Theft Robust Crypto

[josh]

Also, in a BOD type situation, sporadic attempts to communicate from the isolated party will be able to raise alerts about conflicting client versions & other correlated facts related to that.

[josh]

When communications are problematic, the server can attempt to recover from dropouts by storing near Levenshtein distance versions of the correct set & reconciling with a suitable protocol.

[josh]

The efficiency of picking the best scoring function & prefix could be implemented for decoder sets of realistic size by a format that maps each significant digit/bit of the score to some binary or 10 way partition of decoder space.

[josh]

Remote BOD are asking for this util to seed:

Create a set of questions. Share the set among concerned parties. Software privately records answers to each Q. Without Revealing the answers, a protocol can establish the subset where there is agreement & then pick a crypto seed based on a function of the subset & the matching answers. The seed should be matching if the criteria is exact match & the subset is mutually understood. The MITM has not seen the answers, though he may be able to combine knowledge with brute force to guess them. It’s theoretically happier if there is a protocol that makes the subset unknown & adds enough questions, accepting that some are not personal & match may be chance.

[josh]

In other threads we’ve written a lot about trying to combine the themes of tamper-proof, biometrics involved in authentication, & portable multi-factor security. Here is an opportunity to think about a wide set of use cases for an embedded chip that rolls the crypto part in the context of [HERES BIOMETRICS AT TIME T],[HERES PROTOCOL MSGTYPEX TIME T] – Respond at T1… The multi-factors can be secure pki on the hardware itself, biometrics, & running conversation integrit with other pki entities that has interwoven data depending on the other factors.

What looks like a cheap but flexible design for various fobs?

[josh]

With biometrics, for example, Org i in Situation j might only ask to be (silently?) alerted when biometrics change between T1 and T2, while Org k & Situation l is storing some biometric fingerprint that might be used in checking for id recovery in case of lost tokens or sabbaticals, or moving between ids.

A lost fob can be like losing a key or a gov id or a check book – depends on what sort of authorization & identification is system configured. What are the requested use cases?

I imagine people might want to have multiple fobs that the can visually tell apart – some sort of unique external patterning might be a good feature to support, though not necessarily part of the same ic.

[josh]

Refocusing a bit: The premise here is that a standing PKI or one that has to come from a mafia run or compromised vendor is insecure due to exposure of the private key. We consider alternatives that make the the task of acquiring & keeping up with the private data that makes crypto to be much tougher & more burdensome for the kind of attackers we worry about today.

Connected to that was the correlated concern of getting started remotely without sneaker net. This discussion is incomplete without emphasizing the obvious utility of using fresh, one-off, PKI PK pairs to bootstrap. Doh!

[josh]

Note that a MITM can still intercept the sharing of public key, inserting his own forms in each direction, but once we get some shared secrets we are confident in, then the one-off pki dominates the race to the larger basis set, held in secret.

[josh]

Another wrinkle: Incorporate demonstrating control of a relatively fresh public key in various semantic correspondence – say there is some sort of begin-chapter structure and sign-block structure within the chapter at various points. A remote party can then manually select chunks of content based on semantics that they consider to be trusted with high probability. If they are not wrong, then those are usable PK.

[josh]

Colleagues suggest thinking about a social distribution application for the decoder selection format. One could have, for instance, an archive of youtube type videos or podcasts to select from on a public or semi-public site and, semi-independently, a local ability to select different decoders for processing a given set of archived content, resulting in different playback experiences or text/data extraction. The combined functionality involves different forms of social privacy/sharing, data compression, & plausible deniability.

Note that Deep State spy harassers are often loathe to reveal to affiliates the full extent of their spying or motivations for doing so.

In the realm of data compression, alternative systerms of basis functions for approximation can lead to different, parallel codes.

[josh]

Similar ideas can be used to secure crypto secrets inside of a running process on a modern OS – the current secret is not available outside of the running process, where it is encrypted in memory by secrets that are not available outside of the running process. It’s thrown away when processes at either end of a communication link go down.

Example usage – a traditional data center or devel environment that does a lot of transactions where it is not convenient to access hardware tokens each time or due heavy new security. Whatever security is max happens at the startup of long running processes – including full logging, muti-factor, location specific, etc. Then the door is shut while the processes run.

[josh]

Remphasizing – the basic idea here doesn’t solve problems of authentication. It focuses on keeping a strong crypto link with parties you decided to authenticate based on….X…
Developing systems for picking X is also a key project. PGP originally focused on face to face meetups, which is better than online anonymity or using corporations run by the genocidal military mafia to intervene. Associating crypto credentials with writings that are endorsed for both key style & political content can help. Using biometrics to assert continuity of id between different instances of authentication or positive behavior can help. The system describe above with startup agreement about questions linked to unwritten shared history can help. Recommendations from trusted friends can help.

Keeping track of continuing identity vs. switched identity will continue to be important no matter what else is achieved.

[Author]

Posts