HNHacker News
TopNewBestAskShowJobs

chaxor

1,725 karma · joined August 29, 2020

submissionscomments
chaxor··on Meta-analysis: On average undergraduate students intelligence is merely average
A portion of that argument is probably rooted in a very good intention: the Gini coefficient is far too high right now. One good way go reduce the Gini coefficient is to reduce the praise for roles like CEO, manager, physician, etc and reclaim more respect the technicians of the world.

University doesn't really connect anyone with "remarkable" people, but if you come from a rural or low populated area, it does help to let you interact with a more diverse (many professions and ideas) set of people.

You're correct that universities don't really have high IQ people (even the grad students, PIs, and professors likely do not have a high IQ these days), but the incentive structure isn't there anymore to allow for this. The publish-or-perish paradigm has basically annihilated the incentive structure for intelligence and now the focus is much more on 'doing' (without much thought). It's pretty simple to see that when looking at the choices in grad students and faculty at places like Harvard, MIT, Stanford, etc. They're not chosen for drive for novelty, or their cleverness, they're chosen for tirelessly working towards very incremental adjustments to ideas that are then marketed as highly impactful and novel - but the biggest determining factor here is being a "hard worker", not intelligence.

chaxor··on Cleaning up my 200GB iCloud with some JavaScript
"640KB is more than anyone will ever need"

It's a little absurd to think people don't need more than 2TB - especially on HN. Gamers will likely have 2TB in games alone, videographers often have many TBs of videos and photos from weddings and events in their life, many that care about health may have a few TB in genomic data mirrored on their computers to analyze, etc.

I would imagine it's hard to find people that wouldn't have TBs of data, if they were allowed to do so. The reason many people don't have TBs of data is they're limited by these exact companies you're claiming 'solve the problem' by offering limited storage.

It is notable however, that having better tools to organize, deduplicate, and compress data would be helpful to reduce some of the size of data that many people have. Over the years I've noticed my family will have multiple tar.gz archives, zip archives, etc, which (after extraction/unencryption) will share 20% files here, 10% files there, a 4kb jpg that's the same as a 100MB PNG here and there, etc. So yes, those 10TB archives may end up being 5TB if someone spent the time to really comb over, understand, make good decisions, and organize that data. But I have not yet seen anything that can scratch that surface yet, other than perhaps https://github.com/jjuliano/aifiles - but I won't use it until it's local only and has guarantees not to destroy data without explicit permission. An overlay filesystem that shows compression/deduplication with LLM capability like aifiles is probably the best option here.

However, I wouldn't imagine that most people's life data is less than 2TB even with all of this - it's mostly imposed as an artificial constraint by these companies.

chaxor··on Go: What we got right, what we got wrong
It's ok that Go didn't work out for scientific crowds, since Julia works better for the scientific community as a replacement for hacky Matlab/c++/Fortran conbled-together scripts.
chaxor··on Thinking About Internet History
Are there any developments towards helping these efforts out by decentralizing / backing up parts of the archive and way back machine data? I know torrents are hugely important in helping to decentralize and maintain many important files in academia and such (e.g. NN weights and CERN data) but I think cephfs is also trying to allow decentralized data storage with redundancy.

It seems like there is some solution which can provide a huge data source to be decentralized over arbitrary number of nodes, where each node can hold or back up just some part of the data, and allow for a dashboard view that shows the level of redundancy over all of the data for each of its parts.

chaxor··on Latent Dictionary: 3D map of Oxford3000+search words via DistilBert embeddings
Typically these types of single word embedding visualizations work much better with non contextualized models such as the more traditional gensim or w2v approaches, as contextual encoder-based embedding models like BERT don't 'bake in' as much to the token (word) itself, and rather rely on its context to define it. Also, often PCA for contextual models like BERT end up with $PC_0$ aligned with the length of the document.
chaxor··on Why are Apple Silicon VMs so different?
It would be silly to install Colima for this though.

If the argument is that Colima --calls--> Lima --calls--> {a ton of different things including kubernetes and docker and ...} --calls--> a QEMU command somewhere deep in the code, then the only thing that is required here is QEMU. Not kubernetes or any other junk on top that just adds complexity and potential insecurity.

One QEMU command should be all that's required.

chaxor··on Why are Apple Silicon VMs so different?
I'll check this out. There are many different systems out there like UTM and such, but I want the most basic / minimal amount of dependencies, which will work basically anywhere - which is just QEMU. Not UTM, or maybe parallels, sometimes Lima, for Mac and then virtualbox for windows, and QEMU Linux type of nonsense. Just QEMU should suffice everywhere, and it's much more secure that way.
chaxor··on Why are Apple Silicon VMs so different?
Man is this the case.

I have been trying to figure out how to have a single command to make a Qemu VM on an M2 Apple silicon chip for like a year without much luck.

