I would imagine that it's reached the point of diminishing returns -- it would take months of hard work to squeeze another 1% compression out of the format, in which time hard drive space per dollar has gone up by 10% with no effort (on the part of the software developer) at all.

(There are probably still compression technologies waiting to be invented; but applying them within the constraints of an existing format is hard or impossible, and creating a new format means you need a /massive/ improvement to make switching attractive for the consumer)

Level 0 uses less CPU (and improves battery life).

Only (significantly) on the compression side. Decompressing is about the same whether you used 0 or 9.

I think they wanted hardware support, which means that much of it is set in stone.

The same holds for MP3, does it not? Come to think of it, the only CODEC stuff you really hear about after its launch is video related.

MP3 itself may not change much, but it's been superseded in the commercial world a long time ago by AAC.

The same applies though - how often do you hear AAC news? In the last few years I remember maybe one or two stories about improvements in slightly-incompatible variations; no changes to the base spec. Looking at wikipedia, even AAC+ v2 (which I believe is considered current?) seems to have been stable since 2006.

That's not really true. MP3 and AAC are designed, like many video codecs, so that the decoder is standardized and unchanging (except for performance tuning) but the encoders can continue to tweak and experiment to increase quality, reduce size etc.

The LAME encoder was the x264 of its day, and there's forums where people who care test competing audio codec implementations and newer encoder revisions to see how progress is going just like they do for video.

I'm not saying it really matters to end users in a big bang wow sense that you can hang a blog post news item on, but codecs gradually getting better benefits everyone and is continually happening in the background.

Not quite designed, but the effect is the same.

For most lossy encoding formats, there are many ways to compress a single file, and in general, we do not know how to pick the best size/encoding quality encoding of all possible encodings. You can get better compression/better quality at the same compression by searching harder or by searching smarter.

Typically, compression settings deal with the 'searching harder' approach.

Searching smarter is what the successive generations of encoders do, but sometimes, this also is controlled by the compression setting because an alternative approach may be better in compression quality, but so much slower that an quality-wise inferior but faster algorithm is kept as an option.