7
\$\begingroup\$

I've had two challenge ideas floating around in my head for a while. I'll spare the details but as a general overview:

The first is a challenge that should, in principle, be fun but in practice it is so easy for cops to make solutions that are virtually impossible to crack that it couldn't be fun in practice. Similar to how it is quite easy to make a riddle that no one can crack, but hard riddles are not good.

The other is not really even a cops and robbers question, but would involve collecting OEIS sequences (and some other information) submitted by users and then having a competition based on the submitted sequences. There can't rally be a winning criterion for OEIS sequences.

Here there seems to be a common thread. I would like to post a two thread (cops and robbers style) question that is competitive and does have a winning criterion but only for one thread. In the second case the goal of the non-scored thread would be to maximize your own score on the scored thread, but the first is in even murkier water where the ostensibly the only goal of one thread is to make the other thread "fun" (whatever that means).1

Now perhaps each scenario should be treated separately, but they are similar so I thought I'd put them as one meta question.

Can a two thread challenge have one thread without a winning criterion? If so what restrictions should we put in place on such challenge types?

I have thought of making it a pop-con and robbers but I really, really don't like pop-cons for a number of reasons, and I'd rather not ask the question than ruin it by making it a pop-con.

\$\endgroup\$
  • \$\begingroup\$ It's hard to say for sure without knowing the details of your challenge. Why not post it in the sandbox? \$\endgroup\$ – DJMcMayhem Jul 26 '17 at 17:31
  • 2
    \$\begingroup\$ @DJMcMayhem I thought it might be a good idea to tackle it on the meta in a more general case so that if/when I post such a challenge we've already gone through the debate. I'll sandbox them later today. \$\endgroup\$ – Wheat Wizard Jul 26 '17 at 17:33
  • 2
    \$\begingroup\$ I would be interested in resolving this without any particular challenge in mind. I also have a sandboxed C&R challenge that I'm wondering how (and if) to score for cops, and I'd like to hear general thoughts. \$\endgroup\$ – xnor Jul 27 '17 at 0:28
  • 2
    \$\begingroup\$ "Collecting OEIS sequences ... and then having a competition based on the submitted sequences" -- isn't this like the time capsule questions, where people submitted characters and then there were challenges based on the complete list? Could a similar methodology work for your idea? \$\endgroup\$ – DLosc Jul 27 '17 at 1:09
  • \$\begingroup\$ @DLosc Thanks I hadn't thought about the time capsule. That is pretty much exactly what I wanted to do. I'll look into restructuring into that format. \$\endgroup\$ – Wheat Wizard Jul 27 '17 at 1:26
  • \$\begingroup\$ @xnor I looked up your sandboxed C&R (wow! you have even more pending sandbox posts than me...). Pinging you in case my answer here might be of use for your challenge. \$\endgroup\$ – trichoplax Jul 28 '17 at 12:53
  • \$\begingroup\$ If there is no scoring for cops, what's the incentive for cops to post answers? \$\endgroup\$ – Luis Mendo Aug 6 '17 at 22:25
  • \$\begingroup\$ @LuisMendo In the first of the two scenarios, it is just for fun, no real incentives exist. On the second on the incentive for cops is that it helps them score better as robbers if they post good cops. \$\endgroup\$ – Wheat Wizard Aug 6 '17 at 22:32
4
\$\begingroup\$

Tailoring the incentives

Not a direct answer to the title question, but a potential solution for your first example case.

If you want cops to have an incentive to produce difficult problems, but also an incentive to avoid problems that will never be solved, you could design the scoring mechanism with this in mind, rather than not scoring at all. This would avoid the grey area problem with banning cryptographic hash functions (what counts as one...?). The scoring mechanism itself will give cops an incentive to avoid things that can't be cracked.

There may be many ways of including this incentive in the scoring mechanism, that suit different types of challenge.

Example

A cop's score is the length of time between posting the cop answer and having a valid robber answer posted.

Now if a cop misjudges this and posts something that never gets cracked, they score zero.


