I started writing this document as an internal memo, hence the distinct style, but realized that it could be kept abstract and thus shared … with you! Please enjoy this analysis of a common business term, what it means, what it’s for, and how it can go wrong.
Introduction: What Is Alignment?
“Alignment” is a term often used in business contexts to refer to an informational or communicative state of a group. As a first pass, “alignment” could be defined as “the property of a group in which all members know that all members are working toward the same goal.” So an “aligned” group is in a state of high alignment, and an “unaligned” group is in a state of low alignment, which may mean that the members are NOT working toward the same goal, or alternatively that they ARE working toward the same goal, but they don’t know that everyone else is also working toward that goal.
In this sense, producing a state of high alignment is a two-tiered problem:
How to ensure that all members are working toward the same goal? In other words, how to direct the group, producing “buy-in” or affirmation of the direction.
How to ensure that all members KNOW that all other members are working toward the same goal? In other words, how to produce common knowledge of shared goal-orientation.
Combining these two ideas gives us a simpler, clearer definition of “alignment”: an “aligned” group is a group with common knowledge affirmation of its direction. It’s important to emphasize that alignment is of an epistemic character, insofar as it refers to a specific state of knowledge or understanding maintained by individuals within a group.
In the quest to align a group, there are two fundamental failure modes: failure of direction, which is a failure at the first level of “working toward the same goal,” and failure of common knowledge, i.e. failure of communication, which is a failure at the second level, of “ensuring all members know other members are working toward the same goal.”
To better elucidate these failure modes, I propose to take as an example a common method for producing alignment: ratification. The idea of ratification is that a “proposer” puts forward in a document to a group that describes a goal, which the group members then either reject or sign off on as part of a ceremony attended by all. Once all members sign off on the proposal, it is “ratified”, representing a commitment by all members of the group to work toward the goal. At the moment that a proposal has been ratified, the group can be said to be highly aligned toward its described goal, although drift into a lower state of alignment may occur over time. Proposals often include plans of action as well as goals, and ratification often implies a commitment to carry out the plan of action described in pursuing the common goal.
Failures of Direction
How do the two failure modes above manifest in ratification? A failure of direction occurs when, for whatever reason, a member declines to sign off on the proposal. The declining member typically does so because at some level, they disagree with the proposed goal, due to a mismatch between the member and the proposer somewhere along the chain of logic that originally led the proposer to the goal. At the highest level, a failure to sign-off can indicate a fundamental difference in values, which can be irreconcilable, but more often it indicates a mismatch in assumptions between the proposer and the rejecting member. If the disagreement is truly at the level of alignment of value, then the group contains a radical structural fault, in that alignment of value among organizational authorities is the basis of the organization itself (e.g. all business leadership at a company agree that the goal is to make money, although additional value nuances may present themselves around questions of ethics).
Assuming that the members are indeed value-aligned, the proposer has two options when a member rejects his proposal: either he can attempt to persuade the member, walking through his assumptions so they can come to an agreement, which may or may not require revision of the proposal, or he can appeal to force or loyalty in order to get the member to sign-off anyway, in which case the member has chosen to disagree and commit, in that the member may disagree, but still commits to maintaining alignment with the group in terms of how he will orient his decision-making.
The former technique, of persuading through revealing mismatches in assumptions and rectifying them through discussion and revision, serves to produce alignment of thought, insofar as it leads others to the same conclusion through the use of reason. The latter technique, of appealing to disagree and commit, instead produces alignment of intention, which is often but not always natural result of alignment of thought, insofar as two value-aligned rational agents with the same informational inputs will arrive at the same conclusions (cf. Aumann’s Agreement Theorem). In either case, the aim is to achieve alignment of action, i.e. the state in which the ratified goal determines or justifies the group-relevant actions of any member, and the capacity to produce affirmed direction amongst a group is called leadership.
Vagueness can also be considered a failure of direction: members may agree to ratify a document, but walk away with different understandings of what the document meant. In this case, the group cannot be considered aligned, even though they have gone through a formal process of commitment. Thus it is essential to specify direction with a high degree of clarity, regardless of the level of detail involved in the proposal.
The outcome of a failure of direction is a lack of clarity within the group regarding the goals of the group and/or the means by which they will achieve their goals. This lack of clarity produces internal conflict and precludes effective planning and execution. The group will act inefficiently, if at all, as they lack a means to evaluate the outcomes they produce.
Failures of Communication
The establishment of an agreed-upon direction cannot occur without a means for communication: even the perfect vision must be shared amongst the group intended to execute on it. The means by which a direction is disseminated and agreed on is as essential as the direction itself, as the direction can only drive action to the extent that the director can produce shared knowledge.
Failure of communication, in contrast with failure of direction, is the outcome of knowledge asymmetry. Consider for example: rather than signing off as part of a ceremony, the proposer solicits sign-offs in individual meetings with each member. A classic example of this failure is the situation where an unscrupulous proposer solicits each individual’s sign-off by claiming falsely that each other member had already signed off, stating that the current individual is the final vote needed to ratify the proposal. In this example case, a failure of direction is “covered over” by misdirection under the cover of knowledge asymmetry.
