By watching a user browsing its directories the server provider can guess the directory tree. Thus it can guess which downloader downloads a share and which one only downloads to fake traffic.
Do we accept this or do we try to implement download obfuscation (e.g., by always downloading all meta files at once)?