At Pebble we build very dynamic websites, by this we mean that most of our clients' websites have the ablility to add and remove web pages at will, what does this mean for the humble xml sitemap?
Our websites all contain what they call an xml sitemap, basically an xml file that contains a list of all the web pages available so when your friendly search engine bot comes by, they take a look at the sitemap.xml and it gives them some indication of where all of your pages are. Although they do not guarantee to crawl all the links.
The question then came up, isn't that what a html sitemap does? the answer is yes! our websites build the html sitemap automatically as well, at the same time, whenever pages are added, news items etc etc
So we concluded that even though there is no logic behind having both xml and html sitemaps we will continue to add both until we find the ultimate answer from our peers.
There is also a cool video from Matt Cutts we found shortly after to reinforce the point.