The kind of sad club

[MagnumPU]

My ants sure know how to collect food, but the hungry little guys like to pass up unguarded hills as well. They weren't very smart to begin with, there wasn't much room for mistakes. Anyone who made detrimental last minute changes, or forgot to remove debugging code (my bot is going to timeout on the big maps because of one line of code), feel free to respond. It was fun, but ugh, hard to watch. Come on guys, get that hill, just by accident why dontchya?

[MagnumPU]

Or don't respond actually, it might be bad for future employment prospects. Just read that other post about jobs. I have some good suggestions for interview questions actually, especially for a phone interview: "Do you have a big mustache, and have you ever done the google ai challenge?" A good follow up so that you don't get the wrong guy, "Do you think using email for source control is a good idea, or have you ever done that?"

[BenJackson]

[WildUtah]

1. Make a useful seeming but actually detrimental change the day of the end of the contest
2. Notice the problem and submit an improved version of the change an hour before the contest
3. New update doesn't compile on servers (difference between gcc 4.6 and 4.5?)
4. Revert to old version and include the major change that was useful by backporting and submit again.
5. Notice in last minute test that you've broken advanced new distant food/symmetry algorithm in backport
6. With five minutes left, submit the very final update for the food part.
7. Browser crashes in the midst of upload.
8. Force quit, force quit, kill -9.
9. Sorry, submissions closed.
10. ???
11. Profit.

[franperuzzi]

Heck, I'll probably implement a formal release schedule next year.

[amstan]

[salar2k]

It is like you are releasing a software in a very hurry situation.
I've been in that situation and i had a schedule for this contest.
I had a better and fast algorithm based on matrix but it didn't finished until 1 day before the contest, I stopped working on that because I knew it won't be finished and safe!

Last minute big changes are such a big risk in software release cycle, so take this as a lesson.

[DrClaes]

Count me in! I've dismissed my timeouts during the contest to unspecified "java" or "server" problems... then, after timing out in my first finals game, I found a piece of code that's left over since my first attempts at exploration which I've clearly marked as "too slow, fix later" but then forgotten all about and even stopped using the result of... and when I fixed it (2 minutes worth of profiling and three lines of code commented out), it no longer times out in those situations when I run the maps locally ... but too late to resubmit, of course...

Lesson learned, I guess!

[pkmiec]

Yes ... rushing features, no matter how seemingly simple, before a deadline is risky (especially without an automated test suite). I stopped development couple days ahead of the deadline ... but the temptation was there.

[MagnumPU]

To anyone who feels the way I do, I think we'll do better next time. And to Ben, it's a hard lesson but thank you pointing out the silver lining. And to those guys who got in under the wire, I hope their final was gold. And to those with better sense than me, I'll get there. See you guys next time!