I have worked extensively with DjVu (and wrote the text extraction extension Calibre). I consider DjVu the better alternative for scanned multi-page documents, especially for efficiency reasons. Compared to the JPEG encoded pages you get in PDF or TIFF files you often have a 20x smaller file size.
For non-professional users this size difference often is not a problem, but for dessimination in the professional world it is. Just imagine having to buy 20x more storage and increased network capacity. Because a PDF reader is already necessary for many other documents (other than scanned), such users already have a PDF reader and I have seen a reluctance with non-professionals to install something just to be able to handle a more efficient file format.
DjVu is hindered by sub-par open source implementations of the encoder and AFAIK also by patent issues. My expectation is that at some point some replacement technology will become available that has all of the advantage and non of the problems, just like PNG replaced GIF. That will be a multilayer encoded image format.
As long as I will have some decent, commandline, conversion software for DjVu, PDF and any other format, I don't see any of that as a problematic. It would only be nice if I could tell Thunderbird which users only have a PDF reader so that I could attach DjVu and it would convert it to PDF before sending the email...