Accountability
While many governing systems pre-vet their candidates, what about post-vetting them? How does a governing structure impose accountability on those operating it—and on the system itself?
Accountability systems can range from impeachment to demotion or motions of no confidence.
Most accountability systems boil down to five options:
- Automatically triggered systems (e.g., a customer service employee is fired when they fail to respond to a given number of emails in a week or a CEO is fired when a stock price dips below a certain value)
- Single individual watchdogs (who themselves are exposed to corruption)
- Council watchdogs and citizen oversight
- Other elected bodies that judge indications of competence (In the U.S., this can be seen with dynamics such as Article 25 and impeachment)
- Recall elections (in which, through some mechanism, those who voted an individual into power are in charge of monitoring and removing them)
In addition to choosing the mechanism of accountability, one must choose the portion of the system’s hierarchy that is held accountable. For example, if an employee is caught stealing, do you punish only the employee, or both the employee and their boss? Having run large companies in the past, we strongly recommend that hierarchical systems always be designed with “single point” accountability and that said point of accountability is at least one level above the level of the action being held to account.
Any automatic accountability system should target the boss of a person who fails at something and never the individual. An individual’s direct supervisor should determine how that individual should be held accountable. This allows decisions about accountability to be made more dynamically by those closer to the infractions and ensures that managers always have a sense of full responsibility over their reports’ actions. This would hurt day-to-day productivity by incentivizing more stringent monitoring mechanisms, however manager accountability can dramatically reduce corruption and the risk of catastrophic failure.
In general, punishment dealt by the upper hierarchy should always target a single point and never be distributed. If you target management, they will naturally distribute some of the pain associated with being held to account and that pain will be more intelligently distributed.
Backup accountability systems also work in some instances—specifically those tied to safety-related tasks. Backup accountability systems use buddy systems, where if a task is not properly completed, not only will person A be punished, but person B will be punished as well. For example, if a valve on a dam were left open by mistake, both Dave and Darla will be held fully responsible, even if it was Dave’s turn to close it (because Darla was responsible for making sure Dave did close it).
Backup accountability works best when applied across multiple teams instead of two people who always work closely together. Doing so reduces the risk of alliances forming in which team members collectively decide to be lax about the rules (especially if they’re related to low probability outcomes). Such toxic dynamics contributed to the Quintinshill rail disaster, which killed over 200 people in Scotland. Leading up to the tragedy, railroad signalmen traded when some would arrive on the job and what their responsibilities would be, optimizing for convenience over safety.