Quick: name a composer whose "late period" music is less complex and adventurous than their "early period" music. Not many. It's easy to imagine why—the more music you write, the more fluent you get with music's constituent elements, and the more comfortable you are experimenting with them. Often there's a trade-off: the surface level becomes less active and dense, but the relationships between elements and the formal structures become more intricate and subtle (Brahms, Carter). In other instances, the style remains relatively constant, but the vocabulary becomes more challenging (Ravel, Copland). Sometimes the musical ideas are concentrated into utterances of great brevity and density (Webern, Stravinsky). And sometimes everything just gets really complicated (Beethoven).
This is such a common career arc for composers that I think it's affected the way we look at the entire history of music. I'm talking about the idea of historical progress, the notion that an historical style of music supersedes another by virtue of its later chronology. There's been no shortage of conscious debunking of this notion (Kyle Gann summed it up nicely a couple months back), but it still hangs around the subconscious, influencing musical value judgments all the time.
There's an old trope (it made some guest appearances in a lot of the articles surrounding the recent Steve Reich birthday celebrations) that atonality was a "wrong turn" in the history of music and that, say, Reich et al. had returned music to its proper path. (Here's a few examples.) What this implies is that there was a "right turn" that somehow everybody missed. That's nonsense. In the first place, we only have one datum point for the evolution of Western music—for all we know, it happens this way, in this order, every time. (Compare a creationist's anti-evolution argument based on the "improbability" of human existence.) But more importantly, the history we do have bears a lot more resemblance to a random walk than an goal-oriented path. Baroque to Classical, Romanticism to Impressionism, Serialism to Minimalism, music history is filled with sharp tacks and weird digressions. If there's an ultimate goal to all this, we've certainly taken the roundabout way there.
Why is the chimera of musical progress so hard to let go of? I know why someone like Schoenberg would espouse it—the Fichte-Hegel idea of the Mind-as-Being inevitably progressing towards the Absolute must have been hard to escape in 19th-century Germany. But I also think it has to do with the similar artistic evolution of individual composers. We subconsciously assume that ontogeny recapitulates phylogeny: the development of the individual somehow mirrors the development of the entire species. If a particular composer's music gradually becomes more advanced (and you can even see this happening today among the old-school minimalists) then, somehow, it must imply that music as a whole must somehow advance and progress. Haeckel's recapitulation theory is largely discredited in biology, but its elegance and simplicity have made it a hard intellectual habit to break. Especially in this goal-oriented society: we can't just be driving around aimlessly; we must be headed somewhere. At least as far as music is concerned, though, it's better to sit back and enjoy the ride.