Six degrees of vampire smut

See the blog post

Post your lowest numbers here.

zombies>anthology>american>history>archaeology>mesoamerica

It's invalid, we're going from Mesoamerica to Zombies, not the other way around and anthology -> Zombies does not work.

Alas. You can do anthology > horror > werewolves > zombies, however. Making it:

mesoamerica > archaeology > history > american > anthology > horror > werewolves > zombies

8 steps. I didn't solve challenge 1 or 2 very well, but challenge 3 can be done in VERY few steps.

Whoops on the Mesoamerica>Zombies. How 'bout this for tarzan>iceland?

tarzan>africa>europe>middle ages>vikings>iceland

In the meantime:
mesoameric > Latin America > Caribbean > Haiti > zombies

With one less link. Turns out "horror" and "egypt > mummy" are dead links...

8: vampire smut > vampires > historical fiction > history > European history > socialism > fascism > Spanish Civil War

6: Tarzan > adventure > historical fiction > Middle Ages > Vikings > Iceland

6: Mesoamerica > mythology > fantasy > horror > werewolves > zombies

I tie eromstead for 6 on Tarzan > Iceland:

tarzan > apes > africa > exploration > polar > iceland

Standings:

Challenge 1: Kezinge in eight.
Challenge 2: Tim and Eromstead tied in six.
Challenge 3: Circeus in five.

#2 in 5 (which sounds like a good working minimum for obscure pairs, IMHO)

Tarzan > historical fantasy > mythology > norse > Iceland

Arg! I was just about to post that! I'll bet it can be beat, however.

Now 7: vampire smut > vampires > historical fiction > Italy > Italian history > fascism > Spanish Civil War

Nice one Kezinge! I have only managed 9 so far...

*searches*

Vampire smut > mystery > England > France > Spain

YAY!

So that's 6 with Spanish Civil War, circeus?

That's six. You gotta get to Spanish Civil War. Good job, though.

There are many, many variations at 6 for Vampire Smut>Spanish Civil War. Here's another:

vampire smut>vampires>history>europe>spain>spanish civil war

@ Kez and tim

Oups... I knew I was forgetting something...

Just got a five anyway:

vampire smut > strong women > historical mystery > Spain > spanish civil war

Observe how there is no mystery > historicalmystery link.

mesoamerica > science fiction > horror > werewolves > zombies

5!

The "rules" fail to explicitly state that you cannot add tags to your own books to create a new path.

The rules also do not state that you are forbidden to write a book about Mesoamerican Zombies and wait until it gets tagged.

Unfortunately, the tags aren't updated fast enough for your technique to work...

1: Circeus in 5.
vampire smut > strong women > historical mystery > Spain > spanish civil war

2. Circeus in 5.
Tarzan > historical fantasy > mythology > norse > Iceland

3: Stellar dust in 5.
mesoamerica > science fiction > horror > werewolves > zombies

I thought I had 5 first in #2?

Didn't I say you did? Confused.

Well, tag me with "idiot", I meant #3 (cf. messages #6 & 8)

edited because I first typed 2 again...

Shoot. Circeus is in the lead on all three. I KNOW that they can be improved upon. Don't ask me how i know.

Tarzan > Fantasy > mythology > Norse > Iceland (still five, still working)

Thanks Tim, for yet another wonderful procrastination tool! :-)

Another Tarzan in 5:

Tarzan > africa > exploration > polar > iceland

Another 5: Tarzan > Africa > Kenya > Denmark > Iceland

Mesoamerica > writing systems > fortean > zombies.

how strange.

And how about

tarzan > Essential > useful > iceland

?

w00t! Two records down, one to go. =)

PS... is it cheating that I wrote a program to find these? hehe.

Oh, it is *SO* cheating!

Tell us about it.

PS: And--hey--don't tie up my servers recapitulating the traveling salesman problem. :)

It didn't seem to be having much of an effect on the servers. Sorry if it did. I noticed things seemed to be slow around 5:30, but I wasn't running the program then. I won't run it again if it's a problem.

