All of the definitions of this word that apply to
image compression,
scientific experimentation, and
Rational Unified Processes skirt around the more general definition of this term.
An artifact is something that is made by humans -- not simply in the anthropological sense, but generally, in all contexts.
Lossage in image compression is an artifact resulting from the application of a human-made algorithm; the true image* is distorted indirectly by the programmer.
In scientific experiments, artifacts can be from several sources, usually related to the observation of the primary data (mis-calibrated equipment, etc.) but can also be due to environmental factors (EM radiation from overhead fluorescent lights, heavy metal ions in laboratory water, etc.). In short, anything which obscures the observation of the true phenomenon of study is an artifact.*
Of course, anthropologically, anything that humans make and leave behind is an artifact. In this sense, bones and skeletal leavings, ancient as they may be, are not truly artifacts: they are remains.
* This usage brings up some
thorny philosophical issues. For example, what, exactly, is a "true"
image? Can something so
abstract really
exist if all of our attempts to transfer and manipulate it eventually lead to something which is not the true image? Similarly, in science, what is the "true" phenomenon being observed? If our equipment interferes with it, can we ever be sure of what we're seeing? Perhaps this is
redressed by considering the
Heisenberg Uncertainty Principal: We inevitably change anything we try to observe (or apply
JPEG compression to). I also think that
Jean-Paul Sartre and others spent a great deal of time thinking about this problem. But that is best left for another node.