Holy closeness, Batman!
11.15.06 9:45 PM


By Churchillio
TopCoder Member

Before round two began, an even more epic battle was taking place in the dining hall. Four souls locked in battle, trading wheat and wood for ultimate glory. Myself, tomek, Abednego and Multifarious were playing Settlers of Catan (skip this paragraph if you've never played it). Having laid his settlement first, tomek took an early lead with plenty of wood and brick to lay his 'roads' to victory. He also had a 6, which yielded him triple Ore, which kept getting rolled. Myself and Abednego bided our time while tomek and Multifarious were struggling over the longest road card. Multifarious had a quick 3 cities with his abundance of ore, but I had a more sinister plan. I ended up with the largest army, and on one turn I managed to buy two development cards, which were luckily two victory points. With tomek and Multifarious each one point from victory, two lucky last rolls netted me 8 sheep, and I was able to trade in all 8 + an ore for a card which was a victory point, and the win! Anyways, having proved my might, on to the competition.

At 3:30 pm the competitors all started setting up their macros and editors. misof pulled a neat trick where he finished early and started to walk around checking out other people's macros on the public terminals. It was really funny when I turned to see who was reading misof's screen and it was actually him.

The coding began, and after a while Im2Good, mathijs and krijgertje each submitted problem 1 within 2 points of each other. mathijs was however the only person to immediately open the next problem and start working, the other two stayed and looked at the solution for a few minutes before moving on. Soon after, krijgertje opens problem 2 and 4 people each submit problem 1 within the span of a minute or so.

Problem 2 has the room buzzing with 'I wish I was in this round' as it proves to be a fairly easy solution. Some quick math is done, it is found to be brute-forcible and many outside competitors have already solved it. misof, FatSimon and bjoeris however opt to open the 1000 point question first while the others work away on the 500. andrewzta has the first problem 2 submission, and with an hour and one minute left the 1000 point problem seems well in reach, but he ends up waiting a few moments before he opens the 1000 point problem.

mathijs submits the 500 and moves on to the 1000 right away without any checks, at about the same time krijgertje submits the 500 pointer and the last person submits the 250 point problem. At this point everyone has solved the first problem, and 4 have solved the second.

Many minutes go by and a crowd starts to gather over misof's monitor since he is still working on the 1000 point problem. There is also a crowd around OpenGL's monitor at this point, who is currently in 5th place. With 34:44 left, mathijs submits problem 3 and is 630 points ahead of 2nd place. There are only andrewzta and Multifarious left who can realistically overtake the lead, but their chances quickly tick away and first place is guaranteed to mathijs if all his problems are correct.

With about a half hour left, misof submits problem 3 and moves into 2nd place. This is insanely close. misof is now 600 points away from first place, with the 500 point question still unopened. With 25 minutes left, andrewzta submits problem 3 and for some reason his total score does not update. His final score should be around 1250, but it still says 600.

misof is now in 7th place and people once again gather to his monitor. There are some comments on his solution to problem 2 with some people saying he was wasting time with unneeded checks for certain domino configurations. With 15 minutes left kia submits problem 3 and moves into 3rd place behind andrewzta with 1182 points, just 2 points above krijgertje's 1180.96. misof then submits problem 2 and moves into 5th place. andrewzta's score has been removed from the board temporarily in order to fix his final score bug. When he reappears he should still be in 2nd place.

Time passes slowly as people gather around the big scoreboard. With only 4 minutes left krijgertje resubmits his 1000 point question and moves down to 9th place from 3rd. Third through 10 place are now separated by only 200 points, with 3rd through 8th place being only 80 points apart. You can feel the tension in the room as the players start to gather for the break before the challenge phase. Eighty points, for a live semifinal -- that's nuts!

The break fixes andrewzta's score, and so the pre-challenge scoreboard now looks like this:



The challenge phase has rundown goes something like this, with the times signifying how much time is left in the 10 minute challenge phase:

9:12 gevak successful challenge
8:55 Macsy unsuccessful
8:40 FatSimon unsuccessful


At this point I'm sitting Indian style on the floor up in front, actually nervous about the top 8 people. Just look at the scoreboard again... crazy.

4:10 gevak makes two unsuccessful challenges in a row.
Im2Good makes a crucial mistake! He unsuccessfully challenges and goes from 6th to 8th place.
3:40 Mike Mirzayanov unsuccessful.
2:50 krijgertje success.
2:00 Im2Good has another unsuccessful challenge, although it doesn't change his position.

There were 3 other challenges before time ran out by FatSimon and gevak, but none of them changed any of the standings. After the challenge phase the only real difference in the scoreboard was Im2Good's fall from 6th to 8th place. Everyone was now either sitting or standing as close as they could to the front stage with the results monitors. Those were probably the longest 10 minutes some of these guys can remember. Any system test failure (besides mathijs' 250 pointer) would knock someone out of the top 6 spots. The system tests rolled up one by one, and as you now know by the results page, only OpenGL had a failure, bringin him from 7th place to 8th place behind Im2Good, whose failed challenge knocked him one spot short of a wildcard spot. I didn't think that the first round could have been topped for suspense, but this one did a damn good job trying.

Here's a video of the system tests and final results: