HNHacker News
TopNewBestAskShowJobs

lordnacho

31,598 karma · joined September 10, 2014

Low latency c++, performance tuning. Quant investment strategies, hedge funds. Bonds, options, futures, various instruments.

HFT strat consulting recently.

Lots of coding and trading.

efincp specialemailassociatedcharacter gmail.com

submissionscomments
lordnacho··on Don't be fooled–LLMs don't reason
I thought I saw a paper recently explaining that LLMs have a global workspace. Is that not like having the internal state that he's talking about?
lordnacho··on Vote on which of Hacker News' challenges for AI have been met
It wasn't one-shot, I answered a few questions to do with the structure. But all the hard word was done by Claude.

Not sure how I'd share it. I suppose I could order a copy for you.

lordnacho··on Vote on which of Hacker News' challenges for AI have been met
I guess there's a difference between a coherent book, and a good book?

I've gotten AI to generate an entire book, with the plot points kept coherent (this character knows that when this chapter starts, that kind of thing). It didn't impress my wife, but it was a book.

lordnacho··on How Singapore's government-run dating service works
I think this would be better than commercial services, which have all the wrong incentives.

I would think just a coarse sorting would do, paired with a free meal. Every couple of weeks, you go for a random date with a somewhat well paired person, and see what happens. Maybe you don't meet the love of your life, but you will meet someone roughly your age living in the same society, a perspective worth having regardless of whether you're compatible.

However, if you know anyone from Singapore, you must think this is a rather small plaster on a gaping wound.

Singapore is like a permanent examination hall. Everyone is competing to do well on exams, and then they try to get the best internships, and then the best jobs. It's not exactly conducive to starting a family. You'll lose a step in the rat race, and then your kid will need to join this rat race.

lordnacho··on Solving Factorio Quality
I played Factorio a fair bit before Space Age came out, it's awesome. I played it for ages with my kid, and it's a great tool for having him think about software decisions.

However I hit a wall with the game that has a relation to how AI has upended the programming world.

Basically, I got bored, and it might be because of the automation.

In the early game, you are running around with an axe, manually putting down inserters and assemblers, manually crafting items. You eventually make some assembly lines to automatically build stuff, and the outputs then go into more and more complex recipes for which you need to organize the flow of resources.

Along the way, you gain the ability to build trains to move goods from faraway places. Along with a number of other mechanics that are introduced, this creates a bit of novelty, because you discover some slightly different requirements and you get to exercise your brain in how to connect things together.

But then at some point, you learn how to make robots. Robots can place factory components as well as just moving goods around. You make a blueprint, plonk it down, and then the robots get all the necessary pieces, and place them down for you. This completely alters the game, because a blueprint can have a huge number of pieces while still being modular. There are people with libraries of square tiles encompassing every aspect of the game, that can simply be snapped together like legos.

I got to a point where I was assembling libraries and caring more about building a good set of components than actually wiring things together into a factory. I got an auto-placement mod, which I figured was a bit like predictive text in an IDE, it saves you from having to put down inserters and turn them, that kind of micro thing that can mess up your factory if you screw it up, but is actually tedious.

Armed with my new library of everything and my auto-placement mod, I destroyed my enjoyment of the game, somehow.

I had played the game enough times to understand what was needed, and so I got to a place where I just needed to place down a mall (a special kind of factory that mainly makes components) along with the bus (assembly lines of outputs), and things kinda couldn't go wrong. I didn't do any more of the typical early-game fiddling with making ad-hoc designs, because I knew what would work, and I didn't want to do the micro. I just satisfied myself that a few belts and rail lines would be enough to glue my libraries together, and that was that.

What does this have to do with AI? Well, I hear a lot of people's complaints with AI is that they were no longer actually coding. Without the need to type out anything, the micro is gone. People were already glueing libraries together, but at least there was some delight in writing the little bits of knitting. Now, you can just execute your architecture with a prompt, and the LLM will just do it all for you. There's still a necessity of checking what comes out, but the character of the work has changed.

lordnacho··on Does Reddit have an astroturfing problem? What the data suggests
An LLM would easily understand that, no?
lordnacho··on DraftKings is using AI to behaviorally target chronic gamblers
> Are humans not in control of what they get addicted to?

When you see an attractive woman, do you have a choice about whether you think she's attractive?

I would say that is so instinctual as to not be a choice, even though different people have different opinions about who is attractive. And yet we know that men in particular go to great lengths to be with attractive women, some to the point where it becomes an obsession.

What about food? Do you not like things that have a bit of fat, sugar, and maybe some crunch to it? There are clearly people who, despite having all the necessary information, will eat too much, to the point where they become ill from it. How is that a choice?

lordnacho··on DraftKings is using AI to behaviorally target chronic gamblers
Well I agree that 2 and 3 are settled matters, but 1 is simply hiding an unexamined definition of what choice is.

