HNHacker News
TopNewBestAskShowJobs

eterevsky

1,390 karma · joined August 28, 2017

submissionscomments
eterevsky··on Code Is Not Literature (2014)
This is a bit weird to me since I never thought someone would just go and read source code of some program from start to finish. It would be something like reading an encyclopedia starting from the letter A.

That said, I personally DO read code. Besides code reviews that were mentioned in the article I read the code that my code depend on. Especially when I want to better understand some API, I read not only the headers, but the actual implementations of the functions that I'm using.

Sometimes I read code to better understand some algorithm. I only understood how Transformer architecture worked when I read some of its implementations.

So yeah, reading code is cool. We just don't read it as we read books.

eterevsky··on Far Manager: files and archives in Windows
I fail to understand how two-panel file managers didn’t become the default. It seems that in the 90s Norton Commander and it’s derivatives were widely used on DOS PCs. They made many file management tasks so much easier. Why don’t we see them around nowadays?

I know that FAR and probably Midnight Commander are still around, but I just don’t see anybody using them.

eterevsky··on MilkyTracker 1.04
That's very cool! Back in the 90s when my computer wasn't able to play MP3's yet, MODs and XMs were the best sounding music that it could produce.
eterevsky··on Stop Using Brave Browser
I don't think it makes any sense to apologize for beliefs that you once had since beliefs are not a matter of choice. You could only admit that your prior beliefs were wrong.
eterevsky··on Stop Using Brave Browser
I've been using it for a couple of years and I'm quite happy with it. What I like:

- In-browser adblocking, which is not dependent on the changes in extension API or extension policies.

- Reader view, which is comparable to that in Safari.

- Sidepanel with tabs.

I also liked the idea of distributing the payment to the websites that I'm visiting, and I'm sad that this didn't pan out.

Tangentially, I also like Brave Search, which is comparable to Google in quality, doesn't have ads and has nice AI summarization.

eterevsky··on Stop Using Brave Browser
Something like half of this piece is guilt by association. Literally the first argument is that 15 years ago Brendan Eich donated to some cause that the author finds repugnant, hance you shouldn't use his browser today. I also oppose Eich's views, but I believe that he has the right for his political belief, and anyway it shouldn't have any bearing on his professional contributions.

The author lists a number of crypto features, some of which failed and some turned out inconsequential. I don't see how this should have any effect on the users who use Brave for other features like adblocking or tabs sidepanel.

eterevsky··on The Physics of Readability
There were multiple attempts over the decades to move the code from files to other organization structure, and in my opinion they all made matters worse.

It's so much easier to just see a directory structure which reflects the module structure of your project and be able to quickly open any file in any tool and not just in the IDE of your chosen programming language.

eterevsky··on Andrew McCalip demonstrates synthesis of LK99
Is it possible to distinguish SC levitation from diamagnetism, based on this experiment?
eterevsky··on Splitting the Web
I think the author exaggerates a bit. Where on their scale between commercial and non-commercial web would you fit say Substack? It's obviously a commercial service, but it has it bells and whistles mostly tuned down, presenting to the users more or less pure texts of the articles.
eterevsky··on Computer scientists discover limits of gradient descent (2021)
Binary search only works on sorted data.
eterevsky··on Computer scientists discover limits of gradient descent (2021)
Here's the original paper: https://arxiv.org/abs/2011.01929

And here are the complexity classes that are mentioned in the abstract: https://complexityzoo.net/Complexity_Zoo:P#ppad and https://complexityzoo.net/Complexity_Zoo:P#pls.

eterevsky··on Welcome to Wikifunctions
I would view this as Laplace smoothing or Additive smoothing for binary distributions (https://en.wikipedia.org/wiki/Additive_smoothing). I use it all the time when I'm working estimating rates of some events from a limited amount of samples.
eterevsky··on The first room-temperature ambient-pressure superconductor?
I think it would still change the world less than computers have changed it. Without computers we wouldn't have this conversation. So much in the modern world is basically impossible without computers.

