HNHacker News
TopNewBestAskShowJobs

heisenzombie

1,607 karma · joined June 19, 2012

submissionscomments
heisenzombie··on New Zealand teachers shocked at leaked curriculumn – 'Where's the science?'
Thanks for this, sounds right to me.

It’s interesting to me that the ideal curriculum changes as a function of the resources available.

heisenzombie··on New Zealand teachers shocked at leaked curriculumn – 'Where's the science?'
Absolutely you need both. We agree there. And any reasonable curriculum - in practice - will be a combination. But the curriculum will still be organized in some fashion, it's a very high-level document after all. And that's where one or the other flavor will come through.
heisenzombie··on New Zealand teachers shocked at leaked curriculumn – 'Where's the science?'
That's interesting, what makes you say that? Everyone I have talked to in the New Zealand education system has seemed to me to be genuinely trying to get the best outcomes for students, given the constraints imposed by pretty severely limited resources and the challenges imposed by out-of-school circumstances.

We have a pretty decent history of being relatively early adopters of evidence-based changes to our education system in New Zealand. That's reflected in, for example, the high desirableness of New Zealand-trained educators in the prestigious international-school circuit.

What do you think the politics is pushing for?

heisenzombie··on New Zealand teachers shocked at leaked curriculumn – 'Where's the science?'
I don’t see how your comment is related to mine?

Schools have to make choices about pedagogy. I think this is a debate about which choices give the best outcomes.

You’re pointing out that access to educational resources is allocated by wealth to some degree, which is obviously true but I don’t understand your point beyond that?

heisenzombie··on New Zealand teachers shocked at leaked curriculumn – 'Where's the science?'
I am sure that there exist good criticisms of this curriculum, but I don't think these are them. As per the HN commenting guidelines, I think they're incurious and attack a weak version of what's being proposed.

It seems to me that the disagreement here is about whether to structure a curriculum around bottom-up or top-down pedagogy. Do we fundamentally structure things:

- like the tech-tree from a strategy game: I think this is how a lot of us conceive of traditional learning. "Congratulations! You've done enough calculus problems to unlock Newtonian physics!" is the extreme position.

- or like a Kanban "just-in-time" system: We start with a large theme and learn things as we need them. "Well, we've been learning about `conflict and civilization` for six months, and you've been trying to build a better trebuchet for a while but in order to for you to improve it any more I'm going to have to teach you this thing called calculus".

Obviously both approaches have strengths and weaknesses. Also obviously, any real learning is some combination of both of those. A purely "tech tree" approach would require hellish discipline, and human knowledge is way more messy than a clean DAG anyway. Also no-one would stare at a group of kindergarteners who can't count past 10 and say "lets build a large language model!"

I happen to think that the ideal system is closer to the to top-down approach, but there are big challenges to implementing it. I think it's probably much easier to do bottom-up learning if you're treating large classrooms of children like widgets on a conveyor belt. I.e it may be that bottom-up wins on efficiency and so given the current resources of our education system, a much more top-down system would just fail at the implementation stage. Whereas if our education system had way more resources, then top-down would win.

heisenzombie··on Show HN: I created Units Converter that contains 5000 units across 78 categories
Right, you can explain it but that doesn’t make it elegant. This is the equivalent of a function with a comment in it saying “Here be dragons!”.

One very desirable property of a units system is that incompatible quantities have incompatible units. So if I try to add 1 Joule (== 1 kg⋅m^2⋅s^−2) to 1 meter, I’m obviously doing something wrong and I just can’t continue.

But in SI I can take 1 Hertz (==1 s^-1) and add it to 1 radian/second (== 1 s^-1) and SI says sure, that’s 2 s^-1. But it’s not, that’s a totally unphysical thing to do and the answer is nonsense.

heisenzombie··on Self-checkout machines now ask for tips in latest squeeze on customers
My self-checkout operator is pretty slow and doesn’t pack my bags very well so I never tip him. Maybe he’ll learn one day.
heisenzombie··on Surges of cosmic radiation from space directly linked to earthquakes
The OP might be making a more subtle point, but my understanding is they absolutely can do work. We can measure the deceleration of binary neutron stars with a rate predicted by the emission of gravitational waves. Deceleration == work.

However, we can also measure things that don’t do work, for example, a static magnetic field does no work on a charged particle (it cannot change its kinetic energy) but look at the swirls in a cloud chamber and it’s clear as day if there’s a magnetic field there.

heisenzombie··on Show HN: Zsync, a Reddit Alternative with the Goal to Reward Quality Comments
I have two slightly challenging comments, which I hope you will take as constructive prodding to clarify your thinking.

Firstly, you have a bit of dissonance in your thinking that you’ll need to clear up. You say lots of complementary things about HN and suggest that the quality is what you’re aiming for. But you also say you “don’t believe in censorship”. Depending on what you mean by censorship (people have wildly different definitions!), you’ll need to reconcile the fact that HN has relatively restrictive community guidelines, and very active moderation. Many people would credit the high quality of discussion to those things.