So, what does it mean to choose something with full agency?

lordnacho··on DraftKings Is Using AI to Behaviorally Target Chronic Gamblers
But the two big players in question do not do this, do they?
lordnacho··on Does Reddit have an astroturfing problem? What the data suggests
This got me wondering whether you could train an automod to act like Dang.

Dang is more or less what keeps HN from becoming Reddit.

lordnacho··on Coding is not solved
My thoughts on this:

- Coding in the small is solved. I have a current state, I want to change it, and I know how I want to change it. Eg, I have a blocking TCP handler for some reason, and I want to make it async. I can either fiddle with it or just let LLM make the changes for me.

- Coding in the larger sense is never solved. You need judgement to decide what you want made. No matter what you're building, there will be decisions to make (Who/what is it for?) and those decisions change over time. LLMs can take some default decisions for you, and if you're fine with those, you get the default (great for POCs). However you might not even realize what it decided to do for you. At some scale, you will be spending a lot of time going over those decisions. But what we have now is that the friction of changing the decisions is quite a lot lower. You can now test a lot of things that previously were very time consuming.

- The point that LLMs are probabilistic is not as important as it's made out to be. If I ask a junior dev to code up something, I also don't know what he'll make. Heck, you can be sure that you are able to solve something, yet you yourself don't know what the solution will look like. Maybe it turns out the library you were going to use isn't appropriate after all. You don't know what you will use in the end, but you do know that something will fix the issue. There can be more than one solution to a problem, and it doesn't always matter which one you find.

- I STILL think that LLMs are at their best mostly as advanced predictive text. In the sense that it's mostly good at implementing things that you've decided are needed. This can mean a heck of a lot of code, but you have to know the tradeoffs. What was decided, what were the costs of those decisions in terms of maintainability, money, time to change it, and so on.

lordnacho··on AI companies in race to demonstrate their model most threatening to humanity
You can download some weights and run your own model on your own device, and apparently only be a few months behind SOTA. While things are cheap, people will pay for the premium product, but if they ever try to crank up the prices, there's a fallback that is quite competitive.

As for training, sure, they can do that but they get distilled by the open weight guys. Plus, AI is useful even if it freezes at today's level. I can still find use for a local AI that never improves, it's good enough to do a bunch of tasks.

lordnacho··on AI companies in race to demonstrate their model most threatening to humanity
Reminds me of that Lynx deodorant ad where the guy is chased by hundreds of women.

I think you're right, they've realized there's no moat, especially against open-weight models, so they are going to need to lobby government to ban any unapproved AI services.

lordnacho··on Owed a billion dollars in Nvidia stock
The angle I haven't heard, yet I think would make the most sense, is that you both understood the agreement correctly, at the time.

Somebody did the paperwork wrong, but paperwork isn't the agreement. You agreed what you agreed, thought everything was in order, and then discovered an error in the documents.

It doesn't seem like there's a claim here.

lordnacho··on Excel now supports multiple values in a single cell
Users might see it as a good thing to have yet another feature, but this sort of type-softness is why Excel is bad. In general having every feature is a bad thing, you need constraints to design things, to have idioms.

Making everything flexible is what allows you to make spaghetti that nobody understands.

lordnacho··on 'That's so AI ' What gen Alpha's biggest insult tells us
What's ironic is that while kids have seemingly always had a nose for the inauthentic, they have just about no way to express anything of their own identity.

Even way back in the 90s we could tell that Coca Cola isn't actually a requirement for a good Christmas, and your friends who wore platform shoes were not somehow unique.

But still, when you're young you're in a prison. There are very few things you can do that aren't connected to some corporation whose main goal is just to make money.

lordnacho··on Where's the Beef?: The lab-grown-meat revolution that wasn't
Also we still don't really know what we mean by intelligence, even though we've now invented a version of it. You'd have thought that question would need some more investigation before we could build it, but here we are.
lordnacho··on Where's the Beef?: The lab-grown-meat revolution that wasn't
I'm a bit surprised that we got disembodied intelligence before we got deathless meat. If you'd told me in the year 2000 that one of these would happen, I'd have put my money on the meat.
lordnacho··on Writing Rust code that's fast by asking agents to make the code faster
Can you give it the valgrind suite to loop over? I have yet to try that with AI, but maybe the cachegrind tool is enough to help it.
lordnacho··on Writing Rust code that's fast by asking agents to make the code faster
I've been having great results with this kind of thing. I find that really just need a sensible framework within which the optimization can take place. Essentially just providing the measurement harness, and some sort of motivation for what I'm doing.

The great thing about LLM is that it seems to have the checklist for everything. If I rattle off a few things like "don't allocate on the hot path" and "remember to pin the cores" it will come up with a few items of its own that I might have forgotten.

Eventually, it will have gone through the whole list with me, while having documented all the measurements along the way.

But it's still guided by experience. If I see unusual numbers, I might say "hey did you forget to compile it in release mode?" and it will apologize and fix that. If I don't, it may just continue exploring without realising everything is wrong.

lordnacho··on Study: Young users (9 to 18Y) ditch Google for AI, with unknown consequences
Why? AI is a lot closer to the original way of doing things, which was to ask an expert.