Traveling salesman problem? Please, it was just breadth-first search. =)

fun challenge (and great site!), btw. =)

No, just don't hit it more than 1/sec. I'm interested in how you did it. What language, etc.?

Challenge number 1: Vampire smut to Spanish Civil War in 10 steps. I'll bet someone can do a lot better.

Vampire Smut>vampire>Ann Rice>New Orleans>Travel>Spain>Spanish Civil War

7

Tim: OK, I will throttle it back a bit before running it again (I think at the moment it probably does more like 2/sec). I wrote it in Perl, since with regular expressions it requires minimal work to extract the "related tags" info. Basically, I give it a starting and an ending tag, and then it does a breadth-first search from the starting tag -- for each tag, it looks up the tag page on LT, extracts the names of all the related tags, and shoves them all in the end of a queue. Then it takes the next tag out of the front of the queue and does it again. As it goes along it remembers which tags it's visited (so it doesn't revisit any), and how it got there, so when it finds the tag it's looking for it can reconstruct the path. It took about five or ten minutes of run time to find each of the solutions I posted above (which are, of course, guaranteed to be the best possible).

Random question: how many unique tags are there on LT? I'm sure it would be too many to analyze all of them remotely at a speed of only 1/sec, but it might be fun to modify my program to analyze the link structure between the top, say, 1000 most frequent tags, and produce a nice graphic. Or something.

In theory there are 314,490 different tags. But there aren't that many used in the related-tags algorithm. Because relatedness involves examining two-tag overlaps AGAINST a tag's overall statistics (total count and user count), it uses a summary table, periodically regenerated. The summary table has a cut-off; a tags needs like 5 users and 20 instances--I can't remember the specifics. Of these there are 17,525. Neither includes the tag combinations, which make a big difference in the clouds. The actual number of combinations is modest (2,900), but a lot of big tags are affected.

I'd be glad to provide the data in API form. Do you have any experience making such graphics?

@ Byorgey
Nice work on the mesoamerica one, but methink "useful" and "essential" both are of the sort of tags we are supposedly not to use.

@ all
And you know what's fascinating? As LT grows, some of these will eventually come to change and not be valid anymore, or receive shorter solutions.

vampire smut > mystery > England > Europe > Spain >Spanish Civil War (still 6)

Mesoamerica > writing systems > fortean > zombies

was actually the one I knew. I was playing around and was amazed at the connection. The others I didn't know before-hand.

I say we do a new challenge once or twice a week. I'll get "related authors" working better. That might be interesting, since it would require more understanding and there are fewer 'bridge' authors than 'bridge' tags. I wouldn't be surprised if you just could get from Ann Coulter to Aeschylus.

vampire smut > strong women > historical mystery > spain > spanish civil war (5)

Whoops, sorry Circeus I see you beat me to that one :-)

Tim: I do not have experience making such graphics, but I do know of a few Java libraries for graph data structures/visualization etc. that I've been wanting to try out. Do *you* have experience with such things? Also, the "related authors" game sounds fun. But how would you determine (from an algorithmic point of view) which authors are related?

Circeus: you're right about my tarzan > iceland path. I was just excited that my program worked. =) I kind of doubt there is an "interesting" solution in only four steps for that one, although I don't know for sure.

Also, my program just spit out the same solution that jbd1 gave above for vampire smut > spanish civil war, which is proof (assuming my program is correct) that it cannot be done in less than 5 steps, hence proving that circeus is the winner on that one. =)

You can do similar things in your own library alternating tags and books. There are sometimes surprises, but never the absolute strangeness that mixing different folksonomies gives.

I briefly considered a bookpile entry based on a related idea. [* * * * * * * * * * * * * *] But I am just too hopeless with a camera and the necessary lighting to make all the spines legible.

I got a five on #1, also:
vampire smut>series>historical>europe>spanish civil war

Oooh, Good. I'm amazed "Europea" goes right to "Spanish civil war."