Malicious cartels just might be lurking on your blockchain.

At least, that’s the latest finding from Cornell University researchers Philip Daian, Tyler Kell, Ian Miers and Ari Juels, who reached the conclusion in a paper published last week on a vote manipulation scheme it termed a dark decentralized autonomous organization, or “dark DAO.”

Describing the dark DAO as an entity set up using smart contracts, it would be undetectable, buying users votes in order to overwhelm governance systems, issue false signals or engage in market manipulation. According to the paper, such an attack would have far reaching-consequences in that it’s applicable to any project that uses a form of governance in which those who own the coins would have a say in decisions.

Adding weight to the finding, is that this distinction applies to an increasingly large amount of cryptocurrencies, including those with valuations in the billions.

Projects like EOS, Tezos, Tron, Decred and Polkadot, for instance, have all deployed various forms of blockchain voting in an effort to formalize decision-making on their software.

Several of these systems rely on a technology called delegated proof-of-stake, which requires a certain number of nodes to be chosen to validate transactions on the network. As such, token holders are allowed to stake their coins – basically posting them to the blockchain to prove they control them – in an effort to make their votes go further.

Others seek to overcome the governance hurdles faced by major blockchains by allowing stakeholders to vote on technical changes – or what Tezos calls a “self-amending crypto ledger.”

And while some of these projects have already hit roadblocks in their experimentation, according to the Cornell researchers, a dark DAO could cause havoc in a way that surpasses what’s happened in the past.

“The whole decentralization enterprise is founded on democratic ideals, so voting seems a natural governance mechanism,” Juels told CoinDesk. “Unfortunately, it’s hard to get right and until a catastrophe occurs, people tend to assume that theoretical problems won’t materialize.”

The co-author pointed to The DAO hack in 2016, where a malicious user drained 3.6 million ether from the first DAO built on ethereum, adding:

“In a post-2016 world, the fact that election systems can and will be subverted should be crystal clear.”

Past precedent

According to the researchers, this particular dilemma is another case where entrepreneurs in the blockchains space seem to be turning a blind eye to past analysis.

Ethereum founder Vitalik Buterin and ethereum researcher Vlad Zamfir, for example, have criticized on-chain voting mechanisms as “plutocracies,” whereby the wealthy – those that own more coins – rule.

The paper states:

“The blockchain space today, with predictable results, continues its tradition of ignoring decades of study and instead opts to implement the most naive possible form of voting.”

According to the paper, a dark DAO works by essentially dominating voter participation, which is especially disconcerting since many of these votes have suffered from low turnout.

One of the “attack flavors” the paper describes is that of the impact of “trusted hardware.” Because such hardware allows computation to occur in an “enclave” or private setting during which time it’s still submitting proofs, the authors argue this would allow nefarious actors to participate in the attacks without their identity being revealed.

This also means that the manipulated votes couldn’t be detected either.

“Potentially nobody, not even the DAO’s creator, can determine the DAO’s number of participants, the total amount of money pledged to the attack or the precise logic of the attack,” the paper states.

Such a cartel could overwhelm a cryptocurrency, “covertly collecting coins until it reaches some hidden threshold, and then telling its members to short the currency,” it continues.

More attacks

But that’s not to say that systems different than those employed by on-chain governance blockchains are particularly safe either.

For instance, the researchers also detail a bribery attack that could be committed against ethereum’s signaling tool, called Carbon Vote. (A proof of concept for the attack was published to correspond to the release of the paper.)

In this example, a smart contract simply offers to buy votes and can do this in a private or a public way.

The blog post warns that as blockchains begin communicating with each other – also known as interoperability – such incentive-based attacks between competing blockchains are likely to become more frequent.

“In a world with only one smart contract system, ethereum, internal incentives may lead to stable equilibria,” the paper states, adding:

“With two players, and the underdog incentivized to launch a bribery attack to destroy their competitors, such equilibria can be disrupted, changed and destroyed.”

While Jake Yocom-Piatt from Decred acknowledges that these kinds of attacks stand to be highly problematic in the future, the issue is one for both systems that deploy both on-chain and off-chain voting mechanisms.

He told CoinDesk: “It is difficult to defend against vote buying, and it is currently an open research topic how to best defend against it.”

Mitigating the threat

Speaking to CoinDesk, representatives from several on-chain governance projects – Decred, Polkadot and Tezos – said a crucial defense strategy is to raise the cost of attack.

Arthur Breitman, co-founder of the Tezos project, said, “At the end of the day, the only viable protection mechanism is ensuring that decisions involve sufficient skin in the game to ensure accountability to the network.”

Breitman also said that research into futarchy, in which decisions are made by future markets, could help on-chain governance going forward.

But according to the paper, the only defense against such attacks is more trusted hardware, “to know a user has access to their own key material (and therefore cannot be coerced or bribed), some assurance is required that the user has seen their key.”

Still, Juels noted that the reliance on trusted hardware will seem “anathema to a lot of the cryptocurrency community.” As such, he suggested the possibility of “social mitigations” or “community-implemented deterrence to election subversion.”

However, he and Daian warned of the complexity here.

“The mitigations for such threats are primarily social, in many cases imperfect, and in many cases likely complex enough to introduce additional vulnerabilities or attacks,” Daian told CoinDesk.

According to Daian, oversights of this type are common within the industry:

“In general, the blockchain space is extremely myopic: many of the ideas currently being put forward are not sustainable long-term, and only work because the systems being secured are either small or uninteresting to sufficiently motivated adversaries.”

Yet, the Cornell researchers plan to publish another article soon to discuss other available schemes that could eliminate, or at least diminish, the chance of these attacks being perpetrated.

Daian said, “I would strongly caution against direct reliance on any voting scheme vulnerable to vote buying or coercion in decision making.”

Not scared of the dark

Still, while ominous, other researchers don’t seem particularly fazed by the paper.

Griff Green from Giveth, an ethereum-based charity organization, said that little experimentation has gone into smart contract-based autonomous organizations since The DAO hack in 2016. As such, the likelihood that a group has created a dark DAO is slim, according to him.

“DAOs are built to decentralize decision making across stakeholders over shared resources. If that shared resource is ‘circumventing an on-chain election’ then sure, of course, it might be done one day, but we don’t even really have DAOs out in the wild yet,” he told CoinDesk.

“There is no foundation to really draw any conclusions on how DAOs can be used to circumvent other DAOs in their own elections,” he continued, dismissing the paper as “mental masturbation.”

Luke Duncan from Aragon, an ethereum application for building DAOs, seemed similarly calm.

While he admits the connotation around dark DAOs is negative, the industry is interested in protecting the privacy of organizations or individuals using the technology, so looked at in a different way, the research could point to positives.

He added:

“With any of these powerful technologies there’s how it can be used for useful applications and censorship resistance and then how people can use the same techniques to do more nefarious things.”