SAFE Network concerns from an old employee

[Nigel]

@nailonhead no I don’t believe sentinel is a part of the network although it seemed so cool in a way, although secure serialization is a way to protect against these kinds of malicious injection attacks. I believe I remember in an update somewhere that David had said there was a small niggle to be fixed or implemented in that crate (edit: found it) Any thing a pen tester can help out with? - #71 by dirvine . For a little more on secure serialization read here. What is secure_serialisation] - #6 by dirvine

[Nigel]

This. Nothing else needs to be said after reading this.[quote=“dirvine, post:7, topic:10960”]
As we pass through Alpha we are providing the functionality, as this is rolled out we are spending a ton of time on security, which is implemented bit by bit, for instance currently routing meetings are very focused on security, but in phases. So data republish, group security and hop security are all part of that. Network joining is also another part of this, so you create keys and get relocated based on that, but even this relocation does allow a targeting right now, but not by creating a key to do so, but in terms of continually joining and seeing where you got located to, then do this many of times with different nodes to get close. This is prevented though using many methods, such as forced relocation with breadcrumbs, relocation with safecoin and many others. We have not selected the relocation mechanism yet as it’s not time. right now differing sized capabilities are the issue to fix and we are on that. That is a much more difficult issue and probably should have been something way up in a critique of somebody wished to be damaging, as we can prove that. There is also secure_serialisation crate we created which is not fully used either, this introduces encrypted comms in crust but still requires incrementing counter for replay prevention. We have a few parts like this otherwise we would be launched already
[/quote]

[anon40790172]

@vtnerd

That single proxy could generate forged messages easily and the client would not be able to tell even if it knew the prefix of the group it wanted (because it could be easily generated by the malicious proxy node).

In this case you still can’t do a thing. Yes, you fooled someone in a fake network as a client connected over 1 proxy (will be several in the future). So now I’m connected to your false node, whats’next? You still can’t steal my coins, you can’t know my login details. You only prevented me from using SAFE. Not a good thing, but as no private key or log in detail ever leaves my computer not a big deal. Yes, you might see I request chunk “ABC” and maybe look that chunk up in a list. And when I log in to the network you could forward to the real network and provide me my personal files but what’s the purpose of that? You can’t make a dime doing it. it actually cost you money and resources.

@vtnerd

The client could connect to multiple proxy nodes in an attempt to collect multiple responses.

Yes, this will be the case. Like 4 proxies into the network. You even end up in the same group using these proxies (relay_nodes). So if 1 real_node is evil it would be exposed quite fast as the other 3 nodes connect you to some client_managers and the evil one is not as it doesn’t have a clue about the network.

@vtnerd

The client could connect to multiple proxy nodes in an attempt to collect multiple responses. If it gets different results, which one does it trust? The “closest” nodes (previous attempt) is probably not a good idea.

You can always bootstrap using the Maidsafe Bootstrap node. Their public key is in the client. And you could create a fake account and become my “close nodes” in a forged network, but that doesn’t mean you have any client info about me. The close nodes in the real network do. They even know I have some Safecoin left to PUT some data. A group of fake nodes can’t know that info.

@vtnerd

You are absolutely correct. I was speaking to the current implementation specifically. I stated this earlier, my first thought (thinking like an attacker) would be finding a way to generate a key I wanted, then find a way to get nodes in a position to forge a network join. I would need more detailed information than this diagram though.

Even in the current implementation there is a “GetNetworkName (request)”. I’ve seen it come by in my Vault. So again, feel free to do as much POW as you want. But you can’t create or insert your own address in the network and become that address.

@vtnerd

But one thing to think about - the number of groups is directly related to the size of the network and the target for the group size. How many groups are expected? How difficult would it be for an attacker to take 50 machines and slowly attempt rejoins until they are in the same group? Its going to be easier with a small network.

This is a whole different discussion and it actually makes my point again of not being able to attack the network with POW . I started this topic some days ago and it addresses that subject. It’s nothing new as it came up several times already and the devs are thinking about it as well. No surprise as this could be a weakness for SAFE if not addressed. But it get’s addressed.

[dirvine]

@sfultong

I don’t know the details of how disjoint groups work, so maybe you’re onto something here.

That’s because it’s not

[vtnerd]

So I went back through, and read what a disaster the particular discussion about network IDs have been. At first I was discussing the current implementation (close group), then I began discussing the disjoint groups without a login process, and then I was confusing people because I was answering questions they likely were not asking. It took me some time to realize that.