This might even apply to in a similar way to your second example case. If the cops thread is effectively a list of golf challenges, you could score in a way that gives an incentive to find sequences that are harder to compress. For example: A cop's score is the length of the shortest corresponding robber answer. So robbers try to win by posting the shortest code for each cop, and cops try to win by posting sequences that require more bytes of code than other cops' sequences.

There isn't a concept of "uncrackable" in this case (you can phrase the challenge spec to ensure all sequences are well defined and can be described in code), but you could still put an upper limit on code length if you wish. For example: A cop scores zero until the shortest corresponding robber answer is less than 100 bytes (or 1000 bytes, or whatever suits your challenge). This gives cops an incentive to be difficult but not too impractical.

Loopholes

The scoring mechanisms may need some refinement to avoid unwanted incentives and the possibility of loopholes. It might take a few test challenges to iron these out.

For example: a cop might be able to inflate their score by looking at the timezones/frequently active charts for the users in the robbers thread, and posting when fewer of them will be online. This might be mitigated by making the score the floor of the number of whole 24 hour periods, starting at the cop posting time. There's a similar potential problem with weekends too.

\$\endgroup\$
  • \$\begingroup\$ I'm not sure the first solution would work. Robbers could make their solution take an extremely long time to crack (e.g. because it requires brute-forcing), which would guarantee them a high score. \$\endgroup\$ – Esolanging Fruit Aug 4 '17 at 4:50
  • \$\begingroup\$ @Challenger5 Here I'm using "cop" for the post intended to resist cracking, and "robber" for the post that cracks it. I believe that's consistent with the tag wiki. Yes, the idea is to make cracks take as long as possible. Posting something that requires brute force risks never having a crack, and scoring zero. This is still a far from perfect scoring mechanism, and someone could still post something that can definitely be cracked but takes months, which would probably be less fun for most people. \$\endgroup\$ – trichoplax Aug 4 '17 at 10:22
  • \$\begingroup\$ Individual challenges may make compromises to try to mitigate this, for example setting a time limit after which a cop scores zero regardless (almost the opposite of the current approach of cops scoring by becoming "safe"). \$\endgroup\$ – trichoplax Aug 4 '17 at 10:24
  • \$\begingroup\$ On your first point, that's right; I intended to write "cops". On your second point: posting anything risks "never having a crack and scoring zero." As long as the robbers have an incentive to crack submissions, there's the chance somebody will take the time to do it. The problem is that this method doesn't distinguish between a genuinely challenging and fun submission and a time-consuming brute-force submission. \$\endgroup\$ – Esolanging Fruit Aug 5 '17 at 3:36
  • \$\begingroup\$ The second idea about code-length makes a lot of sense, but the burden of proof should be on the cops (for example, you could require the cops to provide a hash of their solution, to be revealed once a valid robber's solution is found). The other problem is that robbers might refrain from posting a crack if it's above the code-length limit, which there are many good reasons against. \$\endgroup\$ – Esolanging Fruit Aug 5 '17 at 6:07
  • \$\begingroup\$ I fully agree that a scoring mechanism can't completely exclude boring cops, but providing an incentive still seems useful. \$\endgroup\$ – trichoplax Aug 5 '17 at 10:00
  • \$\begingroup\$ I also agree with that meta answer about minimum scores, but it applies to language agnostic code golf challenges. I think for a cop answer in a single language, it would be less of a problem, especially if each cop chooses the limit. However, note that the code length limit is an afterthought ("if you wish") and isn't part of the main idea of the suggested scoring mechanism. \$\endgroup\$ – trichoplax Aug 5 '17 at 10:08
  • \$\begingroup\$ The last line of that meta answer points out that it is rude to say "if you are not good enough, you cannot compete". I definitely agree with this. In this case, however, it is not intended to exclude robbers, but to prevent cops from being unreasonable. If it turns out to come across as exclusionary then a different approach will need to be found. \$\endgroup\$ – trichoplax Aug 5 '17 at 10:20

You must log in to answer this question.

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