A major driver for keeping lossless archives is that you can re-encode to the format-du-jour and at an appropriate bitrate/quality (for lossy codecs) for the device-du-jour without any generational loss. For example, keep FLACs around on your media server and encode to 192kbps MP3 for your cheap MP3 player that can't do any other formats and has limited space, 256kbps Vorbis for your high-end Android phone with lots of storage, AAC for your iPod, etc. Even a relatively anemic PC can encode about as fast as the flash memory in most portable devices can be written, anyway, so there's no real downside to this aside from needing the space on the media server, and 3TB hard drives are cheap these days.
I've found that, at what are now "normal" bitrates for common CODECs (lame -v0, ~192-256kbps for Vorbis and MPEG4 AAC) the lossy version is transparent to most listeners on even high-end reproduction gear. The issue is that going between those lossy formats, potentially even once as might happen to support a device with limited CODEC support where the archive copy is lossy, will eventually produce artifacts that even somebody unfamiliar with psychoacoustic compression will easily identify. If you always start from a lossless archive, that's never an issue, and you can always play the lossless version on things that support it for gits and shiggles.
That's the value of FLAC or other lossless formats. Nice thing is that, even if FLAC does fade into nothingness, as long as you can make one last dying gasp with the decode, you can, without any loss, decode it into whatever format is the new hotness. I know plenty of people who did that with their old Shorten and APE archives to turn them into FLAC.