Abhinav Saxena
Glossary

The problem of many hands

Also called: many hands problem, problem of many hands, diffusion of responsibility, distributed responsibility

Definition

The difficulty of holding anyone responsible for an outcome that many people contributed to, where the group can be said to have brought it about but no individual’s part is enough to make it theirs.

By · Updated · 2 min read

A government policy does serious harm. A minister approved it on advice, the advisers summarised an analysis they did not write, the analysts used figures supplied by another department, and hundreds of officials carried it out. Each can truthfully say that their part was small and that others decided. Dennis Thompson named this the problem of many hands in 1980: because many different officials contribute in many ways, it is difficult even in principle to identify who is morally responsible for the outcome.

Van de Poel and colleagues later gave a sharper definition. The problem occurs when a collective can reasonably be held responsible for an outcome while none of its members can.

Many hands in software

Helen Nissenbaum applied the idea to computing in 1996, as the first of four barriers to accountability in a computerised society. Software is written by teams, assembled from components made by other teams, run on platforms owned by others, and used by people who did not build it. When it fails, there may be no one whose contribution, taken alone, explains the harm. Her other barriers reinforce this one: the acceptance of bugs as inevitable, the habit of blaming the computer, and ownership without liability.

Machine learning adds further hands: those who gathered the data, labelled it, chose the objective, tuned the model and decided where to deploy it.

How it differs from the responsibility gap

The two are often run together. The responsibility gap concerns a system whose behaviour its makers could not predict or control, so that the usual conditions for responsibility fail for everyone. The problem of many hands needs no such system. It arises with ordinary tools and ordinary organisations, from the division of labour alone.

Where I stand

I think much of what is presented as a new responsibility gap in AI is this older problem, and that the older problem has known answers. Thompson rejected both easy ways out: blaming only the person at the top, and blaming only “the system”. Organisations can assign responsibility in advance, so that for each foreseeable kind of failure someone is designated to answer. Where an institution has not done so, the absence of an answerable person is not a puzzle about machines. It is a choice the institution made.

My essays and papers on it

EssayThe Responsibility Gap Is a Map, Not a Hole

Related terms

The responsibility gapThe worry, due to Andreas Matthias, that learning machines can cause outcomes for which no one can fittingly be held responsible, since no one could control or foresee them.ResponsibilityThe relation between an agent and an outcome in virtue of which the agent can rightly be praised, blamed or asked to answer for it.Moral crumple zoneThe position of a human operator in a highly automated system who absorbs the moral and legal blame when the system fails, although they had little real control over what it did.Meaningful human controlThe requirement that people, and not automated systems, remain in control of and morally responsible for consequential decisions, in a sense stronger than merely having a person present who could in principle intervene.AgencyThe capacity to act: to do something, rather than merely have something happen, for reasons that are in some sense one’s own.

Further reading

  1. Thompson, D. F. (1980). Moral responsibility of public officials: The problem of many hands. American Political Science Review 74(4): 905–916.doi.org/10.2307/1954312
  2. Nissenbaum, H. (1996). Accountability in a computerized society. Science and Engineering Ethics 2(1): 25–42.doi.org/10.1007/bf02639315
  3. van de Poel, I., Nihlén Fahlquist, J., Doorn, N., Zwart, S. and Royakkers, L. (2012). The problem of many hands: Climate change as an example. Science and Engineering Ethics 18(1): 49–67.doi.org/10.1007/s11948-011-9276-0