People often seem to think of abstraction (the process) as a process of cutting away detail. It's in the word: abstract is abs- (away) + traho (pull; cognate with "drag"), i.e. pull away (or pull out, "extract"). Some specific thing has "inside" it the general structure, so you reach in and pull out the general structure and leave behind the specifics.
That makes it seem like abstraction is in some sense a loss, a retreat or ascension to something thinner--something "lightweight" (hence wieldly, general, flexible). On the other hand, a mathematician might talk of high abstraction as "heavy machinery". The heavy machinery does have its own weight to it. It might be able to blast through more substantial obstacles, and it might require more effort to wield.
As a baby, you watch the sprinker go, and then the sidewalk gets wet. Being a baby, you are unburdened by heavy doors of perception, and are vulnerably exposed to the full true nature of the episode: a big blob of shiny noisy moving color smeared across time. Later you would say: "The sprinkler makes the sidewalk wet.". You couldn't have said that before because you didn't have «sprinkler», and since you didn't have «sprinkler» you couldn't go and do something to the sprinkler specifically--recognizing, considering, and then acting in relation to the sprinkler, are activities that are structured by «sprinkler».
(Yes, detail is also thrown away, in a sense. So we could say "abstraction is not strictly a loss". But this is a post-hoc analysis; with the benefit of the new structure, you can go back and describe the abstraction process as throwing away what was superfluous around the abstraction. But you couldn't have done that prospectively. In your previous ontos, without this new abstraction, you couldn't separate the abstraction from everything else--the stuff you apprehended explicitly was skew to the abstraction. Such a boundary would have to be quite uncomfortably intertwined with the boundaries that would result, if you were to view the episode/thingy in question as just being detail glommed onto each (in turn) of the many abstractions polyinstantiated by the episode/thingy.)
From "Time is homogeneous sequentially-composable determination":
A scientist unlearns or unconditions on preconceptions, and so "returns" to a more eternal point of view. But curiously, this sort of archaeology always has to first happen "in reverse". Abstraction is a gain of structure, not a loss; minds start parochial, and then expand their domain of discourse.
From "Explicitness":
For another rich example of a generally applicable idea congealing from its appearances in concrete contexts, see "The concept of function up to the middle of the 19th century" by Adolph-Andrei Pavlovich Youschkevitch (pdf). Note that abstraction is a gain of content, not a loss of content (extraction is a pure loss of content).
From "The possible shared Craft of deliberate Lexicogenesis":
Abstracting. By using preexisting material as a metaphier, new paraphrands are created——the substructural isomorphism that constitutes the metaphor points at a new idea, the shared substructure.