Stephan Neuhaus

Security guy, sometime guitarist, pun aficionado

Public Key

npub1mfqfz6dss97mcpg38flkr2twze8vz3y0c2upk03wuh2cd85kd7rs38k4gu

NIP-05 Address

sten@chaos-social.mostr.pub

Profile Code

nprofile1qqsd5sy3dxcgzlduq5gn5lmp49hpvnkpgj8u9wqm8chwt4vxn6txlpcpz3mhxue69uhhyetvv9ujuerpd46hxtnfduqs6amnwvaz7tmwdaejumr0dsg5rv08

Publishing to

relay.damus.io

nos.lol

Author Public Key

npub1mfqfz6dss97mcpg38flkr2twze8vz3y0c2upk03wuh2cd85kd7rs38k4gu

Show more details

Last Notes

2026-06-11 15:24:24 UTC

- reply

@nprofile…u76z For what it's worth, I bought the bundle and liked your stories very much. If I had to pick a favourite, it would be either the mad scientist with his dead wife (no spoilers!) or the man who returns to his less-than-loveable family as a boy. They're both very good.

2026-06-09 05:04:27 UTC

- reply

@nprofile…w7t7 Von der Leyen tried similar things when she was the German interior minister, earning her the nickname "Zensursula" (roughly "censoring Ursula")

2026-06-06 09:57:40 UTC

- reply

@nprofile…huat something to keep in mind, though it does not fully answer your question: There is (apparently, can't find the reference right now) some legal precedent in the US that copyright does not apply to AI-generated works.

2026-05-22 06:07:26 UTC

- reply

@nprofile…3qha Shameless plug alert: see also https://eprint.iacr.org/2025/1237, by Peter and yours truly.

2026-05-15 13:49:05 UTC

- reply

@nprofile…naxu Können wir den ersten Satz mal einrahmen?

2026-05-15 13:29:37 UTC

- reply

@nprofile…mzet Als ob die Probleme bei der Digitalisierung bei zuwenig KI und zuviel Datenschutz zu suchen wären.

2026-05-05 14:20:15 UTC

- reply

@nprofile…dgzx Kommt drauf an. Die schreiben: "Human Hantavirus infection is primarily acquired through contact with the urine, faeces, or saliva of infected rodents or by touching contaminated surfaces. Exposure typically occurs during activities such as cleaning buildings with rodent infestations" und dass Übertragung von Mensch zu Mensch selten ist. Da macht Handhygiene schon mehr Sinn, als eine Maske, oder? Es sei denn, die wittern wieder eine Tröpfcheninfektion, wo keine ist.

2026-04-25 18:07:25 UTC

- reply

@nprofile…dv93 Als ich gelesen hatte, dass Frau von der Leyen selbstzerstörende Nachrichten verwendet, wurde mir plötzlich klar, wieso Signal unter Politikern so beliebt ist.

2026-04-11 06:05:58 UTC

- reply

@nprofile…y6rq These proposals are broken so quickly that it would be a good idea IMO to put the brakes on any attempt at standardising them. But if one absolutely must standardise, then at least standardise hybrids, for crying out loud.

2026-03-31 20:10:05 UTC

- reply

@nprofile…44dh The article claims that factorisation takes exponential time, I.e., is proportional to e^(an) where a is a constant and n is the number of bits in the number to be factored. But that's not true: the best factoring algorithm known to date, the General Number Field Sieve, takes subexponential time. This class of functions grows faster than any polynomial, but slower than any exponential. One example of a subexponenrial function is e^(√n).

2026-03-31 20:00:38 UTC

- reply

@nprofile…44dh Additionally, factorisation is not exponential, but subexponential.

2026-03-28 07:58:27 UTC

- reply

@nprofile…dv93 Aaah, nice! Danke :-)

2026-03-28 07:51:22 UTC

- reply

@nprofile…dv93 Was für amtliche Lichtbildausweise gibt es denn noch? Trage mich mit dem Gedanken, so einen mitzuführen, nur für den Fall.

2026-03-25 20:10:44 UTC

- reply

@nprofile…44dh Not my discovery, BTW. I have this from a blog by quantum computing professor Scott Aaronson https://scottaaronson.blog/?p=9615

2026-03-25 19:19:27 UTC

- reply

@nprofile…44dh A followup on this one. Turns out that Shor needs so many qubits because it needs to compute x^r mod n for all r in range 1..n-1. Shor does this with quantum magic and lots of qubits.
The giant reduction in qubits for the claimed breakthrough is that they compute these classically and then load them up into the quantum computer. Of course that saves on qubits, but only at the expense of exponential running time. So, not actually a breakthrough, just more trickery.