Secondly, I think your “tags as subreddits” misses the point of subreddits. They’re not a filtering mechanism for which links you want to see, they’re a way to let users associate into small(er) communities where community norms, standards, shared experiences, running jokes, moderation expectations, and generally a community identity and spirit can develop.

heisenzombie··on Excel Labs, a Microsoft Garage Project
It's annoying for me to share the spreadsheet, but I'm basically implementing the math in:

https://lhc-div-mms.web.cern.ch/tests/MAG/FiDeL/Documentatio...

To do with calculating multipole coefficients from measured data and then doing coordinate transformations on them.

heisenzombie··on Excel Labs, a Microsoft Garage Project
There’s something unbeatable about the purity of a spreadsheet. For certain kinds of problem, it’s one of the most understandable ways to communicate a program.

But there are the obvious problems, as formulas get inscrutable and you really want some more powerful data types.

So I’ve been playing with the new Excel features - lambdas and the new kind of array formulas. And they’re kind of great! I ported some non-trivial analysis algorithms from numpy to excel and it makes for a highly shareable and havkable programming environment for non-coders.

There’s all sorts of crazy excel warts (I’m doing maths with complex numbers, and the handling of those is a true “WAT”)

It’s kind of almost-great. I can’t put my finger on it but I feel like it’s close to a really winning programming environment for certain kinds of algorithms-transforming-data programs. I think Excel probably has too much baggage to get there, but these experiments are still really interesting.

heisenzombie··on Why do recipe writers lie about how long it takes to caramelize onions? (2012)
Serious recommendation: Back in the day (like more than a decade ago) I learned to cook using rouxbe.com and it really set me up for a love of cooking. Can’t vouch for any changes since then, hopefully it’s still good.

Nonserious recommendation: Just drop >600USD on

https://modernistcuisine.com/books/modernist-cuisine/

… and end up spherising your breakfast or something, idk.

heisenzombie··on LLMs are not greater than the sum of their parts: researchers
This leverages chatGPT plugins, which are not generally available yet.
heisenzombie··on LLMs are not greater than the sum of their parts: researchers
Prompting makes a huge difference. For example, here is an interaction:

I will ask some questions, treat them as logic puzzles. Give me your working steps and then finally answer the logic puzzle.

First question: If it rains, I drive home. I did not drive home. Did it rain today?

ChatGPT: Let's break down this statement into its logical components.

The first statement, "If it rains, I drive home", is a conditional statement, often referred to as an 'if-then' statement. It can be denoted as:

If R (it rains), then D (I drive home).

The second statement, "I did not drive home", is a negation of D:

Not D (I did not drive home).

In logic, we have something called the contrapositive. The contrapositive of a statement "If P then Q" is "If not Q then not P". The contrapositive of a statement is logically equivalent to the original statement - meaning that whenever the original statement is true, the contrapositive is also true (and vice versa).

So, the contrapositive of the first statement "If R then D" is:

If not D (I did not drive home), then not R (it did not rain).

Given the second statement is "Not D" (I did not drive home), and from the contrapositive we know that "If not D then not R", we can conclude that "Not R" - it did not rain today.

So, the answer to "Did it rain today?" is No, it did not rain.

heisenzombie··on The King doesn't own all the swans in Britain
This starts out with a strong stance, asserting that this is a "myth".

But, it sounds to me like the king does actually own pretty much all the swans in Great Britain. He has every swan in England, Wales, and Scotland except: - those on the Orkney and Shetland islands - those "marked" with leg bands as being owned by the Worshipful Company of Vintners and the Worshipful Company of Dyers.

heisenzombie··on European DSA Recipients of Services Report
A phone and a tablet are just wildly different devices. Screen size —sure— but also usage patterns, input methods, workflow conventions, etc.

The “mutationes” that must be made reach deep into the functionality of an app, not just UI-level stuff. The whole purpose of the app should probably change. It’s therefore a big and challenging job to make an app scale seamlessly from one device to another. That’s irreducible complexity. The tools available will always be at best “necessary but not sufficient”.

heisenzombie··on Building a ChatGPT-enhanced Python REPL
People also can’t really recreate an earlier mental state, and have to resort to post-hoc rationalisation. E.g. https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3196841/

I feel like the answer to the question “give me a possible reason that a person/llm would give this answer given this situation” is all that’s ever really possible!

heisenzombie··on Glucomate: an app for people who record, track, and monitor their blood glucose
I love the clean design and healthkit integration, but for me it can't beat Glooko. Mainly for the integration with my T:Slim x2 insulin pump, so I can correlate what control-IQ is doing to my insulin, with my blood glucose readings from my CGM. That's super important to me to make decisions about pump settings.
heisenzombie··on ChatPDF – Chat with Any PDF
Yes, I wonder if there needs to be a level of recursion to solve for this problem:

1. User enters question 2. Semantic search for relevant sections of input material 3. Prompt LLM if it needs any further context to answer the question 4. GOTO 2 5. Finish

heisenzombie··on Immutable Linux distributions
It’s much less plug-and-play than the options in the article… but I’ve had fun with a related idea, which is running Nixos on storage where root is erased on every boot.

