5
\$\begingroup\$

I posted a King of the Hill challenge recently, but I'm not sure if this answer is in the spirit of KoTH challenges. It chooses a random number at the start of the challenge and repeats that number throughout the challenge. This means that each time I run the game it changes the playing field drastically while other random bots average their results over time.

I would probably ban this type of bot if it didn't seem like a narrow arbitrary ban. I'm also relatively new to the site so I'm not sure what to expect.

What is the community consensus about this sort of bot?

\$\endgroup\$
4
  • 1
    \$\begingroup\$ You can ban random bots, by requiring bots to be deterministic, but it might be a little unfair to do that so late in the competition \$\endgroup\$ May 1 at 10:23
  • 3
    \$\begingroup\$ Your challenge seems to be one of the few where this can happen - I haven't been able to find any others. Also, is it really that bad? It just means half the time you have an extra Crb, and the other half an extra "Game Theory is stupid anyway". \$\endgroup\$
    – emanresu A
    May 1 at 10:35
  • \$\begingroup\$ @Ausername I ran both cases, producing extremely different standings. If it's an extra Crb the 7eleven wins consistently. If not, it's between AverageAverage, Histogrammer and IQ. \$\endgroup\$ May 1 at 10:53
  • \$\begingroup\$ @cairdcoinheringaahing I like a bit of randomness, as long as the results can average out over time. However, that's quite a subjective measurement and I'm not sure how to make it objective enough that it can be a rule in future challenges. \$\endgroup\$ May 1 at 11:16
9
\$\begingroup\$

Ban generally, not specifically

I'm very much opposed to the idea that "OP has final say" (I.e. if the post author dislikes an answer, or doesn't think it's ok, they get to disqualify it)\${}^*\$. Answers should be objectively valid or invalid by anyone who has read the challenge rules.

Therefore, I suggest that bans be more general than specific. You can ban random bots in by requiring them to be deterministic. Beyond that, you can ban specific, objective behaviour (e.g. may only use uniform random distributions), but not specific answers, unless they break the rules.

Unfortunately, this is an all-too-common reality of challenge writing on CGCC. There are very often answers which completely screw up the challenge, but which are perfectly valid. Typically, they are hyper optimised solutions, or clever tricks found in the mathematics. Here, however, the randomness has affected it in unexpected ways. Really, the only advice I can give is that randomness is often compensated for by running a lot of rounds. Try running a sample competition and see whether 1000 rounds "fixes" the extremes produced by Rude Random.


\${}^*\$ And I have, and do, vote to close challenges as "opinion-based" if the OP tries to claim that they have final say

\$\endgroup\$
4
  • \$\begingroup\$ Running a lot of rounds doesn't fix it. It chooses one number at the start and repeats it. I like a lot of the other random bots because they randomise each round, but I'm not sure how to objectively ban this category of submissions. Also, this isn't for this challenge, but probably the next one. \$\endgroup\$ May 1 at 11:53
  • 1
    \$\begingroup\$ @EnderShadow8 This is why most KotHs consist of "games" and "rounds", where each game resets everything \$\endgroup\$ May 1 at 16:30
  • \$\begingroup\$ That was something I was considering. Glad to hear someone else confirm that it's a good idea. \$\endgroup\$ May 2 at 0:11
  • 3
    \$\begingroup\$ I don't think "may only use uniform random distributions" would be meaningful. All RNGs I know obtain non-uniform distributions by post-processing uniformly distributed outputs. \$\endgroup\$ May 2 at 17:55
4
\$\begingroup\$

The end of the competition is drawing near and I'd say that ad hoc changes like this would make the KoTH challenge even more unfair. Within the parameters of your challenge, results can be very inconsistent and if you don't want the inconsistency, you have to think of the possible solutions people would submit and take them into account while setting the rules for your KoTH challenge. This would require a lot of thought—simply banning the Rude Random bot wouldn't cut it.

\$\endgroup\$
7
  • \$\begingroup\$ I don't think changing rules this late is possible either, I was really asking for what the community thinks of submissions in this vein. \$\endgroup\$ May 1 at 11:14
  • 1
    \$\begingroup\$ I can't speak for the whole community but I think the submission is perfectly normal. In competitive programming, problem setters have to think of possible submissions and design the test cases so that wrong submissions wouldn't be able to produce the correct outputs for the test suite. Test suites that don't fail wrong solutions are called weak. Rude Random is playing within the rules, and if Rude Random makes the results inconsistent, that's your challenge's fault, not the bot's fault. \$\endgroup\$ May 1 at 11:37
  • \$\begingroup\$ How would I create an objective rule to ban this sort of bot? "Must average in the long term" isn't very objective. \$\endgroup\$ May 1 at 11:38
  • \$\begingroup\$ It also means I can run the game several times to pick the result I want, which, while I'm probably not going to do that, isn't fair on everyone participating either. \$\endgroup\$ May 1 at 11:40
  • \$\begingroup\$ I can't think of an objective rule right now 😂 Because it's extremely hard to do so and maybe requires some lateral thinking. Anyway, I think the challenge is perfectly fine in its current form and I wouldn't mind the inconsistency at all. \$\endgroup\$ May 1 at 11:47
  • \$\begingroup\$ While the results might not be perfectly fair, this is your first challenge and you shouldn't worry too much about things you couldn't get right on your first attempt. \$\endgroup\$ May 1 at 11:48
  • 1
    \$\begingroup\$ I should probably take the upvotes and stop complaining lmao \$\endgroup\$ May 1 at 11:54
0
\$\begingroup\$

As long as a bot follow/fulfill the rule/tasks, it’s valid

I agree that changing rule late isn’t the best solution for edgy bot like rude random, instead, all restriction should be clear at the beginning.

Option: Let randomness come in

I don’t fully agree with banning extreme bots, just like normal coding challenges, there is always edge case(game) that you code should still suitable for.

For example, there could possibly have a bot that get high score for both when Rude goes 1 and when Rude goes 100, just like a code that suits large test and empty data. There is always randomness in most of Koth, but the interesting part is how to solve those inconsistent.

In the end, there is half of games with 1, other half 100. Then there is no favor to any bots, so the bot which loss less on both case wins, suits Koth.

Thi is only my option, the final choice is up to you

\$\endgroup\$

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .