Six years ago, Gurevich defined the class of "sequential
algorithms" by help of only a few, intuitively convincing and
amazingly general requirements. Despite their evident simplicity,
the expressiveness of sequential algorithms exceeds classical
computational models. Gurevich proved that sequential algorithms
are covered by the computational model of "Sequential Abstract
State Machines" (sequential ASMs). During the last years, this
result has been extended to various variants of ASMs including
nondeterministic, parallel, and interactive versions.

In the spirit of these results, I present a class of "distributed
algorithms" defined by likewise intuitive requirements. To this
end, I introduce "actions" as locally confined changes of a state.
A distributed run then represents a partial order of action
occurrences. I present a theorem stating that this class of
distributed algorithms is covered by a distributed computation
model derived from sequential ASMs.