Axelar vs LayerZero security and trust assumptions compared
Thread591 words22 likes
Compares how Axelar (PoS validator set with threshold signatures) and LayerZero (oracle plus relayer) verify cross-chain messages. Notes voting power concentration at Axelar and LayerZero's closed codebase and censorship risk.
Key claims
Axelar's active validator set is 75, not 50 as first stated (corrected in thread)
Only 20 of 47 validators wielded significant voting power at time of writing
LayerZero's default oracle is a Chainlink DON with a threshold signature among three parties
0/ I have been talking about cross-chain interoperability & bridging quite a bit lately ๐
after covering both @LayerZero_Labs & @axelarcore, today's thread will provide a high-level comparison of the security & trust assumptions underlying these technologies โ ๏ธ๐
๐งต (0/12) ๐
1/ if you're new to the interoperability / bridging space, check out the below threads first ๐
a ๐งต on bridging solutions and why light client-based state verification via zk proofs might be the future of x-chain interoperability ๐ฎ
x.com/expctchaos/status/1663563684722147329
a ๐งต on trust-minimized cross-bridging ๐
x.com/expctchaos/status/1675882205598580738
as this thread is focused on a comparison of the security/trust assumptions underlying the generic message passing protocols @axelarcore & @LayerZero_Labs, it might also be worth checking the two threads below for a high-level introduction into the workings of these protocols ๐
an introductory ๐งต on @axelarcore ๐
x.com/expctchaos/status/1674477862307631105
an introductory ๐งต on @LayerZero_Labs ๐
x.com/expctchaos/status/1648038651178409984
but let's dive into the security models ๐๐
2/ ๐ @axelarcore is its own blockchain network and underpinned by @Cosmos-based proof-of-stake principles. Validators, elected by token holders, receive voting rights proportional to their delegated stake
3/ ๐ Verification of cross-chain messages is carried out by @axelarcore using a (t,n)-threshold signature protocol. This demands the cumulative voting power of the signers, normalized to n, surpasses t (the protocol threshold), for message authentication
4/ ๐ณ๏ธ Currently, @axelarcore permits a maximum of 50 validators, requiring a minimum of 66.67% voting majority for message signing. These parameters can be adjusted through a governance vote
5/ ๐ In theory, the maximum number of validators is unrestricted. However, practical data shows a skewed distribution of voting power due to the lack of a requirement for validators to manage nodes for every supported chain
6/ ๐ Among @axelarcoreโs current 47 validators, only 20 wield significant voting power, which reduces further when focusing on a specific chain. For instance, considering nodes validating @auroraisnear, at least 8 are needed to successfully transmit a message
7/ ๐ @LayerZero_Labs is a cross-chain interoperability protocol, simplifying trustless communication between chains by assigning distinct roles to two entities: an oracle and a relayer
8/ ๐ The oracle sends the block header to the destination chain, while the relayer forwards the transaction proof. Together, they certify the message's authenticity and origin
9/ ๐ก Users enjoy the flexibility to use @LayerZero_Labsโ default oracle and relayer or to design and run their own. The default oracle is a @Chainlink DON using a threshold signature protocol amongst three parties
10/ ๐ก๏ธ @LayerZero_Labs proprietary & undisclosed codebase limits transparency of its implementation. However, its own audit acknowledges that itโs possible for apps to create & manage its own oracle & relayer, potentially submitting invalid transaction proofs & block headers
11/ โ๏ธ The trust assumption with @LayerZero_Labs' bridge depends on the actions of the oracle and the relayer. If they operate independently, it should prevent the transmission of an invalid message. However, they can arbitrarily censor messages by withholding data
12/ ๐ In conclusion, both @axelarcore and @LayerZero_Labs are offering new approaches to cross-chain interoperability, using different mechanisms to handle message verification and validation
which security model do you feel more comfortable with? ๐ค
tagging some chads to follow ๐ซก๐
@sergey_nog
@Flowslikeosmo
@DeFiMinty
@Only1temmy
@Slappjakke
@Louround_
@thesaint_
@0xSalazar
@CryptoShiro_
@ShivanshuMadan
@0xShivani
@hmalviya9
@daddysether
@bitsplaining
@kunalgoel
@hooeem
1/2
@redvelvetzip
@0xDinoEggs
@WinterSoldierxz
@0xSalazar
@CosmosHOSS
@TheDeFinvestor
@Cephii1
@RobinWhitney_
@cosmosibc
@Curious__J
@Cryptocito
@speicherx
@JayJaboneta
@zmanian
@sunnya97
@IBC_GANG
@Thyborg_
@johnniecosmos
@CryptoLikeMo
2/2
Correction: @axelarcore's active validator set consists of 75 validators, not 50 which is falsely mentioned as the maximum in this thread ๐จ
thanks for the heads up @galenmoore! ๐ค
3 images on X