Blind tagging vs. Suggestive tagging

Another topic in the Tagging book that I found to be of particular interest was the section on how tagging systems can support users. Some systems, like LibraryThing, offer only 'blind tagging' - the system doesn't offer any tagging suggestions, nor does it show you tags that others have already applied to the item until after you've applied your tags. Other systems, like last.fm or Delicious offer some support to their users by suggesting tags, either from the users own set of tags, or from tags that have already been applied to the item. There are strong arguments to be made in support of both approaches - the blind tagging approach leads to a much more diverse (and potentially sparse) set of tags - while with the suggestive model tags tend to converge more quickly on a standard vocabulary.

I've read one paper that talked about tag vocabulary convergence as a function of users being exposed to other's tags, (Cattuto, Semiotic dynamics and collaborative tagging) but it didn't offer any conclusions as to whether blind tagging or suggestive tagging yields better data. Of course, blind tagging is not really 'blind' since users do eventually get to see tags contributed by other users.

I am interested in what taggers and and tag implementers think about about blind vs. suggestive tagging.

Personally I think blind tagging allows for much more useful information to be gleaned. By suggesting tags used by others you automatically skew tags toward specific things. Sometimes this might be the behavior one wants.

At LT, the tagging is supposed to be more personal, putting tags that mean something to the individual, not bandwagon jumping. Of course, when blind tagging you will need a LOT more tags to start getting useful cross cut information. BUT by doing this you can get a much larger number of catches.

lets take Kleptomania. say I have several books about this subject. possible tags could be 'Kleptomania', 'Klepto', 'irresistible thieving' and probably many others. in a suggestive setting Kleptomania would most likely be the winner. however those who might think of 'Klepto' might not even know it is short for Kleptomania and miss the book or the other way around. By having blind tagging it encourages more variety and where the variety matches you get strong correlations. If what people want are strong hierarchy's or obvious categories (especially for books) that is what subject headings are for.

Blind tagging helps you find others who think similar to you on specific books. If two of you use the tag 'Farm boy and bratty Princess' you might find a lot more books in common that you like, than using the tag 'World building Fantasy'.

I agree with readafew that blind tagging creates a richer folksonomy. If we insist on suggestive tagging then we are actually creating a type of controlled vocabulary and loose the whole point of tags.

On the other hand, the main idea in tagging is for personal use. When I tag something I don't take into consideration what will be good for others who look at my tag cloud. I need to create some kind of controlled vocabulary out of my own tags. If I don't then I am not going to find my books/websites when I need them.
If I "blind tag" I'll end up using different tags to describe the same thing. When searching for my books tagged with "cat" I'll miss those that I tagged with "cats". If the system would have suggested the tag "cats" (because I had already used it or because it was more popular than "cat") I would find all of my books.