Transporters for instance provide the ability to replicate people.
Actually, probably they don't.
I mean, replicators as described could most probably replicate people, if run at very high resolution. But high resolution is expensive: it calls for lots of expensive components running for a long time. Even the best replicators in starshipboard use thus probably aren't built for such high resolution, since most applications don't require it. In some laboratory, PhD Frankenstein could quite well be replicating people using very expensive special equipment, but it doesn't appear to be commonplace, at least not aboard starships. Replication of living tissue is a viable medical procedure, as in TNG "Ethics" and VOY "Emanations", but it is a very nonstandard and exotic one.
But transporters have to use high resolution to keep their users alive. Or do they? I'd argue not. For the same reason replicators aren't any better than they absolutely have to be, transporters probably "cheat" as much as possible in order to be practicable. And the associated technobabble suggests that a transporter breaks the user down to components known as "phased matter stream" and then moves that stream from A to reassembly at B. And that
is very, very different from building something at B from scratch.
Think of moving your son's giant LEGO castle away from the living room before the guests come. You can't carry it in one piece - but you don't want to break it into any smaller pieces than you absolutely have to, or the effort of reassembly will become overwhelming. Every time you disconnect a piece, you will have to make a (mental?) note of where it went. If you break it into elementary pieces, you will have very light loads to carry, but you will also have a giant library of notes needed for reassembly.
A transporter probably doesn't "break" the user into any smaller "pieces" than it has to - and the pieces themselves then carry part of the "assembly information" with them, liberating the computer from having to process that information. So Data can be transported from A to B, because his "pieces" carry part of the "assembly information" with them, but the machine cannot build a new Data at B because even if somebody sent undifferentiated raw materials there, there wouldn't be enough assembly information to turn those into Data.
A duplicate Data would require PhD Frankenstein's very special replicator, plus a scanner that analyzed Data to a much, much higher resolution than transporter scanners do. It could probably be done that way - but there would be easier ways to do it.
Now, we have
actually witnessed the replication of a highly complex machine without the benefit of out-of-the-ordinary hardware or effort. In DS9 "Rivals", a probability-altering device was replicated simply by letting a standard Cardassian replicator scan the original. Moreover, this standard replicator was able to make enlarged
However, we could argue that the alien machine didn't actually require all that high a resolution. Many amazing machines of today are actually built of relatively simple and coarse components, but they would still dazzle an 19th century observer who didn't understand the operating principles and had never witnessed this sort of action in nature. This may have been the case with the alien spheres of "Rivals", too.