This topic contains 15 replies, has 1 voice, and was last updated by 
josh April 23, 2022 at 5:02 pm.
-
Posts
-
joshFor GT, try to think about making a line of pen drive sized dongles with super security contributions, & possibly some other features that don’t damage the fundamental no-tamper, secure mfg. purpose, so that they become long term profitable as a line while doing a lot to enhance the network security mission. They can add biometrics, key pair supply, etc.
What are the best designs? Point is that a modest econ win could be a huge network security win.
-
joshThe word ‘dongle’ is a marketing loser, associated with low volume commercial software from 3rd rate vendors. What is the key concept combo:
a) Hard to get at the info securely stored inside. Spying on screens & keystrokes is not available. Input IO to the dongle is not a viable search probe by priority design.
b) It’s small & ez to carry around on person or lock in a safe. It’s a personal device, not a share.
c) Lots of digital power is now available in the small format. Performance is not an issue.
d) Mass production economy is suitable for wide network usage.
e) No need for any side compatibility/restrictions.
f) It shouldn’t be available to network hacking even if plugged in to a networked device.
-
joshRelevant point: what role, if any should the “dongle” play in the implementation of what I call “the last 3 feet” of networking – replacing ICs with something secure & more functional?
Say, for arguments sake, 1 implementation is some kind of soft collar or head band I can put on & take off. How does it relate to the dongle for security? Lots of possibilities. Which have the best features along with top security?
-
joshPoint: In a world where secret spying on screen, typing, & SSL sessions is ubiquitous due to totalitarian mafia governments, encryption based on something like a private key that is only stored in the innermost sanctum of a hardware device, never leaving, being randomly generated at a random point of init time, gives a different kind of security that is hard to get in other ways.
Using some cryptographically viable secret, 1 can then consider other anti-totalitarian measures – e.g. hiding who is actually communicating with who by having each node due some routing & modifying some screen contents in an authorized way – e.g. cryptographically secure paraphrase of the contents that is hard to automatically match to the other end without knowing the secret.
TAKE THAT BIDEN!
-
joshAnother idea which might help some situations – “Operating System Level Filesystem Dongle” – concept is to work like the hardware dongle embedded in a file system partition – say you create a partion as Ext4Crypto – for most purposes it works just like Ext4 (substitute any other base) – but it has some hidden disk data & init-generated crypto-obscure set of routines for grabbing the private key. The upshot is that an attacker needs to grab the entire partition & also know what they are doing in order to grab the private key, which can also be secured with any other measure that is deemed helpful. How does it help? Make the partition big & the code include machine/specific salts…painful to look at many.
-
joshPartial reads of the private key storing partition can be foiled in the running system by having the OS periodically alter the bit storage of encryption of the key(s) without altering the key itself, derived by a computation. Offline guards could be based on some other strategy like reading from a disconnected drive with one-way function for next time pre-setup.
-
-
-
-
-
joshAlso interesting to think about algorithms & interfaces for how local & central webs of trust could be connected to recommendations & mechanisms for software updates & upgrades.
-
joshQ: How to sub-dimensions domains of expertise within trust?
A: The highest trust network goal is to create dynamically valid opinions with strong reality bases of support. In practice, that means that [Most convenient dynamic interface] should rule design [e.g. menus, sliders, etc]
The amount of weight that any entity gets in a sub-dimension is dymaically *upper bounded* by their global level of trust. Sub-dimensions could be created by suggestion that such & such entity be placed in such & such group. Additional weightings for a sub-dimension might be given. Their affect is to adjust the weighted partition of unity for that sub-dimension which is bounded by the overall level of trust. So if, for example, my NETWORK UPDATE ADVISORY GROUP had only 1 entity and it’s overall level of trust was only 40%, then no network updates are automatically performed or advised.
-
-
joshI emphasize that a A CENTRALIZED APPROACH TO WHAT INFORMATION CAN BE SHARED OR ACCURATELY DESCRIBED IN PROSE = TOTALITARIAN ARTIFACT OF WAR. That’s always bad unless you believe you are in a group supporting your interest in a war with opposition against you. In unseen historical reality, spies overwhelmed accurate info within humanity & created fake news media & totalitarian control of language in order to endless perpetuate false totalitarian schemes claiming a need for control based on false accounts of support & war.
A decentralized approach to trust won’t be a cure for centralized control of info. That has to end because it’s foundationally destructive to everything else. But a decentralized approach to trust can help support a decentralized approach to authority that is mutually supportive with sharing of accurate info.
-
joshWeb of Trust may also participate in secure/emergency messaging system using P2P routing of encyrypted and/or plaintext messages. Forward policies can be set to be helpful unless a given party is recognized as a bad actor based on past behavior or alerts.
GT can potentially mfg. hand held communicators with radios that can episodically trigger mobile telecom, satellite, IoT/utility, Inet-cloud or LoRaWan messages based on context. If the per unit/per usage/ & bit/usage cost is intelligently managed by policies then it can be kept low.
-
joshIn some earlier post about Trust Networks (~Dec 2019) I proposed that the local view be based on views of people that were directly assigned levels of trust by a particular user, considering the assignments of those users as well. This prevents gang astroturfing to a large extent (the naive user can still, of course, be fooled about who is trustworthy).
-
joshThis morning found me amplifying my feelings of dismay at the apparent level (%) to which key GT recruits had been subjected to icy spook replacements with mixed motives – if the person has to answer spook phone & meet them or leave when told then they are always a big risk, regardless of other intentions.
I don’t know how to fix it, but asking people to really invest work in auditing & continually building their local networks of trust, including eyeball contact in many cases, seems like a good part o the way forward.
-
joshQ: Some rumors say that Deep State run nodes are participating in networks which copy or implement the Good Team design, while using their nodes to censor content from GT members that they don’t wish to pass. How should the network design per se be dealing with this?
A: The nodes which censor GT content should be regarded as low trust or problematic from local POV of GT members and from the central POV of GT admins. If that is not happening then it could be due to either a) algorithm problem or b) mafia hack attack on the sources that seeded “central view”. The algorithms should make it hard to do b) without an effective adaptive response. Making the algorithms automatic implies some types of monitoring for this are built in.
This could be related to the “Broadcast Issues”
-
joshMake it easier to compare & analyze things that the user can set with the settings of the people they trust the most & others they might enquire about.
-
joshIn common practice, positive trust & negative trust are often treated in different ways. Positive trust builds up more slowly, over sequences of trust behavior while negative trust can be earned with a single instance of malfeasance. This is built into animal behavior in the way we react to foods that make us sick. For the network purposes, we may also want to treat positive & negative trust in different ways. However we should distinguish between an allegation of negative behavior & a convinction. Allegations from peripheral sources should be logged & investigated in appropriate ways (may be automatic testing… or not, depending) prior to going from positive trust to a negative value at the networking level.
You must be logged in to reply to this topic.