Hacker Newsnew | past | comments | ask | show | jobs | submit | tobiasSoftware's commentslogin

"I find it useful to think “what can I expect from speaking with the dataset of combined writing of people”, rather than treating a basic LLM as a mind."

I've been doing this as well, mentally I think of LLMs as the librarians of the internet.


They're bad librarians. They're not bad, they do a bad job of being librarians, which is a good thing! They can't quite tell you the exact quote, but they do recall the gist, they're not sure it was Gandhi who said that thing but they think he did, it might be in this post or perhaps one of these. They'll point you to the right section of the library to find what you're after, but make sure you verify it!


They are librarians, just that it happens to be the library of Babel.


Book golems


One of the weirdest things Einstein discovered is that time is relative, but cause and effect are absolute.

For example, muons should decay before they hit the ground, but they don't due to time dilation. We see the time dilation when observing the muons, but the muons don't, so you would think that for us, the muons make it to the ground but for the muon it would decay too fast. However, the muons experience length contraction, so they do make it to the ground from their viewpoint as well.

So cause and effect is preserved, even though we would disagree with the muon on the relativistic reason why it is preserved.


Time is relative, sure, but “proper time” isn’t relative.


> For example, muons should decay before they hit the ground, but they don't due to time dilation.

Would you mind expanding on this?



My favorite example is if I search Google for "tide me over vs tie me over" it comes up with "Tie me over is correct". Not only is this wrong, but if you click the link, the source itself says it is wrong! The source is literally on the importance of fact checking, and Google is pulling a quote that the article uses as an example of an incorrect fact.

Google: WARNING: It is a common misconception that the phrase “tie me over” is actually pronounced “tide me over.” Some even go so far as to say the “tide” refers to the ebb and flow of hunger, but this is not the case. Rest assured “tie me over” is correct.

Actual source: https://www.techtarget.com/whatis/feature/Tie-me-over-vs-tid...


Interestingly enough, I searched for the same query myself (signed out) and it came up with

> “Tied over” is a misspelling of “tide over.” Tied means to attach, fasten, or bind something, while a tide is the rising and falling of the sea that takes place twice a day in relation to the pull of the moon's gravity.

So, it did come up with a correct answer... but the answer is to a completely different question. Though the "actual source" you used was the second result.


Honestly I think the root problem is that universities have a degree in computer science, whereas what most people want is to learn to build computer software.

The two overlap most of the time in subtle ways where the science gives an important foundation, such as learning Big O notation and low level memory concepts where exposure helps. I've personally seen this with a smart coworker who didn't go through university and is great at programming but I'll catch him on certain topics such as when he didn't know what sets and maps were and when he tries to sleep a second instead of properly wait on an event.

However, the differences between computer science and building software are problematic. Watching my wife go through university, she's had to struggle with insanely hard tasks that will not help her at all with software, such as learning Assembly and building circuits. The latest example is the class where she's learning functional programming is not actually teaching it to her. Instead, they combined it with how to build a programming language, and so instead of giving her toy problems to teach the language she is having to take complex code she doesn't understand well that generates an entirely different programming language and do things like change the associativity of the generated language. In the end, she feels like she's learned nothing in that class, despite it being her first experience with functional programming.

On the flip side are the things that are necessary for software that aren't taught in university, like QA. For me personally, back when I was in university a decade ago I never learned about version control and thought it was just for back up. Similarly, I never learned databases or web, as the required classes were instead focused on low level concepts as Assembly and hardware. My wife is at least learning these things, but even then they often seem taught badly. For example, when they tried to teach her QA, instead of hardcoded unit tests, they made her give random inputs and check to make sure the output was correct. Of course, checking the output can only be done by rewriting all of your code in the testing files, and if there's a bug in your code it'll just get copied, so that kind of defeats the purpose. Even when the assignments are relevant there is often no teaching on them. For example, her first ever web code was a project where they told her to hook up 6 different technologies that they had not gone over in class, with only the line "I hope you've learned some of these technologies already".


Perhaps you should consider wearing an N95 while visiting the hospital. You can a pack on Amazon for $15 (even in stylish black), and they really aren't that bad wearing. In my opinion, the practice of wearing N95s in medical settings is the one response to Covid that should be permanent. Unfortunately even in peak Covid medical advisors were too scared to advise proper masking and settled for cloth and surgical masks, while KN95/N95 offers far more protection for yourself. Looking back, we could have done away with social distancing, quarantining, shutdowns, and all the other extreme procedures if we had just ramped up N95 production and told everyone to wear them in public while we were waiting on the vaccines.