Superconductivity will for sure enable some innovations and could change how we are building power grids, but I don't see it changing the world to the same extent.

eterevsky··on The first room-temperature ambient-pressure superconductor?
I wrote electronic computers: a bunch of technologies like transistor and integrated circuits.
eterevsky··on The first room-temperature ambient-pressure superconductor?
> the biggest scientific discovery of the last hundred years

I would say that electronic computers would take take the first spot for me, but I don't deny that room-temperature superconductors would be pretty close to the top.

eterevsky··on The first room-temperature ambient-pressure superconductor?
It's a preprint. Also, NYT doesn't publish scientific papers.
eterevsky··on Why don't we get our drinking water by taking salt out seawater? (2008)
Also Scientific American, 8 years later (2016): Israel Proves the Desalination Era Is Here (https://www.scientificamerican.com/article/israel-proves-the...)
eterevsky··on The illusion of AI’s existential risk
Why wouldn’t you call a human brain “statistical pattern regurgitator” as well, if it comes to this? Matching learned patterns is a big part of how brains work.

I don’t see why a strong AI _needs_ to be trained specifically on reality. Imagine an alien race that sees world in X-Rays and neutrinos. An alien AI researcher claims that a “true intelligences” must train on the same inputs, and all those systems trained on the part of the spectrum between infrared and ultraviolet only fake intelligence.

I don’t see why intelligence has to be analogue in any way. Analogue systems are inherently noisy and contain limited amount of useful information, so you can always faithfully approximate them using digital representations.

If I am not mistaken Deepmind has shown spontaneous emergence of communication around 5 years ago when training agents in multi-player team shooters.

eterevsky··on Commercial interests petition FCC for high power allocation on shortwave
Using more power directly increases bandwidth according to Shannon's theorem.
eterevsky··on A physicist who bets that gravity can’t be quantized
I do understand that there's more to it than just closing the gap of 90'000x. That said, the fact that these numbers are on roughly the same order of magnitude makes me cautiously optimistic.

If anything, to me both measuring the gravity of 90 mg weight and putting a crystal that you could see with your naked eye into quantum superposition already seem borderline unthinkable.

eterevsky··on A physicist who bets that gravity can’t be quantized
Yes, I meant micrograms.

90'000 times sounds a lot, but just a few decades ago the biggest objects in superposition were individual particles and atoms, and the crystal from the new study is 10^16 times bigger than that.

eterevsky··on A physicist who bets that gravity can’t be quantized
If they are close enough to each other and far enough from everything else, then why not.

The smallest objects for which we measured their gravitational interaction weighed just 90 mg (https://arstechnica.com/science/2021/03/researchers-measure-...).

The biggest object put in quantum superposition weighed around 1 mcg (https://physics.aps.org/articles/v16/s45).

eterevsky··on A physicist who bets that gravity can’t be quantized
(Not a quantum physicist. Please correct me if I am misunderstanding.)

From what this article says, his assumption is that gravity is classical, but "fuzzy" or probabilistic: you can't precisely measure the gravitational field of a sufficiently small object.

In the last years we've seen progressively bigger objects being put in quantum superposition. The theory from this article is incompatible with this process continuing indefinitely. When and if we create a sufficiently big object that we are able to entangle it with something else via gravitational interaction, this would immediately disprove this theory.

So the good news is that this theory is clearly falsifiable, possibly even without creating a particle accelerator the size of the solar system.

eterevsky··on OpenWorm – A computational model of C. elegans worm
This project has existed for something like a decade. Does anyone know if there was any progress with it in the last few years?
eterevsky··on Cloud Backed SQLite
I can't see how this could for a readonly database, but how would you resolve collisions on writes and don't make them super slow in the process?
eterevsky··on Cloud Backed SQLite
It's difficult to implement correctly if there's more than one client. There is no single source-of-truth to manage any conflicts. Might be ok if the clients have only read-only access.
eterevsky··on Cloud Backed SQLite
One obvious problem that I see with this approach is that it will break if there is any change in the storage format.

With client-server architecture, the server code owns the data format, while in this storage-level remote access case, you have to ensure that all of your clients are updated simultaneously. Depending on your architecture it might or might not be feasible.

eterevsky··on Cloud Backed SQLite
I don't know much about SQLite internals, but on the face of it it sounds hacky as hell (pardon if I'm wrong).

Wouldn't it be better to make a proper client-server API similar to traditional SQL databases, but on top of SQLite?

eterevsky··on Ask HN: Are people in tech inside an AI echo chamber?
Of course I mean LLMs fine-tuned by the hedge funds, not something off-the-shelf. My reasoning is that sentiment analysis and similar techniques have been used for a while, and LLMs raise them to the next level, so they are bound to be beneficial.
eterevsky··on Ask HN: Are people in tech inside an AI echo chamber?
That's a bit curious. I was supposing that all the hedge funds should have incorporated LLMs in their models by now, since it should give them such a huge advantage. Is it not so?
← PreviousPage 5 of 16Next →