Acrobat has been able to do Optical Character Recognition (OCR) for years, but the quality of the OCR depends on the quality of the scan and other features. It's 2011 and OCR still tends to get most things wrong. PDFs using scans tend to be excessively large, and OCR makes them larger because it simply adds a layer over the original scans.
When I drowned my copies of SOTA64 and Shadows of Europe, I took it upon myself to scan them because it seemed like a good idea, and I was using them often. It was a pain to try and leaf through these books. So I made a PDF of them, and tried OCR. It was a catastrophic failure. Of course, this was years ago. OCR has supposedly gotten better, but I've tried it with those scans since, and other books (since the scans were of warped pages), and I've never had much luck. Although to be fair it also takes a while, and I've never had the patience after the first few pages just fail.
The better idea is to get the original text and art and redo the layout directly into PDF. The problem is that it takes forever, and Catalyst is running short on layout folks. There is also the matter of getting the original files, which aren't always available. But Wyrm's right. Searchable PDFs are great, would be swell for older books, and there is definitely a need and desire for them among the freelancers and the customers. It is a godsend that Portfolio of a Dragon is one of the books they recompiled for PDF.