Most cases of communication failure are not so intentionally malicious: more often, a ceremony occurs but with partial attendance, or alternatively a ceremony is not held at all and sign-offs are solicited one-by-one or in smaller groups, with no misdirection involved. Another example is when the ratification process is unclear: for example, if verbal ratification is used without a specific ceremonial term or phase, members may walk away with different understandings of who actually signed off. A “seems fine” may be understood as a “I affirm” by one member, but as a “I neither affirm nor reject” by another. Hence, formalization of process is key in preventing failures of communication when attempting to gain alignment through ratification, to ensure all members walk away with the same understanding of what transpired during the ceremony.
At a more general level, ratification itself can be viewed as one means to avoid communication failures. Other, less formal means of soliciting alignment lend themselves much more easily to this form of failure. For example, if an individual attempts to produce alignment more casually, through presenting their position and then informally seeking an understanding (to themselves) that each group member has agreed to the position, the rest of the group may remain unclear about which other members, if any, have committed to the position (the information asymmetry is between the solicitor and each agreer). In this case, the group cannot be said to be aligned, even if all members have expressed agreement, as they may not know that everyone is aligned to the same position; in this case, alignment only exists in terms of multiple solicitor-agreer dyads, and not across the group as a whole.
The consequences of communication failures manifest more subtly than directional failures. Confusion and drift begin to crop up across communication channels, where one member may be unsure of why another member made a particular decision, at worst causing conflicts that lack recourse to any understanding of a shared goal.
If direction is a function of vision, then communication is a function of organizing itself, the operational substrate of an organization. Alignment-based communication requires not only transparency but an effort to activate the whole of the communication graph, ensuring that information originating at the source is distributed or pushed outward, and that responses originating from more distal nodes are received and broadcast across the entire network. The ratification ceremony works because it broadcasts a proposal and its affirmation transparently across all participants involved, producing a state of synchronization.
Synchronization is subject to drift because each member operates within distinct streams of information, as a result of their distinct functions within the group. Thus information asymmetry builds up over time, and further, recurring rituals are needed to synchronize the group and retain a state of high alignment. The engineering standup is an example of such a ritual that preempts drift by reincorporating divergent information streams back into a common understanding: information is broadcast by each node, and other nodes update in accordance with the directionally interpretive frameworks provided by leadership.
Conclusion: Why Align At All?
The function of alignment, as hinted at above in the discussion of synchronization, is to produce a state within a group where members can function autonomously, without requiring repeated detail-oriented cycles of feedback. It is in this way that alignment forms the foundation of delegation of responsibility, a decision-making modality for organizing groups known as facilitated leadership.
Consider an alternative organizing modality, which can be described as “command and control.” In a command style organization, all information flows upward to a single node, which then bears responsibility for making all decisions and communicating actions to the group. Alignment constructs a graph-theoretically connected group that can take the place of that single command node, producing higher throughput (i.e. more information can be brought to bear at each decision point) at the cost of longer decision-making times and effort to maintain group synchronization.
In practice, most organizations operate on a spectrum between command and alignment-based distribution of responsibility, and within organizations, different functions operate at different levels of alignment. In business, a leadership team or C-suite is typically highly aligned, with each function operating at lower levels of alignment, “commanded” by the delegate of the leadership team. If the business is public, shareholders use the market as a proxy signal to influence the direction of alignment within the executive suite, although they have the option to use a governance protocol such as shareholder voting to assert direct collective control, as is common in interventions by “activist” shareholder groups.
The example of shareholder voting is worthy of further consideration, insofar as it raises the question of how alignment functions in highly distributed groups, as opposed to the smaller groups that this essay has primarily considered, or the single-node decision-making of command and control. In a large or distributed group such as a DAO, a congress, or other authoritative body, the decision-making protocol serves as a proxy for alignment: the group will definitionally not be aligned due to its size or low member-to-member communication bandwidth, but they can still perform collective decision-making insofar as they run processes which synthesize many directional input signals into a single output direction. Given that a large group is not aligned by default, formal rules are necessary to achieve this synthesis, such as proposal submission protocols, quorum and approval thresholds for referenda, finalization and execution timelines, and so on. The study of creating and implementing these rules or protocols constitutes the field of study known as “governance.”
These three systems of distributed or protocol-based governance, facilitated leadership (i.e. distribution of responsibility), and command and control can be compared along the axes of speed and throughput:
Within the context of organizational decision-making, “alignment” can be considered functionally as an index on the health or capacity of a facilitated leadership team to execute in accordance with a shared vision. A “healthy” leadership team exists continuously in a state of high alignment, maintained by repeating synchronization rituals, with whole-group ceremonies invoked on changes in direction or course.
A highly-aligned leadership team serves as the foundation of the whole organization, permitting coordination across groups that may span across continents, encompassing thousands of individuals. In contrast, an unaligned leadership team produces waste and inefficiency which can, in the worst case, result in total organizational failure. It is from this perspective that alignment can be seen as one of, if not the most important aim of a business, perhaps only secondary to the underlying vision that the aligned group agrees on and maintains.