Technically a successfully KPA is a "crack" of the encryption under most scenarios. In this case though, I think we are looking at a distinguishing attack. [...]
Also, if they already have this sort of attack, it's going to only get easier for them in the future to break the encryption scheme.
Yeah, that was also my interpretation. 👍
I suspect shard-chan would have significant
[INTEREST] in a distinguisher for their DNA obfuscation thing, even if they aren't into cryptography in general.
IRL cryptographers tend to start moving away from stuff that has already been weakened a bit more than expected(ie needs 2**119 instead of 2**128 operations) even when the encryption is still good for now.
Yeah, unfortunately my experience is that getting most engineers to stop using primitives that aren't believed to be secure anymore is... a challenge, to keep it polite:
in bleeping 2023, I'm still getting people going "HMAC-MD5 or HMAC-SHA1 are fine, there's no practical attack on those," nevermind that systems don't only need to be secure
now / at the time they were designed, but also remain secure over their entire lifetime.
The impact of a theoretical break gets worse over time too, since most academic cryptographers lose interest in beating a dead horse, whereas potential attackers get a leg up and have no interest whatsoever in publishing practical cryptanalysis results.
That's also the reason why I'm trying to push for new systems to use post-quantum cryptography
now at least as far as confidentiality is concerned, i.e. in key exchange, symmetric encryption, etc. Authentication & integrity are somewhat less time-critical, since a break in the future won't impact the integrity of communications that happened before it.
In a way, that's what's happening here too: later cryptanalysis broke confidentiality of old encrypted genomes that had been saved away, just enough to confirm the same known-plaintext occurs in both.
Also, your target is obviously an amazing biologist. What if they've also modelled the decryption attempts likely, and the tools likely to have been used? And, juggled things so a covert AI will end-up being run, as a result of the likely decryption process? That could be... bad.
(Won't happen, of course. [...] Unless, maybe, shard-chan can be sold on the idea that covert AIs makes a cyber-organism even more nasty?)
I somehow doubt that Taylor, as she's been depicted so far, would go for making weaponized, intelligent organisms: based on her self-modification, relationship with shard-chan, and interactions with Delamain, I'd expect she'd consider intelligent beings to be people regardless of whether they run on Earth-standard biology, alien crystal shards, or as machine code.
I could maybe see it eventually happen, if the organism's design goal wasn't to cause harm but either study the "bio-cryptanalysis" attempts, or find ways to report the DNA obfuscation was broken and plausibly let Taylor find where: biosafety standards (presumably) don't cover infoforms escaping containment through the Internet.
That said, non-pathogenic flies and alga would most most likely not be studied in an extreme-isolation lab: in principle, a BSL-2 facility would do (tinkertech organisms would hopefully fail the "well-characterized agents" criterion for BSL-1) and those mostly require safety cabinets for things known to (potentially) create aerosols or splashes, and extreme care in handling sharps (but that it should just be common sense). It should be well within Taylor's abilities to make the deobfuscated organism produce a gas that is harmless, innocuous, and unique-enough to be detectable; maybe some unique isomer of an otherwise-common volatile organic compound?
Taylor's not getting programming help from shard-chan. [...]
Does shard-chan even have the Entity tech-databases needed to make hidden, encrypted, covert-attack AIs???
Does it have to be a software AI? A wetware one would play more to her (and shard-chan's) strengths; even though it's not bypassing entirely the concept of biological containment, it would still be pretty hard to keep an intelligent organism contained, especially one with unknown capabilities and which wasn't expected to appear (and whose intelligence might not even have been noticed yet)
I do really like the idea of embedding a "traditional cyberpunk" AI in a biological organism though, but I'm not sure how the AI payload could actually be deployed: before it can run itself on traditional computers, its biological carrier would have to interact with computer systems somehow.
It could be as mundane as "researchers sequence the DNA, DNA contains exploits for most popular models of sequencers, gaining code execution on it, as both a foothold into the local network and a platform to bootstrap the AI... but then it's back to normal infosec, an update for the sequencer might fix the vulnerability and prevent the deployment of the embedded AI, etc.
At the very least, that sounds like an intriguing potential cooperation if Taylor ever meets an AI and they need to infiltrate an agent in a biolab (or a facility that happens to include one)
Might be a question of how good shard-chan is at cryptography? The answer is likely, 'pretty good', as the Entity tech-database would appear to contain the best tech looted from many races. And, an important part of 'best practice' use of cryptography is not reusing cyphers?
From a cryptography perspective, reusing some primitive is perfectly fine; for instance, all communications with websites (or anything else using TLS) all use one of a handful of elliptic curves for key exchange, block ciphers, etc. Reusing a
key would be an obvious problem, but that's not what was going on in
@Torlek's snippet.
Instead, they were only able to tell that the same basic primitive was used to obfuscate both organisms... but since nobody else on Earth uses it, they deduced they were (most likely) created by the same group.
Also, hiding the tech they're using, so it's 'black boxed', and can't be re-used without Tinker help seems an important part of shard procedures.
Here, the tinker tech would have been in the "hive," which was tinker-made (and plausibly actively powered/run by the shard) but it produces flies that run on conventional chemistry and physics, for all that their genome isn't encoded the same way as in Earthly organisms.
Even then, it wouldn't be necessary to understand how two tinkertech artefacts operate, to tell whether they were made by the same tinker: say, it's easy to recognise Squealer's vehicles, and nobody would confuse them with Armsmaster's bike.
Taylor's situation is a bit weird, in that with enough careful work she seems to be able to personally understand the tech of shard-chan...
Yeah, that's been quite interesting: Taylor doesn't have other parahumans to fight, and shard-chan and [best host] are a lot more interested in gaining knowledge (by reading the local Earth's publications, or doing her own labwork) than typical.
My read on that, is Taylor was (and still is) under a lot of pressure not to release obviously-alien tech, so instead she's been forced to design things that are plausible in the local tech base, with shard-chan's helping understand all the biology and chemistry involved.
... and sorry, I only just realised I got rather wordy again 😅