For example, the question “I can forge a message” to me implies that quite literally, I am able to create a message that would appear to come from the valid authority. To most people that probably means “can you easily trick the client into accepting the wrong data.” I then thought people were implying that it would be difficult for an attacker to generate a public-key for a particular group (the proof-of-work fiasco), then I thought it was being implied that I could not generate a forgery (I can, the client cannot tell the difference without knowing what remote network IDs to expect), then I thought people were saying it was impossible to fool the client (it is, provided you control any group along the path).

I’m sorry for the confusion, but when I read statements that say “I cannot forge a message” … well it just means something different. Its a better defended network if the middle groups are literally unable to generate a forged message or if the client can actually determine the difference.

@sfultong

Also, @vtnerd,

I just read @anon40790172’s post detailing Sentinel, and it appears to address your concerns.

@nailonhead

But afaik sentinel is not part of the current design anymore.

@Nigel

@nailonhead no I don’t believe sentinel is a part of the network although it seemed so cool in a way, although secure serialization is a way to protect against these kinds of malicious injection attacks. I believe I remember in an update somewhere that David had said there was a small niggle to be fixed or implemented in that crate (edit: found it) Any thing a pen tester can help out with? - #71 by dirvine . For a little more on secure serialization read here. What is secure_serialisation] - #6 by dirvine

The new validation information should be under message routing in disjoint groups RFC. The client can verify that a majority of machines in some group voted for a particular response. Each member of that group did the same for the next group, and so forth until the target group that “owns” the resource.

@dirvine

That’s because it’s not

@sfultong

Ok, but aren’t all messages signed, so you can validate where they’re from?

As long as signed messages describing the existence of peers are obtained from multiple sources, it seems pretty robust.

If you could take over an entire group, that seems like it would offer some fairly devastating attack vector

This process only verifies what the next group claims. A single malicious group controls its resource and the responses that travel through it. So gaining control of a single group is pretty powerful. If none of them are malicious, then the client is going to arrive at the correct response. Sending requests down multiple paths is interesting, because it would detect a malicious group before the paths converged.

@anon40790172

Even in the current implementation there is a “GetNetworkName (request)”. I’ve seen it come by in my Vault. So again, feel free to do as much POW as you want. But you can’t create or insert your own address in the network and become that address.

@anon40790172

This is a whole different discussion and it actually makes my point again of not being able to attack the network with POW . I started this topic some days ago and it addresses that subject. It’s nothing new as it came up several times already and the devs are thinking about it as well. No surprise as this could be a weakness for SAFE if not addressed. But it get’s addressed.

We are not talking about the same thing. I agree if the network can force you to go through a login process, then you cannot choose your node. But if you can forge a login attempt, calculating a public-key for the desired group is possible. Just trying to “break-down-the-system”, you cannot understand how it actually works otherwise. If it were hard to generate a desired key, a forged login exploit would be less useful.

[anon40790172]

@vtnerd

We are not talking about the same thing. I agree if the network can force you to go through a login process, then you cannot choose your node. But if you can forge a login attempt, calculating a public-key for the desired group is possible.

What log in do you mean here? A Vault joining the network or a client (like SAFE Launcher) log in to the network using a relay_node. In both cases you can’t pick your own address or forge a group. You can’t connect over IP saying: hi there, I’m a new group and here are my nodes.

[digipl]

Very curious the behaviour of Mr. Clagget. It is attributed as own discoveries, problems that were already raised in various RFCs. After his failure in his POW to create key pair, indicating a profound ignorance about the conversion from client to node, now stands on the possibility of creating faked messages.
The truth is that copied, almost literally, this problem from Andreas Flacker that in an RFC wrote:

The idea behind group authorities (close groups) as well as parallelism (sending
the message along more than one path) is to provide security and redundancy.
However, the former measure can easily bypassed if the latter is undermined:
A single node could invent a whole close group and claim that it is relaying
messages from them. The recipient usually doesn’t know the sender: They neither
know their public keys, nor do they know the network layout around the sender,
so they don’t know whether the sender actually is close to the source address,
nor whether the sender is a legitimate node at all.

RFC on Disjoint Groups, of Andreas itself, is designed precisely for, among other things, avoid that problem.

Admittedly Mr. Clagget is working very hard to try to harm Maidsafe. Are reading old post, is studying the RFCs and is analysing the code in search of the slightest error (which, at least, is good for everyone). And do not hesitate to use the classic tricks of any good attacker. Use half-truths, exaggerate trivial problems and avoid answering direct questions.

