[messaging] Do Blockchains solve Zooko's triangle? (was: Another Take At Public Key Distribution)

Jae Kwon jae at tendermint.com
Wed Sep 16 05:02:50 PDT 2015


Blockchains should solve Zooko's triangle, but in reality none do yet.  I
believe that Tendermint has the potential to solve it for good, given its
consensus algorithm design.

>From http://tendermint.com/posts/cases-for-tendermint/

With Bitcoin (and Namecoin), you can verify that “@satoshi” was registered
with a particular public key at some point in the past, but you wouldn’t
know whether the public key had since been updated without downloading the
whole blockchain. This is because the presence of a name-registration
transaction in the blockchain does not imply that later transactions hadn’t
updated the value for that key. In order for you to efficiently check for
the current value of a name, the blockchain should support a balanced
Merkle tree on the most recent name-registry state. Even if
Bitcoin/Namecoin did support such a data structure, you would still have to
download and verify all the blockchain hashes and headers, and if the value
might have been updated recently you’re still vulnerable to a
fork-censorship attack.

With Tendermint, all you need is the most recent blockhash signed by more
than 2⁄3 of the validators, and a Merkle proof that proves the current
value associated with the name “@satoshi”. You don’t even need to wait for
a single commit. If you’re interested, see this link
<https://github.com/tendermint/tendermint/wiki/Merkle-Trees#iavl-tree> for
more information on our balanced binary Merkle tree implementation.

In future posts I’ll go into detail about the consensus algorithm and how
it can provide these unique speed & security guarantees without
proof-of-work mining. For now, you can check the most recent Tendermint
spec on the github wiki here <https://github.com/tendermint/tendermint/>.

On Tue, Sep 15, 2015 at 4:05 PM, Tao Effect <contact at taoeffect.com> wrote:

> While watching one of the Scaling Bitcoin talks I was reminded of this
> thread and wanted to add a note for accuracy about this “attack”:
>
> It’s actually harder to pull off than I described. Both Namecoin and
> Blockstore (apparently) use two transactions per registration to virtually
> eliminate all possibility of this ever happening. The first transaction
> stakes claim to a name but does not reveal what it is. The second
> transaction must be sent some time later (in Namecoin, it’s recommended to
> wait 12 blocks), referencing the first and revealing what the actual key
> and value of the registration were.
>
> Thus to pull off this “attack” it’s not enough to do a 24/7 MITM, you
> would also need to mine a fake blockchain by yourself continuously while
> waiting for someone to register a name. This is detectable because such a
> MITM would be unable to match the difficulty requirement of the main
> blockchain and thus would start producing an alternative blockchain that
> would show a dramatic decrease in difficulty (several orders of magnitude
> probably).
>
> There’s really only one scenario that I can imagine this attack being
> pulled off, and that would be if the NSA/GCHQ were the #1 miner on the
> Namecoin network. All for the small possibility of catching you registering
> something and stealing it from you. lol. Heh, if that were to happen they
> would have 51%+ of the mining power and would probably choose to use their
> power in a more productive way.
>
> Cheers,
> Greg Slepak
>
> --
> Please do not email me anything that you are not comfortable also sharing with
> the NSA.
>
> On Jul 23, 2015, at 6:22 PM, Tao Effect <contact at taoeffect.com> wrote:
>
> I've just demonstrated how an attacker can perform a man-in-the-middle
> attack which lets them publish a malicious key under a name that the victim
> assumes is theirs. You don't care?
> An attacker who can mine a Namecoin fork in Alice's view of split-brain
> world could convince Alice she's successfully claimed the name.
>
>
> What you are describing implies a persistent, 24/7 MITM on Alice’s
> network, waiting for Alice to register her name (assuming she hasn’t
> already).
>
> That already, by itself, is pretty much not going to ever happen simply
> because it is too costly. There are far cheaper attacks this adversary
> could do to Alice.
>
> So I put such targeted attacks on a local network outside of the realm of
> practical feasibility.
>
> Also, even if this happened, Alice’s client could detect the attack the
> second she moved outside of the MITM’d network.
>
> So, the only real option left is for a persistent, 24/7 global MITM. At
> that point you are no longer dealing with the Internet anymore. You might
> as well smash Alice’s computer with a brick and declare a successful
> “attack” on Namecoin.
>
> This is particularly easy right now because very few people are mining
> Namecoin. Since there's so little actual Namecoin mining going on […]
>
>
> Namecoin is merge-mined with Bitcoin.
>
> - Greg
>
> --
> Please do not email me anything that you are not comfortable also sharing with
> the NSA.
>
> _______________________________________________
> Messaging mailing list
> Messaging at moderncrypto.org
> https://moderncrypto.org/mailman/listinfo/messaging
>
>
>
> _______________________________________________
> Messaging mailing list
> Messaging at moderncrypto.org
> https://moderncrypto.org/mailman/listinfo/messaging
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://moderncrypto.org/mail-archive/messaging/attachments/20150916/803c0a15/attachment.html>


More information about the Messaging mailing list