@carykh
  @carykh
carykh | $1,000 Coding Tournament Announcement! @carykh | Uploaded 3 years ago | Updated 12 minutes ago
Original deadline: May 26th, 2021 at 10 PM UTC
Deadline to re-submit, only if you already filled out the Resubmission-Request-Form (closed now): May 28th, 2021 at 10 PM UTC

Spoiler: Simon won!

Link: htwins.net/prisoners-dilemma
The GitHub issues section is a good place to general discussion and more FAQs about this project: github.com/carykh/PrisonersDilemmaTournament/issues

FAQ:
Q: Help! I already submitted a strategy, and want to edit it, but Google Forms won't let me.
A: Oops! I had planned for you to be able to edit your submission (both the file and the questionnaire) until the deadline, but Google Forms doesn't allow resubmitting files. So, if you'd like to resubmit, fill out this form, and I'll delete your first response: https://forms.gle/g2Nf4D7Wa4cCKXZ59 (CLOSED)

Q: Can I work with a friend on my submission for your Prisoner's Dilemma Tournament?
A: Yes! Working with other people is allowed.

Q: Can I submit multiple strategies?
A: No. For this tournament, only one strategy per person is allowed. (As a soft precaution, my Google submission Form only allows for one submission per Gmail account.)

Q: What if people flood the pool with dumb strategies that boost their own true strategy?
A: That should be prevented by the one-strategy-per-person rule. I suppose a competitor could get their friends to submit dummy responses to boost the one team-leader, but I don't think it's likely this will affect the overall pool, since there are 500 strategies at the this point

Q: Can I still submit strategies if I know a lot about Prisoner's Dilemma or I looked through all the source code?
A: Yes! The competition is set up so that even with perfect knowledge of the game, your performance still depends on others' submissions, which you can't know perfectly. So, since there's no "golden ticket to victory", you can submit strategies!

Q: What about colluding/handshake strategies? (Those use an arbitrary code of actions at the start to detect others with the same mindset, and cooperate amongst themselves and defect against the rest.)
A: Colluding strategies aren't allowed. While inspecting, if I see strategies putting arbitrary codes to check for friendly opponents, then I'll disqualify those strategies. I admit this rule is a bit hard to rigorously define. Just assume that there's no shared information between you and opponents.

Q: How many people have submitted so far?
A: As of right now, (May 21, 2021 11:26 AM UTC), between 400-500 have been entered. This is good, because it means that you probably don't have to worry about any single strategy causing your downfall!

Q: Won't randomness affect how many strategies perform? Will that decide the winner?
A: Thanks to advice from Frederikja, I will run the tournament many times (100+) and average the results. This should smooth out randomness, and only award winners who can consistently score well!

Q: Can I use a programming language other than Python?
A: Sorry, but for this tournament, you have to use Python. If the tournament-organizing code and the strategy code is all in the same language, it's easier for me to make sure everyone's playing by the rules.

Q: Can I use AI libraries?
A: I have to say "no", because I'm not fully aware of what every library is capable of. Obviously, if a library can connect to the internet, I won't run it. But, I don't want to have to check through each libraries' capabilities just to make sure a strategy is ok. (numpy and random are all you really need to code any rational agent, so that should be good enough!)

Q: How many rounds is each 1v1 game?
A: 200 - 1,765. It starts with a guaranteed 200 rounds, and then every turn after that, there is a roughly equal chance of it ending on each turn (so you can't strategize by knowing you're on the last turn.)

Q: What if people submit exactly the same strategy?
A: They'll all still be entered into the roster separately.

Q: Can strategies read and write the RNG seed?
A: No. Any reading/writing/tampering with random-number generation is kinda against the spirit of the competition! You should assume the other player is like a black box, even if they do use RNG to determine their behavior. So, don't it. (It is a clever idea, though.)

Q: How long should it take for strategies' code to run?
A: Don't do anything that deliberately slows runtime. As a rough ballpark, try to have each round run in under 10 ms. But, if you're using if-then statements and numpy arrays, I don't think speed should be a problem for you!

Q: Can I read my opponent's strategy files?
A: Sorry, nope!

(I'm typing a FAQ in this video discussion is more helpful than a stream, because then you can read it and find your exact question more easily, that way!)
lazykh channel: youtube.com/channel/UCobNHsxWdWDFCuPqHoBmyUw
$1,000 Coding Tournament Announcement!Evolution Simulator of Creatures Doing MATH (Part 1)Air. Earth. Fire. Water. ...and 27,221 others. (Elemental 3 data visualization)My worst-graphics game gets a long-overdue upgrade!Can AI edit Twitch VODs?Algicosathlon Day 15The Amazing Marble Race - Season 3, Leg 1BAGUETTE 6BTen Words of Wisdom 13A🐍 My homemade game console has snakes in it! 🐍3 neural nets battle to produce the best jazz musicDo cubed cubes exist?

$1,000 Coding Tournament Announcement! @carykh

SHARE TO X SHARE TO REDDIT SHARE TO FACEBOOK WALLPAPER