All I want is to run something like Alpine Linux + Sway WM on Qemu while on macOS or AsahiLinux with one command on cli.

On x86-64 its fairly simple :(

chaxor··on Every major pharmacy chain giving government warrantless medical record access
"screaming about government power but not caring about corporations and billionaires" - that's by design. Anyone falls for this (worries about government power over corporations) is simply not intelligent enough to see past the corporate media's shaping of the narrative. The US needs a massive shift of power towards the government as well as a decrease in the democratic deficit. While there are plenty of problems with unions and the government, overall the state of the US needs a change, and smashing unions was one of the worst contributors towards increasing the democratic deficit, and it would greatly help to increase the functioning of local democratic organizations to improve the national democratic deficit.
chaxor··on Amazon receives FDA warning letter for supplements with undeclared ingredients
Listing what is in a product isn't necessarily as easy as one might think.

Certainly if an elemental analysis is required to ensure that the heavy metal content is lower than 1 ppb, it means something like a comprehensive SEM-EDX is required for every single batch. This should be required to avoid things like not noticing the enormous addition of lead chromate into turmeric, but it does require caring a bit about quality, and caring about quality in this way won't occur in a free market economy without some checks.

The examples of this are everywhere, even in things as basic as water. If you ever want to worry yourself, learn a bit about characterization techniques for determining the amount of heavy metals or organic molecules ("forever chemicals") are in water to ppt levels. It's astounding how much different techniques will provide wildly different answers, and it's unclear which method to believe.

When you look closely enough, it can get pretty scary how little we know about what is in our products.

chaxor··on Scholars asked Harvard President Claudine Gay for her data. She said no.
This tends to be more of the norm for the ones at 'good' institutions, which is aligned with the common knowledge / heavily replicated studies that 'top performers (best students, etc) are the most likely to cheat.

There can still be good academic works (and the alternative in industry has even worse incentives, such as intentionally hiding the ability to replicate) but unfortunately the structure as of now rewards bad behavior.

chaxor··on Lab Notebooks (2020)
Quarto is fantastic for making presentations as well!

If you're a developer and want to have your presentations amenable to being tracked in git, with all of the figures made from code and so on, Quarto is the absolute best you can do.

It's phenomenal and every developer should be using it.

chaxor··on CPNG, a backwards compatible fork of PNG
Maybe this is the genius in it. In order to get everyone to move to better formats, you just break everything in the one everyone uses, so they have to move.

Like Twitter becoming X to push everyone to mastodon. Mastodon is better in every way, so it's a net win.

chaxor··on If buying isn't owning, piracy isn't stealing
I'm confused. What happened to company 12 initially mentioned and content A.4?
chaxor··on Towards accurate differential diagnosis with large language models
They suck the *insurers* dry?

Insurance is one of the major problems in the US, and they're a parasite on all of the US society, so this line of thinking is completely backwards.

However, the idea that physicians should make less is reasonable, in order to get people who are more interested in the real medicine, rather than money seeking people that is often found now. Getting rid of insurance and the structures that involve it essentially solves this issue.

Unfortunately, it's unlikely to occur, so sidestepping everything using capitalism is the only way forward I can see. One good example is making diagnostic tools at the house more common place. It's pretty typical to be able to purchase an entire machine for the price you will be charged at a physician's office.

For instance, a good 12 lead ECG may cost about $1300 to get for your house, but it will also cost that much to get one test done at the physicians office. So it's clear that buying one for the house is the way to go, as you can do it everyday for years for the price of one time through 'the system'. It could also be much cheaper if targeting every household through economies of scale.

This is true for essentially all diagnostic tools. So it's about time people just expand their first aid kits to involve some of these simple diagnostic tools, like urinalysis kits, etc.

The interpretation is of course valuable, as it takes a little while to understand the output of different tools, but much of this is very capable to be automated and sold with the tool or plugged into a foss solution.

Perhaps it's not ideal to some, but it's the easiest to occur in capitalistic settings and where insurance keeps costs at institutions so high.

chaxor··on Marker: Convert PDF to Markdown quickly with high accuracy
Yeah, totally. PDFs are wonderful for archiving.*

They can hold so many different types of data so that they're extremely difficult to parse.

Because of this, you can put several malicious programs into them for RCE.

That way, if someone archives many PDFs, there can be a plethora of different RCE vulnerabilities just waiting for the user to discover.

It's a wonderful dream for any malicious actor.

* /s

chaxor··on Yabai – A tiling window manager for macOS
Yes, absolutely.

Asahilinux >> MacOS

It's wonderful to use the M2 with incredible battery life and performance without having to have a subpar OS.

chaxor··on Yabai – A tiling window manager for macOS
I tried amethyst but it was essentially useless to emulate an awesomewm or sway like experience.

