Hacker News new | past | comments | ask | show | jobs | submit | moconnor's comments login

> I believe sometimes building things is how we self-soothe.

> I have written entire applications just to avoid thinking about why I was unhappy.

I think this is true too. The prefrontal cortex is inhibitory to the amygdala/limbic system; always having a project you can think about or work on as an unconscious learned adaptation to self-calm in persistent emotionally-stressful situations is very plausible.

I wonder how many of us became very good at programming ~through this - difficult emotional circumstances driving an intense focus on endless logical reasoning problems in every spare moment for a very long time. I wonder if you can measure a degree of HPA-axis deregulation compared to the general population.

And whether it's net good or net harmful. To the extent that it distracts you from actually solving or changing a situation that is making you emotionally unhappy, probably not great. But being born into a time in which the side effects make you rich was pretty cool.


Probably very well

o3 solves this correctly and produces a great table illustrating the solution to always keep the cabbage safe.

If you enter:

A farmer has a boat that can transfer up to 500 people or animals. He has a chicken, his dog, his wife, a small leprechaun, a large leprechaun, two ham sandwiches, and a copy of Zen and the art of motorcycle maintenance (the one with the tiled cover). How can he get them all across the river?

You will get a very detailed answer that goes on for several paragraphs that totally misses the point that there is no challenge here.


o3 gets this one right:

"After taking away the two blue cubes, three cubes remain—in order from bottom to top: 1. Red 2. Red 3. Green

With three cubes, the cube in the central (second) position is red."


o3 just writes and executes a python program in the background to correctly answer this...

o3 was the only model to get this right for me:

"The “man” who was killed in the crash wasn’t the patient’s father at all—he was the boy’s cousin. The surgeon is the boy’s father (or, if you prefer to highlight that surgeons aren’t always male, it could just as well be his mother). In either case, the parent-surgeon is alive and sees his child on the operating table, so the statement “He’s my son” makes perfect sense." - https://chatgpt.com/share/680b470d-3a44-800a-9b2e-d10819168d...

gemini-2.5-pro, o4-mini and gpt 4.5 all failed and said the surgeon is the boy's mother.


I don’t see why you believe 2 would be true. I expect a strong correlation between the visual appearance of food and its caloric content etc.


Most restaurants smuggle obscene amounts of butter and sugar into dishes to make them more delicious so you'll come back. It's not unreasonable to assume that every subcomponent of a recipe might have sugar added individually.

In packaged foods, there is a whole science of masking the sugar and fat content to make it more addictive without triggering your inbuilt satiety mechanisms [1]. This is what today's engagement optimisers did for money in the 50/60s.

You could argue that these "innovations" were precisely to subvert the intuition that visual appearance of food (and other natural sensors) can be relied on to assess their nutritional properties.

[1] https://en.m.wikipedia.org/wiki/Bliss_point_(food)#:~:text=T...


> Most restaurants smuggle obscene amounts of butter

God I wish that was true. Butter is far too expensive to do that, so restaurants will use the cheapest alternative (usually soybean oil with butter flavoring) instead.


There is absolutely, unequivocally, 0 chance this can be accurate within any kind of reasonable bounds. I'm guessing you haven't done much calorie tracking if you think this could possibly be true.

I can make two dishes that look identical and have +/- 50% caloric content, easily.


> I expect a strong correlation between the visual appearance of food and its caloric content

This doesn't pass even simple scrutiny. There are so many caloric ingredients that aren't visible in food. You can't tell just by looking whether a rice dish contains half a stick of butter.


Or if it's diet coke or regular coke. Yes, it's drinks, not food, but the same concept applies.

They claim 90% accuracy, whatever that means, but I have my doubts regarding it's usefulness.


How does an app know that this piece of chicken cordon bleu is actually filled with more bacon and cheese than chicken?


Try telling a picture of diet coke from regular coke apart.


This is completely wrong. For example, you can increase the amount of oil or butter in a recipe, doubling or tripling its calorie count, and you would never be able to tell from a picture.


I imagine it just autofills the information and then you can edit it to make it more accurate

You'd have to be kind of stupid to expect it to actually be 100% accurate for all meals


The point of the app is to figure out the calories of a meal automatically by taking a photo.

Without knowing the amount of sugar, butter, oil, etc. is used in a dish, one cannot know if a dish is worth 250 kilocalories or 750 kilocalories.

If I need to manually fill in details of ingredients and amounts to get to the calories to be have an error margin of less than 100%, then the app is not useful and is at best misleading.


... So if you already know the answer you can correct it? I mean, what possible use is that?


Therefore...

4. I don't need this app.


1 tbsp of animal fat has about 900 calories.

1 tbsp of olive oil has 135 calories.

How would the app know which fat the food was cooked in?


> 1 tbsp of animal fat has about 900 calories.

This is extremely false. Please verify your sources better (and apply a skosh of critical thinking).

> 1 tbsp of olive oil has 135 calories.

This is false too, but at least it's in the right ballpark.


> This is extremely false. Please verify your sources better

Sorry! I was using Cal AI


Ok, that made me laugh.


That’s not true. They have the same amount of calories roughly. It’s physically impossible for animal fat to have that many calories. Tallow has 900 calories per 100 grams while olive oil has 884. They are almost pure fat and pure fat has 9 calories per gram.


I thought this was a joke, but he actually did do this first. Impressive!


I'm not sure whether or not he did this first, but it's very similar to an extremely impressive, but old and well-known illustration of the power of Fourier analysis in which you construct a "Fourier epicycle" (think: machine made of circular gears of different ratios) that can sketch any image. 3blue1brown has a great video on Fourier Epicycles but you can also get the idea here https://mathematica.stackexchange.com/questions/171755/how-c...


Also of potential interest is Kempe's Universailty Theroem which states you can draw any (polynomial) shape with a set of mechanical linkages. Like one that will sign your name.

https://academic.oup.com/plms/article/s1-7/1/213/1570315?log...

http://www.koutschan.de/data/link/


damn, I got nerdsniped again


Or check out drawing Homer Simpson with the same technique https://www.youtube.com/watch?v=QVuU2YCwHjw


Incredible idea and execution, very interesting results. Genuinely: what a time to be alive!


Thank you! Very much a labour of love. The next step for us is to try and build a paperclip maximiser in FLE.


Given that Factorio has logic gates and people have built various programs (including Doom, iirc) how long will it take before someone runs an LLM inside the game?


I'm reminded of Hitchhikers Guide to the Galaxy, where the whole planet is one big computer. I would expect that any model directly implemented in Factorio would take up most of the game-world.


Dual 300As overclocked to just over 500Mhz each on a BP6 with Geforce 256 here too! Fastest, smoothest machine I ever used until the M1 MacBook. Quake 3 multiplayer demo ran so fast it gave me motion sickness ^^ Years later I "upgraded" to a 1Ghz Athlon and it felt like a downgrade a lot of the time.


Consider applying for YC's Summer 2025 batch! Applications are open till May 13

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: