Hacker News: tsimionescu

New comment by tsimionescu in "Wit, unker, Git: The lost medieval pronouns of English intimacy"

tsimionescu — Fri, 10 Apr 2026 04:08:27 +0000

Based on the page you linked, they pretty clearly are false friends: Old English unc is unrelated to modern German uns, it is related instead to Old Germanic unk (while modern German uns is just Old Germanic uns).

New comment by tsimionescu in "Wit, unker, Git: The lost medieval pronouns of English intimacy"

tsimionescu — Fri, 10 Apr 2026 03:52:42 +0000

Note that Romanian also has a second person singular formal pronoun, "dumneata", though it's use today is very rare and isn't actually considered polite. This is probably since Romanian, like most Romance languages, often omits the subject in phrases, so the real politeness marker ends up being just the use of second person plural verb forms to refer to a singular speaker ("mă puteți ajuta" is far more common instead of "dumneavoastră mă puteți ajuta" without the omitted subject, while the informal version is the singular "mă poți ajuta", which "dumneata mă poți ajuta" would also require - all of these phrases meaning "can [you] help me").

The origin for both is more "your lordship" ("domnia ta/voastră") than "your mercy", as well.

New comment by tsimionescu in "Claude mixes up who said what"

tsimionescu — Thu, 09 Apr 2026 21:22:00 +0000

Having that property is still a looooong way away from being able to get a meaningful answer. Consider P being something like "asks for SQL output" and Q being "is syntactically valid SQL output". This would represent a useful guarantee, but it would not in any way mean that you could do away with the LLM.

New comment by tsimionescu in "Claude mixes up who said what"

tsimionescu — Thu, 09 Apr 2026 14:38:17 +0000

I think they mean having some useful predicates P, Q such that for any input i and for any output o that the LLM can generate from that input, P(i) => Q(o).

New comment by tsimionescu in "Scientists Figured Out How Eels Reproduce (2022)"

tsimionescu — Sun, 05 Apr 2026 16:47:35 +0000

That just moves the question to how they figured out the genetic sex determination system of eels, or even how they figured out that sex is genetically determined in eels (and is it? many vertebrates don't have genetically determined sex).

New comment by tsimionescu in "Quantum computing bombshells that are not April Fools"

tsimionescu — Fri, 03 Apr 2026 09:17:37 +0000

External access to banks is already using the latest versions of TLS, so no luck needed, they will just need to upgrade as they do constantly.

New comment by tsimionescu in "Quantum computing bombshells that are not April Fools"

tsimionescu — Thu, 02 Apr 2026 16:36:15 +0000

I'm not saying it would be pretty, but it would still be infinitely better than what migtn happen in Bitcoin land - where people will either be able to steal money from those wallets directly, or the owners will permanently lose access.

New comment by tsimionescu in "Quantum computing bombshells that are not April Fools"

tsimionescu — Thu, 02 Apr 2026 10:00:50 +0000

The banking sector can relatively trivially move to a new encryption scheme - this is one of the huge advantages of centralized systems. Also, the banking sector, rather than trying to provide anonymity/pseudonimity, has KYC - they can relatively simply disable remote access if they think a current system is no longer secure, and get everyone to come in to a physical office and get a secure version, after manually verifying they are the rightful holder of the account.

Tldr; Bitcoin relies entirely on encryption, banking does not. So broken encryption is a catastrophe for Bitcoin, but just a bad week for banking.

New comment by tsimionescu in "Artemis II Launch Day Updates"

tsimionescu — Thu, 02 Apr 2026 06:36:41 +0000

You need to take into account your entire 4-vector for speed. You don't just have a speed in the 3 spatial coordinates, you're also moving thorough the "time" coordinate, and that is happening at a slower pace near a large mass like the Earth than it would of you were far away from here.

New comment by tsimionescu in "Live: Artemis II Launch Day Updates"

tsimionescu — Thu, 02 Apr 2026 06:34:04 +0000

It is correct, and you're also right that two rockets tethered to each other would not feel acceleration. The acceleration we feel in Earth's gravitational field is affecting our speed, though - it's slowing down the speed at which we move towards the future.

New comment by tsimionescu in "The Claude Code Source Leak: fake tools, frustration regexes, undercover mode"

tsimionescu — Wed, 01 Apr 2026 05:39:25 +0000

Usually, pre-LLM generated code is flagged because people aren't expected to modify it by hand. If you find a bug and track it to the generated code, you are expected to fix the sources and re-generate.

This is not at all the case with LLM-generated code - mostly because you can't regenerate it even if you wanted to, as it's not deterministic.

That said, I do agree that LLM code is different enough from human code (even just in regards to potential copyright worries) that it should be mentioned that LLMs were used to create it.

New comment by tsimionescu in "Hamilton-Jacobi-Bellman Equation: Reinforcement Learning and Diffusion Models"

tsimionescu — Mon, 30 Mar 2026 14:51:51 +0000

Infinity has properties that finite approximations of it just don't have, and this can lead to serious problems for certain theorems. In the general case, the integral of a continuous function can be arbitrarily different from the sum of a finite sequence of points sampled from that function, regardless of how many points you sample - and it's even possible that the discrete version is divergent even if the continous one is convergent.

I'm not saying that this is the case here, but there generally needs to be some justification to say that a certain result that is proven for a continuous function also holds for some discrete version of it.

For a somewhat famous real-world example, it's not currently known how to produce a version of QM/QFT that works with discrete spacetime coordinates, the attempted discretizations fail to maintain the properties of the continuous equations.

New comment by tsimionescu in "Say No to Palantir in Europe"

tsimionescu — Sun, 29 Mar 2026 15:57:23 +0000

Palantir's technology, as its own name suggests, is inherently dangerous, regardless of who controls it. The right alternative is to simply not build capabilities similar to Palantir in the EU - ideally, to legally forbid building them at all. This type of aggregated data flow simply gives too much control to whoever has access to it, and thus greatly harms democracy.

New comment by tsimionescu in "My minute-by-minute response to the LiteLLM malware attack"

tsimionescu — Fri, 27 Mar 2026 05:59:36 +0000

The whole point is that this would give a false sense of security. Scanned dependencies aren't secure, they're just scanned by some tools which might catch some issues. If you care about security, you need to run those same scans on your side, perhaps with many more rules enabled, perhaps with multiple tools. PyPI, understandably, does NOT want to take any steps to make it seem like they promise their repo doesn't contain any malware. They make various best effort attempts to keep it that way, but the responsibility ultimately falls on you, not on them.

New comment by tsimionescu in "The Resolv hack: How one compromised key printed $23M"

tsimionescu — Wed, 25 Mar 2026 17:32:33 +0000

Intra-bank transfers are instant, so is PayPal/Revolut/Zelle or whatever else, and many inter-bank transfers are also instant or very nearly so in the EU. None of these, except maybe cash, protect you from someone sinply not delivering the physical good (car + car keys) after the transfer completes.

From a legal standpoint, the bank transfer speed is anyway irrelevant - you first sign a sale contract that makes the car yours and the money theirs, before anything actually exchanges hands. If one party fails to deliver the money or the other fails to deliver the good, they are anyway liable. With instant transfers, the buyer is more likely to get scammed; with delayed transfers, both the buyer and the seller are equally as likely to get scammed - that is the only difference.

New comment by tsimionescu in "Slovenian officials blame Israeli firm Black Cube for trying to manipulate vote"

tsimionescu — Wed, 25 Mar 2026 17:23:08 +0000

Any company that sells this type of services exists as an extension of its parent state. Any contract it offers, especially to a political entity in another state, will be scrutinized by state authorities and allowed by them or not. Sometimes, those contracts will be forced on the company based on state-level negotiations.

Noone is saying that the party that contracted this company (if indeed it was a Slovenian party and not the Israeli state itself) for this service doesn't carry blame. But both the company itself and the state of Israel carry just as much blame for offering, permitting, and carrying out such services.

By your logic, if someone were to found a legal private paid assassin company in France, and then the opposition party in Germany hired this company to assassinate the German chancellor, you'd say that it's unfair for Germany to blame France for this.

New comment by tsimionescu in "Slovenian officials blame Israeli firm Black Cube for trying to manipulate vote"

tsimionescu — Wed, 25 Mar 2026 17:16:35 +0000

If you think for a second that the Israeli state allows this company to sell its services to anyone who is opposed to their interests, you really don't understand how defense companies and states work.

New comment by tsimionescu in "Wine 11 rewrites how Linux runs Windows games at kernel with massive speed gains"

tsimionescu — Wed, 25 Mar 2026 16:25:47 +0000

Electron is basically just a GUI framework. The application itself can be arbitrarily complicated, nothing stops you from building a Java + .NET + C++&COM app that includes three Windows Services that interfaces with the Electron runtime just for UI.

New comment by tsimionescu in "Epic Games to cut more than 1k jobs as Fortnite usage falls"

tsimionescu — Tue, 24 Mar 2026 20:26:44 +0000

The majority of games on Gog offer fully offline installers, where the copy you download is enough to run forever (assuming Windows and hardware compatibility, of coruse).

New comment by tsimionescu in "The Resolv hack: How one compromised key printed $23M"

tsimionescu — Tue, 24 Mar 2026 17:34:51 +0000

How does crypto solve this? You still have to send the funds and hope they give you the car - it's exactly the same as a bank transfer.