This morning in the second game of the latest big “Human Champion vs Computer” match current World Chess Champion Vladimir Kramnik did not only lose but was actually checkmated by Fritz. It was a one move blunder. Kramnick made a mistake, then wham, checkmate. That is just plain not supposed to happen to a world champion. Games at this level almost always end with someone resigning, or with an agreed draw. An actual end of the game due to a mate or a by the rules draw is very very rare. So it was kind of cool to see happen. Well… I missed the exact moment because I was on a conference call with Europe, but I saw right before and right after…

Anyway, two games into this match it is dissapointing to me even though the computer is ahead 1.5 to 0.5 at the moment. Cause the first game was a draw, and apperantly some people think Kramnik should have been able to pull out a win. And this second game was again definately a draw, with some people thinking Kramnik should be able to pull out a win… before he made his stupid move.

It has been years now since Deep Blue first scored a point against Kasparov. Now, DB was a custom built computer, whereas the program now competing with world champions can be purchased by anybody and runs on commodity windows machinces. So I guess there has been improvement and all, but I am looking for the point where the world’s best machine can wipe the floor with the world champion the same way that the world champion human could wipe the floor with a five year old who just learned the rules. So far that just isn’t happening.

It is looking more and more like the humans will remain good enough to be able to at least manage a draw if they don’t make big mistakes. That is kind of disappointing. Well, we’ll see how it goes given a few more years.

And there are a few more games left in this match. I have been getting up early to watch them online. Maybe there will be more excitement in the coming games. The actual checkmate and all is cool, but it would have been nicer if the computer had outplayed the human all game long instead of there just being a stupid mistake at the end.

Also… just to repeat a tirade I make every time I watch one of these human/computer matches… traditions and respect for your opponant and all that nonsense aside… a computer in a human computer match should NEVER resign and should NEVER accept a draw. And this game shows exactly why. The human can make a stupid mistake. The computers should accept the human resigning, but otherwise should force the game to go to the bitter end no matter what… either a checkmate or a draw by the rules (Stalemate, Insufficent Material, Threefold Repitition, Fifty Move Rule). Because up until that very last moment, the human could always screw up and let the computer back in the game, even if the computer was losing previously. Computers should NEVER resign and NEVER accept a draw offer. Period.

My house has been invaded by almost ten 6th grade girls for a “Birthday Slumber Party”. A few minutes ago they were doing Karaoke. Now they are running around the house screaming and chasing each other. This will be a fun evening. I just brought the ice cream cake and pizza back. There will be food soon. Later there will be movies on the projector.

I haven’t updated this in quite a long time. Since November of 2004 online actually. (Although I added a couple of data points in between without pushing the changes up to my site.) So it has been awhile. So I dropped the lines between the data points cause they meant nothing.

But you can see that the number of State Department Warnings is higher than any other point I have (although I can’t say for sure it wasn’t higher during the gaps).

Since I last took a data point in January 2005 Libya is safe to go to again, but Sri Lanka, East Timor, Chad, Phillipines, Eritrea, and Uzbekistan have been added to the list.

So I’ll try to avoid any vacations to those places at the moment.

A couple weeks ago I downloaded Nethack for old times sake. I played one game then but don’t remember much. I just played a game a few minutes ago. Here is the entire transcript:

Hello Abulsme, welcome to NetHack!
You are a chaotic female orcish Wizard.

You displaced your kitten.
m – a puce potion.
The kitten falls into a pit!
The kitten hisses!
You frightened the kitten!
You fall into a pit!
You are still in a pit.
The kitten bites!
The kitten bites!
You die…

Goodbye Abulsme the Wizard…
You died in The Dungeons of Doom
on dungeon level 1 with 0 points,
and 0 pieces of gold, after 7 moves.
You were level 1 with a maximum of
11 hit points when you died.

Ah, good times.

This last evening I’ve had the opposite problem. I fell asleep at 3:30 UTC (early for me) but then work up at 8:30 UTC (middle of the night). Then I was up two hours. Then tried to get back to sleep another two hours. I only succeeded in lying in bed tossing and turning. Then just a few minutes ago I decided that even though I would really like a couple more hours of sleep and feel quite tired, that given my failure to fall back asleep and the fact the sun is now up, I should just go ahead and get up even though normally it would be another two (maybe three) hours until I would be stumbling out of bed. Ideally at this time of year given my work schedule, I’d sleep from about 6 UTC to 14 UTC each day. So I am getting up, although I resent it. Yawn!!!

But usually, I have the opposite problem. Hitting snooze for hours and not getting up when I should. People keep coming up with good ideas for alarm clocks to combat this sort of thing, but I haven’t seen any of them actually AVAILABLE TO BUY yet. And this one is no different. Just another design prototype. Come on, somebody please sell one of these?

Anyway, I love the idea.

Alarm clock that won’t give up
Stephen McGinty, Scotsman

The wall-mounted alarm clock can be switched off only when its user climbs out of bed, stands directly in front and repeats, by pressing coloured buttons, a sequence generated randomly each morning. If the user fails to repeat the sequence swiftly, the alarm will continue to blare until the task is completed correctly.

(via Digg)