always predicting the next timestep's reward. Otherwise, how would it know what to do next?
YeGoblynQueenne··on “Next-token predictor” is the wrong mental model for LLMs
>> EDIT: gentler phrasing

Thank you.

YeGoblynQueenne··on “Next-token predictor” is the wrong mental model for LLMs
>> 'pattern matching' is a better intuition that 'reasoning' even though I think nominally, using the term 'reasoning' is perfectly fine in that context. It's just a loaded word that brings too much to the table.

Can I be a little pedantic? It's "pattern recognition" not "pattern matching". "Pattern matching" is what we do when we write a regex to recognise a sub-string in a bit of text. "Pattern recognition" is a subfield of AI that studies how to find patterns in data. For example the outcome of "pattern recognition" could conceivably be a regex that fits a large number of substrings in some corpus of text (essentially a regular grammar used to compress the text maybe). A regex is already a pattern, usually formed by a human (or an LLM these days) while pattern recognition starts without any patterns and builds them up from looking at the data.

It's harder to pin down "pattern recognition" because it's an ancient term that was once its own field, before it got absorbed into modern machine learning, and because there are countless approaches to it, quite unlike "pattern matching" which is basically just regexes plus a couple of other rarer things (like unification).

YeGoblynQueenne··on “Next-token predictor” is the wrong mental model for LLMs
Sorry, I'm not convinced. "make_more_likely" is always maximising the probability of the next token given a sequence of preceding tokens. That's what we mean when we say "next token prediction", that the model returns the token that's most likely to follow the current sequence of tokens. The mechanism used to do that doesn't matter, it's still predicting the next token whether that's because it maximises a reward or because it follows a gradient or whatever else one might think.

Btw it's "make_most_likely" not "make_more_likely". It's just that "most" gets "mosted more" with every pass through the training data.

Edit: the article author's argument is that RLVR is different because it's learning to predict the next token by generating its own token sequences. That makes no difference: what is learned is still the conditional distribution P(token_k|tokens(1,...,k-1). That's the prediction task. Doesn't make a difference where you learned it from or how.

YeGoblynQueenne··on Pushdown automata in statistical machine translation (2014)
Open access pdf from ACL anthology here:

https://aclanthology.org/J14-3008/

Btw, this is an interesting article to me. What was the motivation for posting on HN ... in 2026? Is there some background/ backstory to it?

YeGoblynQueenne··on Francis Fukuyama: The American Right Has Gone Off the Deep End
Huh. That's funny. I have no log in yet I read the article entirely free. I clicked on a link on a post in twitter but then I copied the URL directly from the article. I can't see a difference in the two URLs though but there's probably a redirect. Can't check easily now.
YeGoblynQueenne··on Can AI design circuit boards yet?
Ahem. For you or for the AI?
YeGoblynQueenne··on Mom gets 6-month suspended sentence for letting 5-year-old walk to the pond
Why Japan? How about the UK, or the US? Aren't those a bit more relevant? Those are certainly not racially homogeneous societies but why is it that racial differences are also cultural differences? Like, don't all Americans, black and white, love their guns for example? Or love baseball?
YeGoblynQueenne··on Formalizing Fermat's Last Theorem
Well, if I were a mathematician, what I'd be noodling about with right now is a way to model the number of failed attempts that AI companies must be making for every success they report.

I guess you don't have to be a mathematician to do that sort of calculation, but I'm just proposing it as a way to lift mathematicians' spirits a bit.

Also pay attention to the fact that every time a new model is released there's a slew of new results and then they dry out for a while, which suggests a "throw stuff at the wall and keep what sticks" approach that's incompatible with a kind of system that can just magickally solve all maths right now.

I'm saying that because I get the feeling that mathematicians don't have a good model for the true capabilities of those systems and that can lead to an overreaction, like "woe is me, all of mathematics will be solved and my entire discipline will be rendered obsolete". Coming from an AI background I don't think that's right. I think because mathematicians are not AI researchers they simply don't have a very clear idea of what's going on with those systems. And tbf even many AI researchers (the ones who don't enjoy the benefits of a long tradition that goes back to the 1950's and basically only joined the field in the last 10 years or so) don't understand those systems very well either.

Bottom line: don't panic.

Or, not yet :0)

YeGoblynQueenne··on Formalizing Fermat's Last Theorem
It doesn't seem execution was a problem. From Kevin Buzzard's blog:

I’ve compiled the code base and run comparator on it — it checks out. It is a gigantic proof (over 13.4 million lines of code) and takes nearly 20 times as long to compile as Lean’s mathematics library (on a machine with 96 cores!). Lean can be sluggish when jumping from file to file on a repo of this size (even on a machine with 500G of ram, which Anthropic also gave me access to), but Anthropic also supplied me with some html documents which are easier in practice to explore (clone the repo and open with a web browser).

500G of RAM is not actually that huge tbh (I was looking to buy a used 1TB server blade for some personal stuff a while ago but it was too much hassle) so I don't guess there was too much potential for buffer overflows.

