Suppose that I had a small single-room apartment, and a set of Star Trek style holoemitters which can generate shaped and colored force fields that simulate objects for the purposes of both sight and touch. I could then with a simple verbal command transform my room from a bedroom, complete with bed, bedding and bedside table, to a kitchen with all the appurtanances, to a well-appointed bathroom, to a cozy library, to a living room with as much soft, cushy furniture as would fit. The furniture would be fully usable.
Now, suppose I went to Ikea and bought a lovely new set of sofa designs and uploaded them to my holoemitter. Shouldn't I say that at this very point, when I uploaded the sofa designs, I had a new sofa? If so, then sofas aren't essentially material objects. For my new sofa would, until projected, exist only as a configuring of the memory system of the holoemitter computer. Of course, after activating the sofa projection, the force fields in my room would be sofa-shaped. But they no more would be the sofa than the arrangements of pixels on the screen are identical with the electronic book that I purchased (even in the special case that the book is only one screen long). The fields would manifest the sofa and would at most be part of it.
But perhaps one can resist the above. Maybe I have no new sofa until I project it, and it is the shaped force fields that make up the sofa. But now notice that the sofa can be turned off and then turned back on. There is presumably no numerical identity in the force fields making up the sofa on successive activations. Yet it sure seems plausible to say that it's the same sofa. (We could even have the holoemitter store data on acquired characteristics like dents and scratches.) Suppose we deny that. Then we can still ask: Why is it the same sofa from millisecond to millisecond, when it hasn't been turned off? After all, during every millisecond it is entirely produced by the holoemitter. It has no "existential inertia" (this is not so clear in the Star Trek canon, but I stipulate thus). Yet it would be weird to say that I have had thousands, or maybe even infinitely many, sofas in a single second. But what if I hack the holoemitter to emit two copies of the sofa for a big party, thereby illegally circumventing my sofa license from Ikea? Which of the two sofas being projected will be the sofa I had the day before? There just is no answer to that question.
I think that the more we think about such science fictional scenarios, the more we destabilize our concept of an artifact. And this gives more plausibility to the thought that there really are no such things as artifacts. There is just stuff (force fields, particles, memory systems) arranged artefactually, stuff that serves as sofas, computers and chairs. And we talk as if there are artifacts. But within minutes we would talk in the same way of the holofurniture!
But one might also take this approach to give plausibility to a less radical solution, namely Rob Koons' theory that artifacts are particularized social practices. There will still be tough questions with counting, though. If I buy two sofa licenses from Ikea, the holoprojector computer might not keep two separate copies of the sofa design files. It might much more efficiently store a single copy of the sofa design files with a note that I am permitted two simultaneous manifestations. It seems more plausible that there is a single particular social practice—the pair-of-sofas practice—or two practices, one per sofa. But what if I initially bought a single sofa license from Ikea, and later bought a second license. Did I lose the first sofa, and gain a pair-of-sofas?