Well said Francesco. In reply to attack concerns, Twister is built on Bittorrent, which uses transactional algorithms to guess and block bad nodes, but I don't know if these still work in Twister because of its different behaviour.
In reply about Tor: if I *could* magically guess which packets contained child porn or other outright-evil material, I of course would block those. I would still relay politically nonsense material, but not because failing to do so is "censorship". My refusal to relay bullshit is not censorship, it's *my free speech* not to be forced to speak(/relay) on your behalf. But, because Tor is more important to society right now than my objections, I endorse it. The zero-knowledge nature of Tor routers means I cannot guess material, and I believe it is far more good than evil. If I felt it was significantly about CP, I'd stop helping/endorsing/relaying (see for example Freenet). Twister is not anonymous by design like Tor/Freenet, so I have the ability to see messages I relay. Therefore, I have an ethical imperative to act, I must either filter or stop relaying when I see harm to other humans being celebrated or fetishised. If Twister were zero knowledge, I would not be able to do this. Instead, like Tor, I would need to guesstimate the good-ness of the traffic I relay, and decide whether to participate at all. If it was even 5% CP I'd feel I could not ethically take part. Functionally that's equivalent to 100% filtering as I am no longer relaying *anything*. In this sense, Filtering would help me to relay *more* twister traffic, because II'd go from 0% to 99%. :) On 12 September 2015 08:38:44 IST, Francesco Ariis <[email protected]> wrote: >On Sat, Sep 12, 2015 at 07:39:50AM +0100, Cathal (Phone) wrote: >> If they could, then they wouldn't need help.from the twister team. >> >> The requested feature is a *refusal to relay*, it has no bearing on >other nodes. If this were useful for an *attack* then it'd be trivial >to code a fake node to do just that: connect and not relay. >> >> In P2P unless you trust a node, you never trust a node. So anyone >trusting a set of nodes to tell them the whole story will have a bad >day. If ye want child porn in a world where non-monsters can choose not >to relay it, you'd better keep hunting for nodes willing to share child >porn with you. >> >> The tragwdy of the commons is that in the absence of mechanisms for a >community to protect it, a common resource usually ends up abused. This >is a democratic and anarchic way to protect the commons of free speech >and thought from abuse before it becomes known as just >yet-another-refuge-for-cp. >> > >Hello Cathal, > thanks for your reply. > >If I am reading your message correctly, you say: "It doesn't matter, >because even if it mattered, a malicious user could implement this >trivially by themselves today, without anyone being able to detect it". >I appreciate the explanation and would personally add: "*if* it >matters, >we need to find a way to prevent this kind of attack". Since Twister >is a µblog service, I would consider a successful attack everything >where a message gets delayed by more than 15 minutes (imagine twister >being used to organise a political rally). >But as I said I am quite ignorant on the inner mechanism of Twister, >so I don't know if this is even possible (but you showed it is *not >related* to the particular feature requested). > >I'll share with you another thought I had. I think everyone in this >ML used TOR at least once (I see you suggest it as 'the ultimate >anti-censorship technology' in your blog). >Well, if we are to run an onion-router, there is a risk (I would say >certainty) that a non-zero number of packets will contain illegal, >morally abject or even terrorism related material. If every >onion-router provider said "I don't want to risk retransmitting >those", the whole network would be impaired in its primary goal >(letting political dissenters retrieve/publish information without >being identified, etc.). >I am not saying that this related 1:1 to the case at hand (it doesn't >translate well really, in one case the information is overt, in the >tor-case we just 'peel' one layer of encryption), but that's a >simple example where the _very_ understandable behaviour of many would >influence the network as a whole. > >To Erkan and the proponents of the 'censorship-resilience is not open >for debate' folks: imagine a scenario where a malicious party floods >twister with a ton of gibberish material, rendering the transmission >ineffective or very very slow. Wouldn't it be reasonable to block >those nodes then? And even share a block-list to make the process >speedier? > >What I am trying to convey with those two examples is that it might >be not easy to foresee the results of our intents (those intents >can come in the form of code or behaviours). > >To state how I feel about the issue: I think to let every node >decide what/what not to relay is a valid request, from an ethical >point of view (don't force someone to do what they don't want to >do) and from a technical point of view (you won't be able to control >this anyway). >Twister is a nice piece of software, I hope the community can come >up with a solution everyone can stand behind. -- Sent from my Android device with K-9 Mail. Please excuse my brevity. -- You received this message because you are subscribed to the Google Groups "twister-users" group. To unsubscribe from this group and stop receiving emails from it, send an email to [email protected]. To post to this group, send email to [email protected]. For more options, visit https://groups.google.com/d/optout.
