When I found delicious for the first time a few years ago, I went on a binge and bookmarked and tagged a whole bunch of URLs. Now I have completely stopped using the service. There is too much cruft (read useless bookmarks) in my account. One option is that I can search all my bookmarks by tags to find the URLs I am looking for. But I don't trust that I have tagged my bookmarks correctly or not. Then I tried using the Live Sync Favorites that works with IE but it doesn't seem to work all the time, and when it does it does not seem to be very reliable. It will not sync all my links. Because it is nearly unusable I have not even tried using it in Firefox. A good favorites organizer will be something much more "alive". When I add a bookmark to it, I would not have to specify keywords. It will figure those out automatically. When I search my bookmarks, it would actually go to these links and find the relevant pages based on my query. Kind of like a search engine in a bottle. If I think about it a bit more, I don't even think keyword tagging is something very useful given this kind of a search feature. In terms of display it would rank my favorites in order of last access, date when added (more recent shows up first), total number of times a link has been accessed, etc. Of course this would be a cloud service so it would work on any device and any application. It would make sense to have a native app on my device (phone or PC) instead of having to go online. But then my immediate action is to go online, so maybe it makes sense for this to be a browser based service in spite of the added steps required to access it and the network wait times. One thing that I find useful in browser based bookmarking is that I am able to create folders and categorize my bookmarks. My ideal bookmarking service will have this feature, but with an added twist. It will automatically categorize my bookmarks and I can then label them.

One thing that can be lost in all of these automated features (automated listing, automated categorization, etc.) is predictability. For example I know the link to check Seattle weather is fourth in my list or is in the "environment" category. But because the program keeps mucking with my lists and categories, I'll have now lost this link. It can be argued that a good robust search capability makes this predictability obsolete but I don't think it applies. One way to solve this problem is to provide both the user-defined and automated flavors of all features. But that kind of defeats the purpose of having automated. I don't know how to solve this problem, but if I ever start building this service, I sure want to have solved it. I can actually implement this on Azure. It would be a good sample project.