Research Summary: An Exploration of Governing via IT in Decentralized Autonomous Organizations


  • Decentralized Autonomous Organizations (DAOs) and other platform-based, technology-enabled mechanisms are governed in a fundamentally different manner than traditional (human-managed) Informational Technology (IT). They govern via IT. Established assumptions about IT governance no longer apply.
  • To deal with the challenges that this shift brings up, the authors analyzed selected DAO cases (Aragon, Flare Networks, KyberDAO, MakerDAO, and MolochDAO) and took a grounded-theory approach to build an inductive theory. The authors proposed explanations and alternative explanations, allowing them to categorize the cases. Finally, they inducted concepts for these mechanisms.
  • They organized a system for conceptualizing DAOs, including two key mechanisms with three sub-concepts each that underlie DAO governance, these are 1) establishing an algorithmic organization (enforcing precise rules and steps, encoding trust in the system, surveying the stakeholders and the environment; and 2) taming algorithmic power (granularizing empowerment, triggering reactive interventions, negotiating limits of automation).
  • About the second mechanism, “taming algorithmic power”, the authors found that the governance of IT in DAOs is dynamic, emergent, multidirectional, and granular.

Core Research Question

How are DAOs governed via IT, and what changes should be made to established assumptions about IT governance?


Mini, Tobias; Ellinger, Eleunthia Wong; Gregory, Robert W.; and Widjaja, Thomas, “An Exploration of Governing via IT in Decentralized Autonomous Organizations” (2021). ICIS 2021 Proceedings. 1. AIS Electronic Library (AISeL) - ICIS 2021 Proceedings: An Exploration of Governing via IT in Decentralized Autonomous Organizations


  • IT governance: The framework of decision rights and accountability that is related to Information Technology (IT), which aims to ensure the actions of individual actors align with the purpose of the organization. Under traditional centralized IT application frames, governance usually refers to the development and maintenance of standardized boundary resources or algorithms. In other words, IT governance is characterized by human management. This differs considerably under DAOs and blockchain applications. There, IT governance mainly refers to governance by algorithms.
  • Decentralized Autonomous Organizations (DAOs): Platform-based, technology-enabled mechanisms for governance, in which the participants are governed by a set of algorithms that allow the community to actively shape and re-orient the organization in a decentralized, collective way.


  • Blockchain-based governance faces three challenges when applied to IT governance research.

  • First challenge: DAOs heavily rely on governance via IT that includes a set of sophisticated automated algorithms encoded in smart contracts to perform as decision-making mechanisms, monitors, and coordinators. This characteristic challenges the convention of IT governance, where governance functions are primarily performed by humans.

  • Second challenge: The participants of DAOs can affect governance decisions and modification of code. This characteristic challenges the convention of the governance of IT, where IT function or centralized platform owners and their software engineers are primarily responsible for the design, development, and maintenance of IT systems.

  • Third challenge: Enforcement rules in DAOs are transparent, non-discriminatory, and automated thanks to smart contracts. This characteristic challenges the convention of the governance of IT, where rules are often latent or invisible.

  • Algorithms drive the foundational activities of DAOs, which the authors describe as an “establishing algorithmic organization” mechanism.

  • There are three sub-concepts under “establishing algorithmic organization” described as follows.

  • The first sub-concept of “establishing algorithmic organization” is Enforcing precise rules and steps, which allows performance and enforcement automatically and independently.

  • Several notable characteristics or points of Enforcing precise rules and steps are: 1) Automated control over events concerning the internal capital is a core function in DAO; 2) Smart contracts are used to sequence and bundle transactions to perform a series of actions happening in DAO; 3) Some DAOs set up relays to connect different blockchain systems and activate interoperability for DAO; 4) Some DAOs use oracles to resolve disputes between humans.

  • The second sub-concept of “establishing algorithmic organization” is Encoding trust in the system. Transparency in DAOs establishes trust in their users, which enhances participation and investment.