YeGoblynQueenne··on Formalizing Fermat's Last Theorem
No expertise at all on interactive theorem provers like Lean but I am familiar with Resolution-based automated theorem proving. In that setting, one writes down a theory, in the form of a set of first-order definite program clauses, and then presents a statement to the prover, then the prover proceeds to prove the statement is a theorem derived from the theory.

Is that (other than the language not being definite logic) more or less what Lean does also? In that case, isn't all the work in writing down the theory, and isn't that the step where mistakes can creep in?

Is that more or less what you're pointing out? That FLT is simple enough to state but the theory from which it is to be derived can be mangled and so accept FLT on the wrong grounds?

YeGoblynQueenne··on Formalizing Fermat's Last Theorem
Couldn't the kernels have different bugs?
YeGoblynQueenne··on Mom gets 6-month suspended sentence for letting 5-year-old walk to the pond
Wait. Is it racial or is it cultural? There's a difference there.

Do you believe it's possible to have cultural homogeneity without having racial homogeneity?

Btw I think you're wrong: people can be extremely xenophobic with the folks from the next village over who are their cousins thrice removed. It's nothing to do with race or culture, just who's in the in-group and who's in the out-group, and what those are depends on a myriad of factors that "race" and "culture" are only convenient, but inaccurate, proxies for.

Another example: Greek football fans from Athens will literally murder their counterparts from Thessaloniki, and they're all just as Greek as each other [1]. You just have to draw a line in the sand and people on either side of it will start throwing stones at each other.

__________________

[1] Culturally of course. We're not a "race". Tiny little rant but with Nolan's Odyssey, there were the usual complains about Helen being played by a black woman etc. and people all over the internet were having fights about what real Greeks look like. Oh they have fair hair and light skin. No they have dark hair and olive skin. And so on. The truth is that most Greek families are like mine: one parent is fair-haired and light skinned (my dad: blond hair, green eyes, goes red in the sun; his mom: blond hair, blue eyes, skin like yogurt) and the other is dark-haired and dark skinned (my mom: brown hair, brown eyes, goes brown in the sun; her mom: black hair, black eyes, never sat in the sun). But stereotypes persist so me and one of my friends were asked how come we're both fair-haired with light-coloured eyes when true Greeks are brown and at the same time a friend recommended to me a concealer that was three tones darker than my actual colour even though I was standing right there in front of her eyes with my pasty-white skin. Race is not what it's made out to be. End rant.

YeGoblynQueenne··on Mom gets 6-month suspended sentence for letting 5-year-old walk to the pond
Well, everybody rode their horse everywhere in the Wild West.
YeGoblynQueenne··on Mom gets 6-month suspended sentence for letting 5-year-old walk to the pond
As an international reader I'm missing some context here. Isn't a gated community supposed to be singularly safe? With private guards (like the guy who took the kid to his mom) patrolling and no strangers allowed in etc? Why did the guard tell the kid that "bad people" might take him? Where would those bad people come from? It really all does sound very weird.
YeGoblynQueenne··on Mom gets 6-month suspended sentence for letting 5-year-old walk to the pond
>> It's funny in the morning. One child walks to another child's home then they walk to another child's home and so on until you have a big group of kids walking to school together. In Austria they call it something like a children train.

In Japan they call it a Katamari.

YeGoblynQueenne··on Holden's Lightning Flight
>> Regarding speed jets are obviously different level fast from propeller engine planes, landscape runs below you way much faster than you are used to. Higher speed means less time for decision and reaction, no time for "wait, I need to think something over", because times flies and jet flies with it, fast. Nobody waits for you and your pondering there. Its totally different workload.

I am guessing this accounts for the fact that he couldn't turn the afterburners off until he had lifted off: he didn't have the time to fiddle enough with the controls.

>> During final seconds you learn to feel through your butt the rate of descent - you coordinate butt feeling of descending (less speed means less lift, thus more descending) with hand pulling the stick to compensate decreasing speed with increasing the angle of attack (thus increasing lift), you smoothly balance this dynamic imbalance, you slowly decrease speed, increase lift and slowly decrease the height above the concrete in order to land smoothly from desired height.

That must be what people mean when they talk about "flying by the seat of your pants" XD

YeGoblynQueenne··on Holden's Lightning Flight
I think you're reading too much into the quote. RAF simply acknowledged that the whole episode was an accident that happened during routine work and they wouldn't punish their man for an accident, especially since there was not much harm done (except the repairable damage on the plane).

If he had taken the plane for a joyflight then he would have gotten a right bollocking (to put it mildly) but if it was an accident what exactly could they hold against him?

If there's something peculiarly British about it, it's that their airmen are well-trained professionals and they do not mete out unnecessary punishments just to show who's boss.

YeGoblynQueenne··on Silicon Valley is in denial in face of widespread backlash
How do we know this?
YeGoblynQueenne··on Humanity has the debate about AI consciousness backwards
One is a computer and the other, a stone.
← PreviousPage 6 of 34Next →