Switching to asahilinux was a *far* better solution. Not really sure what the use of macOS for me is when asahi exists now.

chaxor··on Nvidia H200 Tensor Core GPU
I don't think that type of question or logic applies when predicting stock markets.
chaxor··on Show HN: ElectricSQL, Postgres to SQLite active-active sync for local-first apps
This is a great project.

Also, it reinforces that fact that everyone should pronounce SQL as 'squeal'.

'S.Q.L.' is far too many syllables, and 'sequel' isn't much better. 'Squeal' is just one syllable, so obviously better.

Also, I love how this project now has me singing "OO girl, shock me like an electricSQL".

chaxor··on Add extra stuff to a “standard” encoding? Sure, why not
Related to https://en.m.wikipedia.org/wiki/Berry_paradox

Just as a quick example of why it's a bit absurd - I name that number you just defined $zeta$. Now I make $zeta'$ = zeta^zeta. Or whatever manipulation you like. Adding constraints is addressed in the link.

chaxor··on NSA Backdoor Key from Lotus-Notes (1997)
I don't know much about hardware, but is it not possible that there is a small part of a chip somewhere deep in the highly complex systems we have that simply intercepts prior to encryption and, if some condition is met (a remote connection sets a flag via hardware set keys), encrypts/sends the data elsewhere? Something like that anyway. It seems possible, but idk how plausible it is, and if things like the Linux kernel would be likely to not report on it, if the hardware is not known enough.

Anyway, just suggesting something that wouldn't require quantum cryptography.

chaxor··on DeepMind cofounder: Generative AI is just a phase. What’s next is interactive AI
A key point here is that the centralization in this case is from a for-profit company.

If the records were centralized by the government, the issues that we see now would likely not be as prevalent.

There would be other talking points and issues to be sure, but the point is there are many ways to 'centralize' information, even including, ironically, the technologies that are infamous on HN that "de-centralize" information.

chaxor··on DeepMind cofounder: Generative AI is just a phase. What’s next is interactive AI
I don't think that's necessarily what they're saying. It certainly isn't a necessary condition.

Post quantum Homomorphic encryption is one key element to make this type of system work.

chaxor··on DeepMind cofounder: Generative AI is just a phase. What’s next is interactive AI
Is there not a win-win situation through post-quantum homomorphic encryption? I'm not an expert in the area, but I imagine it's possible to set up a centralized system that contains pretty much all patient data for a nation, in a completely encrypted state. Then each institution/hospital could apply for access to make an application which generates only an aggregated metric of interest as an output, such as the map example given above, which could provide real-time notifications to medical personnel in the area.

I realize that many will come to say that the homomorphic encryption capabilities right now require a long time and many CPU cycles to compute some plaintext equivalent, but there would still be a huge improvement in the time required to be notified in the world today.

Additionally, performing studies in academia would likely be a far better experience if the data was available in a single place, and aggregate information could be gathered with a single API.

chaxor··on Linear Book Scanner – Open-source automatic book scanner (2014)
This looks absolutely fantastic. Not only can you digitize your books, but it also shreds them for you for free!

Pretty sweet 2 for 1 deal.

chaxor··on static-server: an HTTP server in Go for static content
I'm (very) unfamiliar with webdev and hosting. Is this 'oneliner' a secure way to host files? I have heard that using a reverse proxy, some 2FA on a cloud gateway that forwards to another location, with a VLAN for the file server, and containerization, etc are all the standard practice now for ensuring security. Is there anything that works toward making this goal more easily amenable (since it's not possible with one project), or should I just keep using python -m http.serve and not care?
chaxor··on Apple’s new Transformer-powered predictive text model
UnilmCtrl seems to imply some dependence on Sochers CTRL models, but I wouldn't be surprised if it's just another name collision. Would be interesting if it were, as that would likely mean they have been working on releasing this for a few more years than the "chat-gpt craze". I am a bit biased, but I always have more respect if someone has worked in the NLP field for at least ~8 years, rather than jumping on after Sparrow and RLHF.
chaxor··on Linux on a MacBook Pro (M1 Pro): How Good Is Asahi Now?
The battery life (i.e. the backlight being either ON or OFF, without anything in between) is one of the biggest points that hinder moving to asahi for me.

Every time I open it up to try it out again, it's so bright that I try to reduce the brightness, which turns off the screen completely, which then has to be fixed by imagining what the screen state is and using hotkeys and typing out commands to get it back to full on blasting the sun out from the screen.

The selling point for this when usable IMO is having a laptop that will last 4x the (already insane) battery lifetime of macos, by having just a basic terminal and no default background processes that macos has.

chaxor··on Horcrux: Split your file into encrypted fragments
I would imagine if you're worried you could just split the files with this, and of you think the encryption is 'weak' you could encrypt again with another tool. Probably not a bad idea.
← PreviousPage 5 of 31Next →