" I don't think there's a JRPG more accessible than this one"

I'd argue there's one more, though I may be biased as it's what got me into RPGs - Final Fantasy Mystic Quest. It's a relatively unknown title as it was an attempt by the Final Fantasy franchise to get beginners into RPGs. It's like Super Mario RPG in a lot of ways, including seeing monsters on the overworld. However, it's much more like table-top RPGs, without mechanics such as timed hits and with more standard RPG monsters and spells. Personally I love the art style they went with in the game, it's the peak of pixel art IMO.


Yeah, my friend and I got MQ when it first came out and beat it in a single day. IIRC, the resurrection item was cheaply available, instant-death to undead enemies, and healed living characters up to full health.

There were some fun box-moving/jumping puzzles though.


There were no resurrection items in FFMQ. There is a Life spell for use during battles, and you automatically resurrect with 1HP after a battle.


Did the Life spell heal you to full if you're alive? If so that must be what I'm thinking of


It's generally a fun game and fairly streamlined, but there are some trash fights (mainly in the fire region) where it's really easy to get stun-locked to death. Or get a game-over from having both party members get petrified by the monsters in between your turns.

With SMRPG, you can at least still block while asleep/mushroomed/scarecrowed.


Mystic Quest was my first JRPG...then FF4, Chrono Trigger, and SMRPG - I think in that order?

It really kicked off a lifelong love of the genre


Mystic quest peak moment: a fight with Naga/medusa with low HP. The medusa sprite would change from the regular giant femme fatale image, to an old lady, distraught about losing her hairpiece.

Odd game for sure

Speaking of Peak pixel Art : Paladins quest. Thats Mystic quest on crack


FF: Mystic Quest was also my first JRPG. After playing it I had to try all the others.


I agree that "throw everything in a hashmap" should be straightforward and is a good interview test. However, his further steps to "optimize" it by saying "Poor candidates load the contents of both files into memory." are terrible. Yes, that might optimize resources, but first it hardcodes the requirement that there are exactly two days breaking the solution if the requirement changes, and second it adds a bunch of finicky fragile code about "if there are two pages or more from day one or if the first page from day one is different from the page from day two".

Great candidates treat software like a business with changing requirements and code that is read by multiple people, poor candidates treat software like a math challenge where the only goal is to use as few resources as possible.


I agree. I hate how these interviews always focus so much on algorithmic time at the expense of flexibility of the code. I agree with the first part about avoiding the n^2 algorithm by using hash maps. However, making your algorithm use half the memory but hardcoding the requirement that a user visited on two days is bad design, especially when it's only halving the memory used. Also not only does it hardcode the requirements, it also makes it much more complex logic wise as you need that "if pages from day 1 >= 2 or first page from day 1 != page from day 2".

My design was to create two hash maps, one for customer to a list of days and one for customer to list of pages, though after reading the article I realized my lists should really be sets. Then you can easily account for any change to the definition of a loyal customer, as all you need to do is use two O(1) lookups and then check the size of the lists. Easy, flexible, and little room for error.


> I agree. I hate how these interviews always focus so much on algorithmic time at the expense of flexibility of the code.

Especially when the question scenario is generating "business metrics" which tend to see a lot of tweaking and iteration.

Having engineers who make contextually appropriate designs and architectures is at least as important as having engineers who are math-whizzes.


The SMTM theory is that obesity is not directly about counting calories, but more like a contagious disease due to a contaminant that causes people to be hungrier and eat more calories. Many people will dismiss this outright, but consider these things:

1. 40% of US adults are obese, which is insanely high for a willpower issue (gambling addiction is 1-2% for example)

2. The vast majority of weight loss attempts fail miserably long term, with success rates somewhere between 5-20%

3. There is precedent for this type of idea with stomach ulcers. We thought they were a psychological cause but the main cause turned out to be H Pylori bacteria

The problem is that even if they are right, it is very difficult to detect a difference between directly eating less calories and not eating a contaminant that makes you hungry so you indirectly eat less calories.


Because the story was misrepresented and it wasn't truly voted on in the normal sense.

In the US Congress, how many people are present matters to the vote. So if a bunch of Congressmen take off, it would give enormous power to the few people who remain. Because of this, there's a gentlemen's agreement that on those sorts of days, the people who are there just clock in and don't actually vote. My understanding is that the daylight savings "vote" was this type of situation where someone violated the gentlemen's agreement and voted on it without a planned vote. It doesn't really affect anything though because it would need to pass both sides of Congress and this was just one of them.

As far as why anyone be against it, I'm not sure, but it wasn't "almost" passed as the media made it sound like.


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

Search: