Plasma MVP, or Minimum Viable Plasma, is the most well-known and discussed iteration of Joseph Poon and Vitalik Buterin's concept of Plasma, first laid out in an August 2017 working paper. In that paper, it's referred to as a "framework for incentivized and enforced execution of smart contracts which is scalable to … be able to represent a significant amount of decentralized financial applications worldwide." Plasma MVP has yet to be launched on the main chain, but when it is, MVP will enable a (relatively high) degree of scalability for transfers of digital assets, but that's about it. No EDCCs (aka smart contracts).
A recently released research paper by Carl Park, Aiden Park, and Kevin Jeong of Onther Inc. titled "Plasma EVM" seeks to move closer to Poon's vision by enabling execution of the Ethereum Virtual Machine's Turing-complete computation power. If successful, an EVM-enabled plasma chain would allow for EDCCs, permit existing clients like Go Ethereum and Parity to run the chain, and make Ethereum development tools available for use.
However, Plasma EVM is very much in its early stages. It's unclear how much time or bandwidth has already gone into the recently released research paper, but the authors make clear that there are security issues and technical challenges that must be addressed before it could be considered viable enough for testing.
For one, the developers point to one known attack possible under their current model where the Plasma operator could withhold blocks from finalization. They suggest a method for addressing this concern using cryptoeconomic signature aggregation (CAS), but note that it isn't enough and request discussion of the issue.
Xuanji Li, a full stack web and Ethereum developer, commented on the Plasma EVM research, echoing the authors' concerns:
The "Plasma EVM" authors suggest, as a more foolproof way of addressing the issue, that if the chain were to run on a Byzantine fault tolerant consensus mechanism like PBFT, "there is no need to consider this attack because at lease [sic] one honest node can provide the data that others are withholding."
Kelvin Fichter, an OmiseGO Plasma MVP coder, lent his input on another thread. Most basically, he called into question Plasma EVM's capacity to effectively run EDCCs – or at least certain kinds of them. He explained the issue:
"Ethereum contracts typically have no clear overall owner, even though individual components of the contract might. This means it's not easy to assign the user(s) responsible to exit the object. Contracts can also often be modified by many, if not all, network participants. Remember that exits on these objects would be blocked by any valid state transition."
"We basically want to break these contracts down into explicit pieces of state with a reasonable ownership model. .... It might not be possible to reconcile contracts where state cannot be reduced to clear ownership."
And, of course, working these kinks out will take time.
The Plasma MVP spec was first released in January 2018, and OmiseGO began implementation around the same time. Despite MVP's relative viability when the spec was first released, there have been 271 commits (public changes to the code) and 90 pull requests on its GitHub since January, and there's still no clear mainchain implementation date in site.
Meanwhile, the specifications for EVM-enabled Plasma are incomplete, with many questions left open to the community (not least what consensus mechanism should be used), and the Ethereum Virtual Machine's days are numbered. Granted, the switch to ewasm is still a ways out, but existing clients are already preparing. We'll just have to wait and see whether Plasma EVM will beat the clock.