[Tagdb] I agree

[Juan Cristian Vera]

Hello again;

This is getting interesting! I think it's my turn again:

I wrote some stuff saying that I like single word tags,
and that implicit prepositions and articles in English create an 
illusion of greater intrinsic relatedness between words in composite 
nouns which are actually loose word associations.
In essence, I said, now i just type in dog & house as two separate tags 
where as before I worried that I needed to keep these two words bound 
together when tagging a dog house.


Then Tim answered, and I paraphrase liberally here:
(please correct me if you feel the need - I paraphrase to flesh out 
understanding)
Multiple word tags are more intuitive.
Naive users expect them.
You have gravitated towards single word tags due to subject matter 
expertise and adaptation.
You can't label Tierra Del Fuego well with single word tags.
I love American Idol. <kidding...>

and Nitin added(idem):
It's possible that there are linguistic subcultures which overshadow 
maturity in the single versus multiple word preference.
Failing to support multiple word tags is turning our backs on more 
literary users.


--

I am not sure that I agree with Tim's first set of assertions. What I 
was originally remarking on is that I read a posting here some time ago 
that talked about multiple word tags and the lack of their support in 
del.icio.us, and was influenced by what I read, so that I started 
tagging on del.icio.us expecting multiple word tags and missing them. I 
also read many other things about the general concept, the most 
influential of which, in  terms of helping me understand how tags work 
in the cognitive brain, was commentary in the form of blogs by Rashmi 
Sinha that talked about  folksonomy and the power of disorganized 
tagging being in immediate storage of cognitive associations without the 
need for analysis, postulating that most classification failures amount 
to a failure in the second part of that process, not the association, 
but the organization of classes into a taxonomy.

So first I think that single word tagging is actually at a lower 
cognitive level - a simpler activity, and that the "thin client 
cognition" is what makes tags attractive in the first place, because 
they enable you to recall something without having stopped to remember it.

Second, I think that our ability to use the wetware to discard 
irrelevant results and hone in on relevant ones without thinking too 
hard about it is greater than our ability to write software that manages 
these relations for us. For example, I appreciate that when I'm looking 
for "new york" I am usually not interested in "Amy's new York Terrier", 
but sometimes I am. Being able to find all entries with the tags "san" 
and "california" is of interest to me at times. After all, and if the 
cloud that is presented to me is well-organized, it should somehow rank 
objects labeled with both terms in the right order and adjacent as 
higher than those with the two terms adjacent but in the wrong order, etc.

I think this is the tip of a much larger iceberg, actually, one that 
deserves a more careful exposition, but the gist is that we expect 
multiple word tags only because we expect tags to behave like pattern 
matches, and that the arguments against spaces in tags are a bit like 
the arguments against spaces in filenames - they spoil a natural term 
boundary in exchange for arguable gain. I submit that the human brain 
does 90% of the work when searching Google - this is why the quality of 
results differs by orders of magnitude between different users, and that 
there ought to be a sort of "dogma 95" of tagging, a set of core 
boundaries of functionality. The idea would be that making clouds that 
do more than this manifesto of functionality is likely to result in 
reduced usability. Sort of an anti-spec, or a "no use case" model.