My home server is set up this way. At the time, I more or less followed this guide, which explains things nicely:

https://grahamc.com/blog/erase-your-darlings/

Been erasing root for almost 2 years I think now. Works great!

heisenzombie··on Apple’s Best Hope for New Headset: a Smartwatch-Like Trajectory
I’ve heard people bring up fitness, but the physical experience of sweating with a headset strapped to my face sounds unpleasant. It also turns the product into “gym equipment”. Is it waterproof? Can I give it a good wash in the sink and dry it? Or am I putting it on later with it soaked in cold sweat? How’s it going to smell after a couple of weeks?
heisenzombie··on Indirect Prompt Injection on Bing Chat
I don’t necessarily disagree with your larger point, but your example isn’t very persuasive. Travelling with an unrestrained dog in a car is pretty reckless (and maybe illegal).

Between the driver distraction, and the fact that they become a deadly projectile in a crash (and of course the fact that even a fairly minor crash could kill the dog), It’s a really good idea to have some kind of car restraint for pets in the car. Buckle him in for real!

heisenzombie··on Lost in the Stock
Highly agree with this. I was gifted an electric pressure cooker recently, and it has been a revelation for home stock making. It makes truly excellent stock in around 60-90 minutes all up.

If you're willing to wait the extra half hour for a slow pressure release, then you get a beautiful super-clear consommé-style stock because the pressurized environment means the liquid never boils. I've never achieved that in a stock pot, even if i'm being finicky about temperature control.

heisenzombie··on Critique of the mind/body problem
Perhaps like the Wheeler/Feynman one-electron universe!

Hm, except that would require anti-consciousness to exist… On second thought, this makes me uncomfortable.

heisenzombie··on Apple TV prompt requires another Apple device
Hm, some context here is that Apple Watch apps were — until recently — actually iPhone apps. The iPhone app was treated kind of like if you wanted to make a widget for your app, except it appeared on your wrist rather than your springboard. The whole app architecture really tightly coupled the products.

This has started to change with “set up for family” watches (which is mostly so you can give a kid a watch and no phone) but there are still lots of weird edge cases there.

I do personally wish they were standalone. If they were I would have gifted one to my father in law this Christmas. Instead he got a Garmin.

heisenzombie··on GPT-3 is the best journal I’ve used
VRAM, you need enough. That’s sort of a binary thing: either you have enough or you don’t. 1x4090 does not.
heisenzombie··on Common Hall-effect sensor myths (2022)
Good story, but it’s unlikely it was the (magnetic part of the) Hall effect sensor at fault here — the field due to a magnetic dipole (like a lifting magnet) drops off as an inverse cube of distance. Unless the machines were IN the scrapyard, the field would be negligible.

More likely boring old electrical grounding issues. Oh how I hate them.

heisenzombie··on Liquid democracy: two experiments on delegation in voting
That's kind of what the experiment was about. They wanted to know: If we run this simple game of chance and delegation with real people, do people delegate more or less than the optimal amount?

They found that people delegated more than would be optimal.

Now -- whether that says anything about liquid democracy applied to real world situations is quite a different question, of course. As you started to point out.

heisenzombie··on Liquid democracy: two experiments on delegation in voting
Assume we're deciding on a binary choice. Also assume that there is one correct and one wrong answer, and the only reason people choose wrong is a lack of information. Next assume each participant in the vote has a particular chance of getting the right answer.

Lets take an extreme case where you have a million people each with a 55% chance of getting the right answer, and one expert with an 80% chance of getting the right answer.

If you have a normal vote, the right answer is EXTREMELY likely to win. But if everyone delegates to the expert (after all 80% is better than 55%) then the right answer only has an 80% chance of winning. Worse outcome.

heisenzombie··on Liquid democracy: two experiments on delegation in voting
""

For example, consider one of the parametrizations we study: a group of 15 voters of which 3 are experts; the experts’ information is correct with probability 70%, while the precision of non-experts’ signals can take any value between 50 and 70%, with equal probability. Then only non-experts with signals of precision close to random should delegate: a non-expert with information that she knows is only 55% likely to be correct should not delegate to experts whom she knows to be correct with probability 70%.

""

This is not about politics, or subjective questions about complicated situations. It imagines a world in which there is one correct answer and the reason that different people vote differently is due to a lack of information about that truth.

This is very much a simplified "game" they're playing with. In this context, the result -- I think -- is more like an outcome of the central limit theorem. Basically, if you get lots of people each with a 55% chance of being correct to vote on a binary choice, they're very likely to pick the correct choice in aggregate. If you have one person who is 70% correct vote, they're 30% likely to get it wrong. So you actually don't want too many of the 55% people to delegate their decision to the 70% person, which is a bit counter-intuitive.

This is sort of interesting, and I think says something useful about the Liquid Democracy concept. However I think the experiment is simplified to the point that it's a categorically different thing than even moderately interesting political questions, and so I don't think it's an argument against applying liquid democracy to politics.

← PreviousPage 5 of 12Next →