The expert would then give you his thoughts on which modern books are most similar to S&S/P&P, and you would have to use your own thoughts to decide whether you agreed.

In many ways I find it similar to tutorials. You can go back and forth with the AI if there is something in the argument you want to investigate.

lordnacho··on What It's Like to Work in One of America's Data Centers
I'm guessing it's to do with regulation. There's probably some policy document that says a certain org needs to be able to access the physical servers within a certain amount of time, and so they need to be nearby, even though nobody ever does the accessing.

DC would be the kind of place with a lot of government contractors that would have this kind of stipulation, is my guess.

lordnacho··on Study: Young users (9 to 18Y) ditch Google for AI, with unknown consequences
AI is my default for just about any search now. From today:

How did Cantor come up with his diagonal proof? It seems to have been dreamed out of nowhere

Why are there advertisements for a shampoo containing caffeine? Does caffeine do something useful other than when you drink it?

What happens in sense and sensibility?

All of them give a decent answer, with the ability to ask follow ups. Eg, how does S&S compare with Pride and Prejudice? What about more recent novels connecting romantic choices with economic choices?

All of them give you links to source material, in case you suspect they are just making it up.

At the start of the millennium, Google search changed everything about how I look for information. Now it seems outdated.

lordnacho··on ZuckOff Is a Free App That Sees Meta Glasses Before They See You
Why do any of the device's network connections (WiFi/BT/etc) need to be on? What's to stop someone recording and uploading later?
lordnacho··on Jemalloc 5.4.0
I always wondered, is it French? "Je m'alloc du memory"
lordnacho··on Learning Programming in an Age of LLMs
I think it's still important for young people to learn coding without the LLM. they need to see the little pieces before they can build big structures. It will be like calculators, just on a bigger scale: you learn how arithmetic works, and then you rely on the calculator when you are multiplying large numbers.

My guess is it will probably take some time to incorporate LLM use into education. People who are graduating right now have a problem, being between two worlds. Those graduating in a few years might have chance to figure out what to do.

> I may have built a system that is above my own level of understanding

If I venture into an unknown area, I end up where the letter-writer ends up when he is visiting programming. Suppose I am curious about an advanced math topic, like Navier-Stokes. The LLM's answer to the news about the new advance last week is strewn with words I don't understand. Asking about anything produces another essay with more things, a loop that never closes. If it were my specialist area, I imagine I would eventually hit some point where the explanation connects to something familiar.

I think this is the wall people run into when they don't have the fundamentals. You eventually get to a point where the machine is asking you for decisions that you won't know the consequences of, and when you are trying to clarify, you end up in a massive rabbit hole. It's not that different from asking a real expert about their area, they will eventually ask you to clarify something that means something to them, but not to you.

I learned programming the slow way. I would run into phrases like "memory barrier" or "green thread" and find an article using the keywords, which led to more searches, which led to more...

There are also many false dawns. Early on, after some success writing some trading strategies, I thought I had it, in the sense that I would be able to write any program required. But it wasn't true, I would run into an iceberg from time to time. Huge areas of knowledge that I hadn't come across.

Obviously I'm not claiming I finally know everything, but LLMs have arrived at a very convenient time for me. For the things I build, there is rarely anything that I don't understand at a fundamental level. When it asks me something, it's an incidental question: what decision should we make? What are the superficial changes that are needed to fit the architecture to the desired product?

I am essentially using LLM as a very quick junior, who knows how the OS works well enough to compile things and analyze logs. These are things that would take a lot of attention in the old days because they can break on very small errors, but the direction was known from the start, and thus for me (having paid the learning cost already) it is just a matter of waiting for the AI to get the code into the desired state.

I have a somewhat usable experience. I was asked to build a trading system a few years ago, which would connect to certain exchanges and show an orderbook. This kind of thing is bread and butter, but writing it up at a new firm would still take weeks.

In recent engagements, I've simply declaratively told the LLM what properties I wanted to see in the solution, waited, and answered a few questions. Since the architecture is the same, there weren't a lot of real decisions. The time difference is immense.

lordnacho··on Swift 6.4 Released
If you have that much experience, swift is probably just a mix of what you already know? Do a couple of tutorials and then you are ready to vibe it.

It's the people who haven't done the grind of learning a new language so have the problem, since LLM will give them solutions but not experience.

lordnacho··on Global bond yields hit 2008 highs, raising stakes for big borrowers
> Maybe we should vote for people who are financially literate, can think long term, and can "conserve" rather than politicians who promise endlessly?

Who could you vote for, in any democracy, that would fit this?

Also, how many voters would have the wherewithal to identify such a person?

lordnacho··on Aligned to whom?
How can you know that verbal articulation is a weak correlate? There must be some other, canonical way to measure intelligence?
lordnacho··on Aligned to whom?
Why could you not use text to measure intelligence? Isn't that what we used to do? Have students write an essay, and then deduce that they had enough intelligence to put some thoughts together?
← PreviousPage 2 of 34Next →