How Guy Zyskind came to the conclusion that Enigma needed a blockchain?

From a user:

Would work Enigma without a blockchain? if i want something computed, then i can also ask someone “without” blockchain and pay for it with bitcoins. can also be automated. the auctioning in this respect would probably be implementable with normal p2p-messaging. "most of the often circulated use cases (“health care”, supply chain, messenger, social media) can be implemented p2p and decentralized without blockchain. only in very special cases do you need a highly complex global consensus on the sequence of events among unknown, unfamiliar, potentially malicious actors. internet money makes this necessary, yes. everything else can also be a different kind of network, and is scalable.) Would MPC be not be safe without Blockchain ? Thanks

Guess i found out:
Here from Guy Zyskind:

Integrating MPC and Blockchain technology
The usage of blockchain technology in scalable MPC and in privacy-enhancing technologies (PETs) in general extends beyond the economical model of IC-MPC. In MPC
in particular, agreeing on some shared public state and verifying who are the honest parties plays a major role in constructing actively secure protocols. The main
idea is to use the blockchain as a settlement layer that can be trusted for consensus, which serves as a deterrent that discourages malicious behavior. The blockchain
stores commitments to the inputs and can verify transcripts of computations. Parties
who deviate can therefore be identified and penalized (or eliminated).
More importantly, the basis of scaling MPC to large networks is based on the
notion of quorums. A quorum operates like an elected committee that is trusted
with carrying out a computation. Several recent theoretical works have dealt with
realizing such protocols [14] [16], but they require an expensive network-wide BA
step for agreeing on some public randomness that is used as a seed for the quorum
selection protocol. Using the blockchain instead allows us to perform this selection
(almost) for free. At the time of writing this thesis, this is the first work to utilize
the quorum approach in an efficient manner in order to allow scaling the number of
parties in an MPC network.