Path: nntp.eternal-september.org!news.eternal-september.org!eternal-september.org!.POSTED!not-for-mail
From: Rich <rich@example.invalid>
Newsgroups: sci.crypt
Subject: Re: AI's take on my cipher...
Date: Mon, 14 Jul 2025 19:07:20 -0000 (UTC)
Organization: A noiseless patient Spider
Lines: 67
Message-ID: <1053kh8$3imh3$1@dont-email.me>
References: <1049c0q$10d0c$1@dont-email.me>   <104mj60$dltj$1@dont-email.me> <4b6e233e7c3fb669fa324151f627c4addbfc9f70@i2pn2.org> <104r7eo$1i08p$1@dont-email.me> <95a6f265f6bdddcd037a7e48cf5258e77cec9b15@i2pn2.org> <104uecv$2ak1k$1@dont-email.me> <8e54a93978459bb7baa6896adc62508b9deb7d78@i2pn2.org> <104uqme$2cu71$1@dont-email.me> <1050ffb$2q09e$2@dont-email.me> <da6f83987b8b26ab31d7548607aaa6529a7a0f06@i2pn2.org> <10536me$3fgpt$1@dont-email.me> <49d6be48c0d398c0f4e4f95d0566ff9a98e94353@i2pn2.org>
Injection-Date: Mon, 14 Jul 2025 21:07:21 +0200 (CEST)
Injection-Info: dont-email.me; posting-host="8a23a2a3293dacde61eee0f701b686e4";
	logging-data="3758627"; mail-complaints-to="abuse@eternal-september.org";	posting-account="U2FsdGVkX1/J5MSvjtBcj+Skzgea7VsE"
User-Agent: tin/2.6.1-20211226 ("Convalmore") (Linux/5.15.139 (x86_64))
Cancel-Lock: sha1:F+DVcO/egPdXVNYNDbRnN619HRQ=

Stefan Claas <stefan@mailchuck.com> wrote:
> Rich wrote:
>> Stefan Claas <stefan@mailchuck.com> wrote:
>> > Rich wrote:
>> > 
>> > > And, how accurately do you think the average person who wants to send 
>> > > an encrypted message will be at typing this:
>> > > 
>> > > KqHtqbSca2hvI02pCMHtdKQLfHhW6OeN7iK1Fg45nMpoT+to8XpwpvARkW6UziY0iyZWUEgP/gol
>> > > gz5p3XpGCe0hZbYV2IYYLDvvRjGWj1k5IHkDX4WshBZvI5fhVssJOqVI3bzqdEW3XLD4NoGKVQg3
>> > > ZeNaSJs2hBySnkBoKGI=
>> > > 
>> > > That's 128 random bytes, base64 encoded.  128 bytes is right about the 
>> > > original "tweet length" of tweets on shitter, so there is a severe 
>> > > limit of the amount of information that can be transferred.
>> > 
>> > That why I have my az and ug program for people available, but it uses 2
>> > bytes, which should be no problem.
>> > 
>> > $ openssl rand 128 | az | ug -g
>> > ZMAXT OPNWC LZWIF OQIMR PNNQV BFQLC BRZDA RUFBT ROLQS GOLKA
>> > KKNJF ULBLO WINNL IIVVK FWTEE XRGBS UJCYS DCMWH JUMAA VLLNX
>> > MJMYS LHSKG ENKLL LUGBN YNDSP AJYMO OXUBC YQNOY QMFYW ABOPH
>> > NUVCJ KMFCM XKDVM EEXYL LVUKO VVGAU UACYV OHKUG GTVAA MWDLO
>> > KCPYN HOWVM DPNHA ZMGHV MFIKW DILNO FYQHK VQELK OMFNL EOLTL
>> > ETMPL S
>> 
>> Yes, easier to enter than raw base64.  But in this case this "easier" 
>> is like the fact that it is "easier" to move 10,000kg of sand 1km by 
>> hand than it is to move one single 10,000kg rock 1km by hand.  "Easier?" Yes, 
>> but no one will actually want to do so either by hand if they have 
>> other alternatives.
>> 
>> No one, except for the very very truly determined (a tiny sized 
>> population), will want to hand type that to maintain proper 
>> air-gapping.  So they will use USB sticks or other methods to "move" 
>> the data, opening up the possibility of transfer of an exploit via that 
>> same USB stick.
>> 
> 
> A 3.5 ich disk drive and disk for it come in handy, because you hear
> every read/write process

You hope.  A NSA level attack could hide a microcontroller and several 
GB of non-volatile memory on an otherwise normal looking interface 
board.  Some of the read/writes could then be redirected to/from that 
non-volatile memory.

Far fetched -- certianly not when such CPU's can be had from Amazon for 
$10.00.

Likely to happen to any individual - well, very unlikely, unless they 
happen to also be in the NSA's crosshairs.

> and can quickly examine the sectors with a disk editor.

The same exploited drive could perform a VW Dieselgate detection to 
detect likely access by a disk editor (the access patterns will differ 
vs. filesystem accesses) and return alternate contents or modify the 
actual return from the disk surface to make you believe anything was 
written to the sectors.  So you'd have to disk edit read on another 
machine that itself was not compromised in some way (and hope the NSA 
didn't swap the drive in that machine for another comprimised one).

And -- I'm ignoring the fact that buying a newly manufactured in 2025 
3.5" drive mechanism is all but impossible today.