2026-02-04 07:17:36 UTC

- reply

@nprofile…dv93 And that's where a lot if the digital sovereignty discussion goes wrong. We should not replace one set of tech monopolies (and accompanying tech billionaires) with another, that would solve exactly nothing. Instead we need robust antitrust and privacy laws, with teeth, and support for open standards. For a start.
The problem is not mainly technological and therefore can't be solved by merely exchanging Office365 with, say, NextCloud. (I like NextCloud!)

2026-01-16 06:30:44 UTC

- reply

@nprofile…w7t7 I agree, focusing on the continent of origin is stupid. The focus must be on open standards and interoperability. That said, AFAIR, Postfix was invented by Wietse Venema, so it already is "European" (whatever that means)

2025-07-17 16:45:33 UTC

- reply

@nprofile…0h8q Ah, so rum, das geht natürlich auch, daran hatte ich gar nicht gedacht.

2025-07-17 16:38:11 UTC

- reply

@nprofile…0h8q Und natürlich müsste man vorher mal darüber nachdenken, was passiert. wenn eine KI eine rechtlich verbindliche, aber politisch nicht gewollte Entscheidung trifft.

2025-07-17 16:05:13 UTC

- reply

@nprofile…0h8q Hashtag dog, eh? Let's see what kind of answers you get for this 😋

2025-07-14 09:28:14 UTC

Have you always wanted to set a new quantum factorisation record but don't have a massive physics experiment to hand? Don't worry, we're here to help!
A new preprint by Peter Gutmann and myself shows how you can replicate these quantum factorisation records with an 8-bit home computer, an abacus, or a dog. (For those not familiar with preprints: these are not peer reviewed. The point is to get them out there and to collect reviews along the way.)
Paper: https://eprint.iacr.org/2025/1237
@nprofile…fs9p

2025-04-27 12:38:48 UTC

- reply

@nprofile…vwdu Das ist bei Büchern von NoStarch https://nostarch.com tatsächlich so. Und die e-Books von NoStarch sind stets DRM-frei.

2025-04-24 06:36:42 UTC

- reply

@nprofile…5lc3 Yep. As I say in my lectures, logs are useless if all you ever do with them is rotate them out every seven days.

2025-04-04 12:59:04 UTC

- reply

@nprofile…0h8q in fact, that's precisely what secure multiparty computation was made for. It comes with proofs.

2025-04-02 09:28:05 UTC

- reply

@nprofile…d2rl Yes yes yes to all you've said in this piece! I'll be forwarding this to my bosses at my institution (a University of Applied Sciences).

2025-04-01 10:49:13 UTC

- reply

@nprofile…nhea I believe that the "non-zero NULL" is in violation of Section 6.3.2.3, paragraphs 3 and especially 4[1]. Unless of course you can guarantee that ((void*) 42) compares equal to ((void*) 0).
Of course the question is now: AITA?
[1] To be fair, this refers to the current standard C23, but I doubt that a null pointer has undergone such drastic changes in meaning since C99.

2025-03-27 11:51:22 UTC

- reply

@nprofile…fs9p Nicht nur das. Zu denen, die das nict wahrhaben wollen, gehören leider auch die Hersteller moderner Autos.
Vor noch etwa 10 Jahren hatte man ein Auto, das mit Computern ausgestattet war. Heute hat man einen Computer, um den herum ein Auto gebaut wurde. Diesen Wechsel verkraften viele Hersteller nicht, weil sie immer noch glauben, sie würden ein Auto bauen.

2025-03-26 07:33:58 UTC

- reply

@nprofile…fs9p I think it cannot be overemphasised that a) there are no (zero) credible threats against public-key cryptography at the moment. There has been no factorisation larger than 21 = 3*7 with Shor's Algorithm since 2012 (I wrote 2016 recently, but it's actually 2012). Also, b) symmetric crypto is safe from QC, provided you use long enough keys (256 bits for encryption, 512 bits for hashing). And finally, c) post-quantum algorithms are already being adopted into, e.g., TLS. Just in case.

