Advanced WoT course : how the WoT is attacked, and how it defends itself.

Friday, 18 April, Year 6 d.Tr. | Author: Mircea Popescu

You are not expected to understand this.

mircea_popescu Let me write this out plainly. A WoT is a graph made out of nodes n1..ni, with their respective relations. A Sybil is a subgraph of this, with nodes s1..sj, which act in tandem, and could in principle take control of the wot even if j << i, by leveraging the fact that the n nodes do not act in tandem. (Acting in tandem is actually a weakness of a WoT, and undesirable. Also called groupthink.) Now, this attack fundamentally depends on the n nodes treating all the s nodes similarly. If each (or even just some) n nodes treat some of the s nodes one way and some of them another way (or further fragmented), the s nodes find themselves under a communication burden, which can easily become very large. If i and j are large enough, they encounter irresolvable translation problems (p,n complete, in fact).
kakobrekla Yeah, but "treating" in this case is a manual procedure is it not.

mircea_popescu Yes. It has to be.
kakobrekla So we figured out nothing new. I mean, the WoT was always manual.

mircea_popescu No question about it, I wasn't saying we figured something new. I was just explaining why me ignoring what I suspect to be sx has an effect. It's specifically because others don't that now sx has to play two different roles, as it were. And also sy, whom I don't ignore, etc.
kakobrekla I'm not sure that help though. You just cut yourself out of more data which you could use to determine stuff.

mircea_popescu Nop. It works specifically because we fragment the dataflow.
kakobrekla Your dataflow, not others'.

mircea_popescu Theirs. As a banal example, this is why teachers are so bothered if kids in class aren't paying attention. The teacher is the Sybil trying to attack the class WoT. The kids not paying attention are... ignoring the teacher. Because they are ignoring it at random points in its speech, soon enough its task of speaking to them becomes impossible.
kakobrekla Yeah, but theres just one problematic kiddo called Mircea :D

mircea_popescu So far. The problem is, of course, that attempts to isolate me are just about the reddest flag one could wave for most everyone there.
kakobrekla Ok, lets assume a simple case, with s1 and s2 which are both legit.

mircea_popescu Okay.
kakobrekla So now each of them herps derps legit things. You see half of it.

mircea_popescu Right.
kakobrekla How does that help?

mircea_popescu As a banal explanation, for intuitive understanding : I lack the context that would make you not ask a question with an "obvious" answer. So I do ask it. It doesn't reduce to this, but this is a good example.
kakobrekla Why does this remind me of a game theory.

mircea_popescu Because it exactly is.
kakobrekla Well I understand your theory, as for the practice I guess we shall see. I'm too dumb to simulate it in my brain.

mircea_popescu Ima post the general parts of this because I want nano to comment. He may be the only one who understands this shit well enough.
kakobrekla Yea.

Category: 3 ani experienta
Comments feed : RSS 2.0. Leave your own comment below, or send a trackback.

5 Responses

  1. No need for any fancy math here or simulations or whatever. If it breaks 'group think' its a good thing.

  2. To the extent that Mircea can't be isolated, and to the extent that he can recruit and maintain a cadre of good actors, this immune system should be able to, as it were, divide and conquer. The tricky part is that Mircea must act as three things simultaneously: T helper cells, Cytotoxic T cells, and Memory T cells. Loss of MP would therefore be like AIDS^3.

  3. Mircea Popescu`s avatar
    3
    Mircea Popescu 
    Saturday, 19 April 2014

    These two comments go well together, I think. Sorta like an ad hoc koan.

  1. [...] Advanced WoT course : how the WoT is attacked, and how it defends itself. [...]

  2. [...] the breath of it. Go re-read the null sets discussion and think about it : if MP's rule that WoTs must not be normalized (ie, there can't be a "common rule") is respected, then the scoring criteria employed by nodes will [...]

Add your cents! »
    If this is your first comment, it will wait to be approved. This usually takes a few hours. Subsequent comments are not delayed.