  • Several notable characteristics or points of Encoding trust in the system are: 1) Typical DAO systems reveal ownership of their holders and extension rules regarding the ownership; 2) A typical DAO system’s code exposes voting and decision execution mechanisms and detailed data; 3) Smart contracts in DAOs provide undisputable and up-to-date information to all holders.

  • The third sub-concept of “establishing algorithmic organization” is Surveying the stakeholders and environment, where software systems constantly monitor their stakeholders and environment to protect the integrity and interest of DAO. This enhances software systems’ central authority through their ongoing execution of routines.

  • Several notable characteristics or points of Surveying the stakeholders and environment are: 1) Oracles scrutinize stakeholder actions in exhaustive detail as a protective approach to DAO stability; 2) Oracles are used to monitor the environment; 3) Smart contracts are used to verify operations; 4) Algorithms continuously monitor and execute transactions.

  • Human intervention can counter the software dominance that the first mechanism establishes, and shape the outcomes of DAOs. The authors inducted this mechanism as “taming algorithmic power”. There are three sub-concepts under “taming algorithmic power” described as follows.

  • The first sub-concept of “taming algorithmic power” is “Granularizing” empowerment. It broadens the degree to which humans can shape their engagement in DAO by increasing decision-making opportunities on multiple levels.

  • Several notable characteristics or points of Granularizing empowerment are: 1) DAOs provide fine-grained customization of authority for a particular action. This reduces decision complexity and enables humans to govern a wide variety of DAO activities; 2) Voting systems allow stakeholders to affect automatable parameters and listings encoded in the smart contracts that impact the DAO’s actions and direction.

  • The second “taming algorithmic power” sub-concept is Triggering reactive interventions. It allows DAOs to recognize emergency scenarios and alert humans precisely and quickly.

  • Several notable characteristics or points of Triggering reactive interventions are 1) Some DAOs have algorithms to alert holders of certain conditions and provide holders an opportunity to propose to the community as a response; 2) Oracles are employed to alert users of high-risk situations; 3) Some DAOs set up human-based mechanisms against risky situations.

  • The third sub-concept of “taming algorithmic power” is Negotiating limits of automation. It implies the limits of algorithms that humans are irreplaceable.

  • Several notable characteristics or points of Negotiating limits of automation are: 1) DAOs often provide users with largely autonomous spaces such as independent control; 2) DAOs often capture the balance between the automation of smart contracts and the mediation based on human discernment, such as providing a court that can change and undo actions in a smart contract; 3) DAOs take different strategies to mitigate the limits of blockchain technology, such as imposing a complexity limit to restrain computational complexity.


  • Theoretical sampling and data collection: The authors used a theoretical sampling strategy to iteratively select DAO cases. The authors first collected an initial sample of 20 DAO cases from public materials. Then they abstracted cases with particular attention to manageable and extreme cases to maximize the variety of characteristics for analysis. By considering relative maturity, complexity, and prominent status in the DAO community, the authors condensed their findings into a final sample that included Aragon, Flare Networks, KyberDAO, MakerDAO, and MolochDAO.
  • Data analysis and approach to conceptualization: The authors took grounded theory to iteratively generate an inductive theory. By using the coding software MAXQDA to perform open coding of every DAOs in the sample, an initial coding book consisting of first-order categories was generated. Then, by identifying relationships emerging between first-order categories, second-order themes were organized. Finally, the authors discussed the results and challenges of the coding, to propose the final mechanisms, which include first-order categories, second-order themes, and conceptualizations.


  • With the principles of the two mechanisms the authors found, humans and machines interact and cooperate to operate an organization and cross-govern each other.
  • The table below shows the two key mechanisms of governance via IT, their meaning and key themes concluded by the authors.

(Mini et al., 2021)

  • The table below shows the characteristics of key themes for the first mechanism, establishing algorithmic organizations. The items in the lower layer are characteristics or meanings of the higher layer items.

(Mini et al., 2021)

  • The table below shows the characteristics of key themes for the second mechanism, taming algorithmic power. The items in the lower layer are characteristics or meanings of the higher layer items.

(Mini et al., 2021)

Discussion and Key Takeaways

  • In the first mechanism, “establishing algorithmic organization”, code is the basis of the routine matters of a DAO, such as executing transactions in the organization automatically, verifying the empowered capabilities of stakeholders, and then executing their emitted decision. The accuracy and transparency establish trust among the participants and machines in DAO.
  • In the second mechanism, “taming algorithmic power”, humans systematically and spontaneously perform designed actions to bend automated processes to align with their organizational will.
  • Algorithms govern participants in DAO, in turn, human members bend them in various pre-designed ways. Algorithms can also act as helpers to human members.

Implications and Follow-ups

  • DAOs synthesize autonomy and alignment by the first mechanism, “establishing algorithmic organization”. DAOs not only respect the autonomy and the voice of every individual participant but also monitor their actions elaborately.
  • About the second mechanism, “taming algorithmic power”, the authors found that the governance of IT in DAOs is dynamic, emergent, multidirectional, and granular. The authors also emphasized their argument that “the more a DAO relies on algorithmic governance processes embedded into the design of the platform to achieve automation, efficiency, and scalability, the more the resulting mechanism underlying governance via IT must be complemented with a mechanism focused on the governance of IT” (Mini et al., 2021).
  • Future research can work on the validation of the two mechanisms that the paper proposed. By leveraging the two proposed mechanisms, future research can find the dynamic interplay and balancing act of governance via IT and governance of IT.
  • Future works can research the underlying economic and social dynamics that justified the reason for DAOs using the two mechanisms to govern.


  • The two mechanisms this paper proposed can supplement the legal requirements in limited liability DAOs.
  • Some people doubt the sustainability of DAOs because of the failures of DAOs that heavily rely on IT, such as “TheDAO,” which in 2016 caused a hard fork of the Ethereum blockchain. The paper found that the mechanism of “taming algorithmic power” can help to increase resilience in DAOs to counter the vulnerability of purely governed by code.

How comprehensive of a system for classifying DAO governance is this? Did you see any problems with the way they divided the categories? Do you think they got enough data for this to be truly comprehensive?

1 Like

Hello, @Astrid_CH, good to see another contribution from you.

The underlying premise of this paper is that the governance of DAOs can be accomplished via automation. The rules for doing so are “transparent, non-discriminatory, and automated,” unlike conventional IT “where rules are often latent or invisible.”

So far, the algorithmic governance of DAOs sounds like an ideal development in human history.

However, unlike conventional IT, where only the ordained priestly class can design and maintain systems, in DAOs the normal participants “can affect governance decisions and modification of code.”

This sounds like the canker in the rose.

Reaching for an analogy, let me contrast Apple’s locked-down iPhone app store with the relatively wild-west Android store.

Apple is well-known for maintaining security in their ecosystem by having an ordained priestly class test apps and control what gets uploaded. In theory anyway, you can’t download a badly programmed iPhone app that will crash your phone, or one that will deliberately steal your address book or use your camera to spy on you.

In the Android world, not so much. Maybe evolution occurs more rapidly in the Android ecosystem, but orderly social relations don’t typically occur at the bleeding edge of raw evolution.

Obviously, Apple is a large centralized company, but that doesn’t invalidate my point: The principle of not allowing “normal participants” to do the equivalent of “modifying code” goes much farther back in human history than the digital era.

Why exactly do DAOs allow users to shoot their system in the foot? Why do they set up elaborate automated governance systems with “transparent, non-discriminatory rules,” only to allow normal participants to “affect governance decisions” and even modify code? Who’s interests does this serve? Is the motive for allowing this an effort to maintain decentralization purity?

And even if the interests of decentralization are at the root of this, why can’t it be implemented in a sane way? Why can’t “affecting governance” and “modifying code” be the province of (for example) a transparently elected and routinely monitored priestly class trained in theory and maintenance that won’t break the system by manhandling it?


@Astrid_CH What about some of the more playful DAOs or ones that are arranged more closely to an investment club, do those still fit into the rubric? Could their categorization be extended beyond DAOs to other organizations?