Tuesday, May 13, 2003


Automated Fun.

Making fun of people who take themselves too seriously. Ernie Svenson: Is pretty easy to do, but coming up with a fresh method is tough.  Don Novello's Lazlo Letters employed the technique of contrived letters from a rabidly adoring fan sent to celebrities and political figures.  Now we have Ali G, whose ruse is to obtain interviews with serious people--who believe him to be another cloying member of the mainstream press--and then proceeding to ask them innane questions.  Maureen Dowd apparently liked his interview of James Baker. [Sam Ruby]

Some of you may remember a famous flame war that existed in the space strategy games newsgroups -- the Derek Smart Flame War.  Some of the posts that flew back and forth in that one were hilarious -- just absolutely tortured in their use of language, circular reasoning, and crazy pull-apart details...

One of the better afternoons I spent (in terms of self-amusement) was grabbing a pile of messages that were written by this one particular guy, Bill Huffman, who was one of the biggest "detractors" of Smart's.  I don't want to take sides in the dispute, but Huffman was pretty hilarious to read (and he didn't intend to be).  I pulled apart a pile of the messages and built a template-driven "Bill Huffbot" message creator program.  Ah, what fun. 

Message #1.

But you know the fun had to end.  I am just such a weenie.  Bill Huffman wrote this, and I felt so guilty and mean I just stopped.  Ok, maybe I did it once more. 

So just remember -- you may think it's fun to make fun of someone, and then you do it, and then maybe you're gonna feel all guilty about it. 

See?  I have a conscience.


4:46:07 PM    

Tuples -> Values.

another awesome bit of software technology is Linda, the tuple spaces architecture by David Gelernter. like Lisp, it provides a simple set of primitives with great expressive power. but i don't think that the Linda approach scales well. the remove() primitive requires some degree of locking that is not suited for distributed environments. i don't think that JavaSpaces does Linda justice - it could have been implemented far better. i built a Linda prototype on top of my last company's Voyager product that was more in the spirit of the original vision. [graham glass: what's next?]

Tuplespaces are pretty cool...everybody who implements one ends up extending it to do batch operations, etc...I did.  The thing is, I am not so sure that a simple tuple is the be-all and end-all...I have been toying with value spaces, which are essentially the same thing.  A tuple is just another kind of value you can put into the space.  The same value cannot exist more than once in the space.  In effect, it becomes a big set.

Once you have that type division, you can do a lot of interesting things to partition the data, route it based on type and/or attributes, and so forth...

The key thing in a distributed knowledge system is an effective partitioning system.


4:30:07 PM