2025-03-23 14:41:04 UTC

- reply

@nprofile…fs9p Der verlinkte Artikel enthält den Text "Quantencomputer nähern sich schon heute der Rechenleistung und Stabilität, die zum Brechen von Verschlüsselungsprotokollen mit öffentlich hinterlegten Schlüsseln erforderlich sind." Das stimmt einfach nicht. Ausserdem ist symmetrische Krypto bei 256-bit Schlüsseln (oder 512-bit Hashes) geschützt vor Grovers Algorithmus, Passwörter sind also mitnichten wertlos.

2025-03-23 14:28:50 UTC

- reply

@nprofile…fs9p In der genannten Arbeit werden 2048-bit RSA-Moduli geknackt, so weit, so korrekt. Nur: Die Primfaktoren dieser Moduli sind entweder 2 oder 6 voneinander entfernt. Genau: nicht 2 BITS voneinander entfernt, sondern 2. Damit ist das Knacken solcher RSA-Moduli ein Kinderspiel.
Wenn man vorher schon weiss, wie die Antwort lautet, ist das natürlich leicht, auch ohne Quanten-Zeugs.
Die Menge an Bullshit, die in dem Bereich produziert wird, ist irre hoch.

2025-03-21 17:59:18 UTC

- reply

@nprofile…fle4 Stupid question: how do you migrate a repo to another server, including all branches? Do you migrate every single branch? I'm asking because I'm looking for an easy way to have a backup repo to which I can simply push occasionally and have it mirror the state of the original repo.

2025-03-21 13:03:49 UTC

- reply

@nprofile…s6hu I have read the travel advice from the German Foreign Office yesterday and I did not find a "warn[ing] against traveling to the US". It can be read here https://www.auswaertiges-amt.de/de/service/laender/usa-node/usavereinigtestaatensicherheit-201382 They say that it has been unchanged since March 18.

2025-03-21 10:01:24 UTC

- reply

@nprofile…8gdf Warum nicht alles drei gleichzeitig: wütend, betrunken und hessisch?

2025-03-20 15:22:20 UTC

- reply

@nprofile…fle4 I made it to about four minutes in and had to stop. Sorry. This is terrible.

2025-03-05 08:56:56 UTC

- reply

@nprofile…ka4x If you attach sufficient amounts of C4 to those columns, they will drop from view pretty fast, I'm sure.

2025-03-04 14:58:54 UTC

- reply

@nprofile…rk93 Warum nicht das Übel an der Wurzel bekämpfen und das Verbot für Umgehung von Schutzmassnahmen kippen? Dann wäre es endlich legal, seinen Drucker (oder jedes andere technische Gerät) so zu modifizieren, dass es das tut, was *ich* will und nicht das, was der *Hersteller* will.

2025-03-04 08:12:55 UTC

- reply

@nprofile…0h8q Ich dachte zunächst, dass das nur Apple-Geräte betrifft, aber der verlinkten Mitteilung zufolge ist das nicht so. Es ist JEDES Bluetooth-Gerät betroffen.
Und es wäre verwunderlich, wenn die drei-Buchstaben-Behörden das nicht total super fänden. Damit muss man damit rechnen, dass selbst nach einem offiziellen Fix von Apple diese Ortungsmöglichkeit in der einen oder anderen Form bestehen bleibt.

2025-02-16 20:46:13 UTC

- reply

@nprofile…g7k6 Nanu, woher kennt denn das Tindlein den Rime of the Ancient Mariner?

2025-02-15 15:58:23 UTC

- reply

@nprofile…0h8q The 1875 missing birth date thing is really odd. I have looked a bit and can't find it anywhere. It doesn't sound like a programming language feature.

2025-02-11 15:59:32 UTC

- reply

@nprofile…0h8q Wir haben neulich die NZZ am Samstag abbestellt, nachdem uns auf der Titelseite (above the fold) erklärt wurde, dass Merz die Brandmauer zur AfD im Weg stünde und dass es so nicht gehe.
Anscheinend reicht es immerhin noch für einen guten Gastkommentar.

2023-08-25 07:19:41 UTC

- reply

@npub1h4u…cnpk Besonders unangenehm ist, dass man sich beim SwissPass noch nicht einmal mit opt-out gegen die Datensammlung individuell wehrenkann. Bei der Ticketprüfung wird der SwissPass eingescant und dann wer weiss was mit den Daten gemacht.

2023-08-09 19:40:28 UTC

- reply

@npub1sly…t8j7 @npub12zn…z0zw Winziger Haarspalt: Von Unix bekommt man Speicher mit brk, nicht malloc.