October 2003 - Posts
For the past few months I've been an advocate for having
some sort of internal blogging guidelines for Microsoft employees.
Unfortunately, every time I start to write out my ideas, I realize that it's
a lot harder than it sounds. These guidelines wouldn't be intended
to dictate what people say, but rather to help them to make the right decisions on what is
appropriate.
My blog, as anyone who reads it is aware, is pretty non-Microsoft. I
don't like to talk about the things I do that are core to my job unless
something really compels me. Instead, most of my entries are about soda can
pyramids, silly one-liners, or steganographic codes for our once and future
overlords
kittens. I try to keep it humorous, which keeps me out of trouble, assuming I'm
careful enough not to cross into dangerous territory.
A lot of people write deeply technical entries,
observational entries, entries that comment on the state of code/tea/etc, and
even more. These are all very different kinds of blogs, so it's difficult to
come up with a single set of guidelines that guide them all. Today, however, I
noticed a post from Scoble
that sums it all up in a nice analogy:
If you're a skier, How do you make rules for skiing through the trees? I don't know how. Anyone who wants rules for skiing through the trees really shouldn't try it.
Maybe it is just that simple.
I've noticed that the root page of http://blogs.gotdotnet.com has
reverted to an older version, which indicates that I am the "Windows Client" for
Longhorn.
Yikes, I don't think I can take that kind of pressure!
My neck is sore and my ears are still ringing
from last night's Deftones concert.
I just got back from the Deftones concert and they were
awesome. Unfortunately, I now smell like pot (apparently people don't
pay attention to the hundreds of "no smoking" signs), which may not go over well
with my roommate, Shawn Nandi (product manager for ASP.NET). Then again, he's
from Canada so he probably won't even notice, eh?
Oh yeah, there was some stuff about Longhorn too, and I guess that was a
pretty good show as well. If you'd like to know more, please check
out any other blog on the Internet.
Anyway, I have to go find out where this ringing noise is coming from. At first
I just thought it was outside the concert, but then it followed me back to the
hotel, up the elevator, and into my room. I'll keep you posted if I find its
source.
One of my favorite bands, the Deftones, had a show in
Seattle on Friday night. However, since I was leaving for the PDC on Saturday, I
decided to spend the evening with my wife (aww...).
Anyway, today, as my taxi was nearing the hotel, I noticed that the Deftones
are playing here in LA on both Monday and Tuesday nights. Thanks to the loose
wireless network in my hotel (labelled "cpa", making me think it's an accounting
conference) I now have a ticket for the Monday show.
If the show is awesome (which it will be) and I'm staying through Tuesday
night (which I currently am) than I may just go back for a second night.
If you're in town and plan to go, let me know and we can share a cab
:-)
If you see me, say "banana". I was going to tell you to
say "hello", but then I realized that I wouldn't know it was really
you.
By the way, if I don't say "thank you", then you should be aware that it isn't
really me.
I was in NY yesterday and stopped into a nice little
pizza place for a few slices. It was awesome. I'll try to remember which one it
was (I'd never been to it before).
I'm watching a marathon of Chappelle's
Show on Comedy Central and it's got me thinking--what would happen if you
crossed David
Chappell with Dave Chappelle? I'm
sure it would be funny and educational.
On a side note, it's kind of odd that Dave Chappelle's web site is at
davechapelle.com (only one 'p'), davechappelle.com seems to still be
available...
I've always found artificial intelligence really interesting. I took a few
classes on it in college, and I'm embarrassed to admit that I was disappointed
when I realized there was nothing magical to it. It is, more or less, just a
series of well-defined algorithms that result in computer actions that give the
illusion of intelligence. Before studying it, I thought there was some
artificial intelligence API you could use that would tell the computer to figure
stuff out (I'm a dreamer). After the first day of class, my hopes and dreams
were crushed by the mathematical exactness of it all and I became very
disillusioned with the idea altogether, getting in a lot of debates about
whether or not "artificial intelligence" was even really possible.
To some, the idea of AI falls into the category of computing learning (such
as neural networks) where an application will get better at a task (such as
character recognition) over time based on feedback or coaching. To others, it's
any application that seems to be intelligent about the way it interacts with the
user, such as in the case of a game that makes the "correct" move given a game
state. This could be something as basic as a lookup table or a complex algorithm
that ends with the same result.
Tonight I was playing one of my favorite hockey games, NHL 2002, by
Electronic Arts (on XBox, of course). The game has amazing graphics, pretty good
dynamic commentary, and lots of television-like features that weave replays and
game themes right into the experience. Sometimes I like to just watch the game
play by itself. Each version of NHL hockey gets better and better.
Unfortunately, there is one apparent weakness that doesn't make as much
improvement in each round--the AI. I play against the computer on the "normal"
level and find that it's very clever about a lot things, knowing when to chase
the puck and when to cover a position, when to shoot and when to pass, etc. Once
you get used to it, however, there are a lot of recurring patterns a player can
recognize to take advantage of the AI. To counter this, the game seems to favor
the AI players, who always seem to move faster, hit harder, shoot better, and
never get tired. I tried an experiment by making a custom team with half average
players and half "perfect" players, and the perfect players were able to evenly
match the AI players for about half the game, whereas the average players were
barely able to get out of their own zone.
This all reminds me of older games on less powerful systems, where the
difference between "normal" and "hard" was simply a function of doubling the
number of adversaries on the screen, making them move twice as fast, or making
the levels twice as long. The other alternative has been to make the application
explore more potential moves, such as the chess games that are excruciatingly
slow when played on high levels. It's kind of disappointing that we haven't
gotten much further in this field. When you look at the advances made in
graphics and other elements of gameplay, it's a wonder what the future of AI
will be like.
Or maybe I would rather have the excuse of "oh man, the game cheated!"
instead of admitting I can't keep up with the games of today's kids :-)
More Posts
Next page »