Their work has been responsible, in a large extent, of the 20% drop in the price of safecoin this last days and, in addition, it has chosen, to reappear, a very important moment in the future of Maidsafe, few days before the BnkToTheFuture equity fundraising.

Coincidences exist but the appearance of Mr. Clagget ,after almost year and a half absent, the chosen time and the enormous work being taken, profitless apparent, are extremely suspicious.

This world moves much money, Mr. Clagget are looking for work since his departure from Maidsafe and a network like SAFE can damage many and very powerful people. Those who want to maintain the current centralized Status Quo, and those, with decentralized or distributed solutions, which may become obsolete in overnight.

Let everyone draw their conclusions.

[Nigel]

Wasn’t dallyshallas big blow up before some of the first test nets? And then Ben before alpha? Now Lee right before bank2tf and soon coming up on alpha 2? What or who is motivating this?

[Audity]

Yep not worth responding.

[draw]

The curious case of Mr. Clagget

[BIGbtc]

@digipl

Their work has been responsible, in a large extent, of the 20% drop in the price of safecoin this last days and, in

I don’t agree this FUD created by Cleggat is responsible for the recent decline but I do agree it has something to do with the equity raise on BNK. IMO there are some people who have an agenda and it’s personal.

I believe it’s going to get worse before it gets better and I hope the the raise is successful.

Cleggat needs to be relagated to Guthub

[sfultong]

@vtnerd

But if you can forge a login attempt

How would you go about doing that?

[Jabba]

@sfultong

How would you go about doing that?

He’s just grasping at straws dude, if he had anything meaty we’d know it by now, don’t feed the troll imo

[vtnerd]

@polpolrene

What log in do you mean here? A Vault joining the network or a client (like SAFE Launcher) log in to the network using a relay_node. In both cases you can’t pick your own address or forge a group. You can’t connect over IP saying: hi there, I’m a new group and here are my nodes.

@sfultong

How would you go about doing that?

@polpolrene and I keep bouncing between two different concepts: (1) can I generate a valid network address for a group of my choosing, and (2) can I get the group to let me join with that address.

I think there is insufficient material for me to say much about (2), but I think (1) is possible.

Unless @sfultong you were asking me to purely speculate based on the diagram posted above, I cannot really tell by your phrasing of the question.

[neo]

@vtnerd

@polpolrene and I keep bouncing between two different concepts: (1) can I generate a valid network address for a group of my choosing, and (2) can I get the group to let me join with that address.

I think there is insufficient material for me to say much about (2), but I think (1) is possible.

It is pretty self evident that you can generate an address that would be a potential group address, how long to generate is another question

###BUT

Unfortunately this conversation is going around in circles

Actually @polpolrene answered this previously.

SAFE Network concerns from an old employee

This implies that when you have a “private key” to create address A2333 you could become part of the group that has that address in it’s range. So it implies you could become part of group A23xx where you are now number 33. This is not correct. You can’t pick your own address on the network. You can’t insert or activate an address, even if you have the private key. It’s more like: I’m a brand new node and my public key is “ABCD”. I can sign whatever to prove I’m the owner of this public (and …

[vtnerd]

@neo

It is pretty self evident that you can generate an address that would be a potential group address, how long to generate is another question

I really should just let this go, but did anybody read about my proof-of-work with easy difficulty section above? Besides just crapping all over it? I mean, really calculate the prefix sizes for the groups, and how long of prefixes people are finding for bitcoin vanity addresses?

[neo]

@vtnerd

really calculate the prefix sizes for the groups, and how long of prefixes people are finding for bitcoin vanity addresses?

The reason no one is commenting on it is that it doesn’t help you gain access to the group, because it still fails which @anon40790172 has pointed out by showing the hoops to jump through to join a group and be recognised as part of the group. But yes it was pretty clear what you meant by by that, just not useful.

[AndreasF]

(I must admit I haven’t read the whole thread, so I apologise if I’m missing the point.)

I mean, really calculate the prefix sizes for the groups, and how long of prefixes people are finding for bitcoin vanity addresses?

The “prefixes” were never meant as a proof-of-work challenge. The RFC has nothing to do with proof-of-work, it’s just a suggestion to reorganise nodes differently into groups, and with that new organisation, the common prefix is simply the obvious choice for a group name.

[BIGbtc]

@vtnerd

I really should just let this go

Its the most reliable claim you have made. Go with it.