4

What is the minimum number of days that is needed to complete the StackEgg game? (It's on the right if anyone didn't notice.) There are four markers (Questions, Answers, Users, Quality) I believe each marker decays according to a function depending on its value and the other markers, but I am unable to figure out the relation just by playing. I know at least that the Answers marker decreases very fast if the Questions marker is low, and so the optimal strategy might be to keep Asking as long as nothing else is dangerously low and Questions is low. Other than that, I guess that all markers decrease much faster if any of the others are low, and so the optimal strategy might be to Answer/Upvote/Downvote according to which is lowest. The problem is that experiments are difficult to carry out because of the low resolution of the markers and the voting system. Markers are capped at their maximum value. Flag is unnecessary and Nothing is a red herring (but many people waste many days because they vote for Nothing!). I suspect using Close is also not optimal. And Traffic seems to be needed to keep the other markers from falling too fast, but I can't figure out how to raise Traffic at the start. Anyone have any ideas?

Notes & Updates

Beta can be completed in 16 days. I think Graduation can be done in 100 days but there was some interruption at the last part so it took 111 days instead.

Winning took 326 days after some more interruption, but using zero flags. The leaderboard shows that Theoretical Computer Science Stack Exchange did it in less than 259. Perhaps they used flags. I still can't tell the trick to finish this stage efficiently without flags.

The strategy I used for the most part is to keep Questions at 3/4 hearts and equally improve the rest (always pick the least) until they catch up. Then for the Graduation stage equally improve all, keeping track of the order in which we do so because the poor resolution does not show which one is really the least. For the Winning stage it seems that one should also keep Answers at 4/4 and Quality at 3/4 for Traffic to go up, which is necessary to get past that point. If we don't want to keep all the flags I think they can be used to get over that point or finish the last part much faster. It is probably bad to use Close and certainly not good to do Nothing.

Most of the time I am half-guessing whether I have reached an impasse when it seems to cycle too many times, before I try a slightly different method to find a way out. But there must be a more systematic way to analyze such black-box processes. And I'm curious to know eventually what the process details are.

StackEgg is gone

See https://meta.stackexchange.com/questions/252451/we-liked-stackegg-let-the-fun-continue. The leaderboard is still at https://math.stackexchange.com/stackegg/leaderboard. I just found a question like mine at https://meta.stackexchange.com/questions/252439/whats-the-shortest-time-for-a-completed-stackegg.

Still, the question of the optimal algorithm remains, though I haven't looked at the code. Someone wrote a bot to play the game, as mentioned at http://chat.stackexchange.com/rooms/240/the-nineteenth-byte. His code is at http://pastebin.com/N5A74eEM.

user21820
  • 53,491
  • 7
  • 84
  • 231
  • It says "Many answers and high quality" attract traffic. – Arpan Apr 01 '15 at 07:08
  • @ArpanBanerjee: Yes I know. Do you mean that it is the rate of increase in Traffic that is correlated to both, rather than directly? – user21820 Apr 01 '15 at 07:09
  • @ArpanBanerjee: And maybe Close is a red herring too... – user21820 Apr 01 '15 at 07:10
  • I haven't got to the late stages of the game yet, so can't comment much. – Arpan Apr 01 '15 at 07:14
  • @ArpanBanerjee: Are you the other one playing now? – user21820 Apr 01 '15 at 07:18
  • Yep, let's talk strategy. Trying to go for max answers and quality to see what happens with traffic. – Arpan Apr 01 '15 at 07:19
  • @ArpanBanerjee: Eh someone used the flag. I saw that some people on the leaderboard didn't use any flag that's why I said it's unnecessary. – user21820 Apr 01 '15 at 07:20
  • Let us [continue this discussion in chat](http://chat.stackexchange.com/rooms/22486/discussion-between-arpan-banerjee-and-user21820). – Arpan Apr 01 '15 at 07:21
  • @ArpanBanerjee: Someone restarted... Okay I'm not playing anymore... – user21820 Apr 01 '15 at 07:30
  • It was going so good. I'm not playing either. – Arpan Apr 01 '15 at 07:31
  • I was trying to do some measure but it seems there is a bot doing upvote and downvote to perform some sabotage. :/ However, if you want to do some tests, you can click on the progress bar and do a inspect element (on firefox), or you can look in the source code at the egg-phase-progress-bar element. The width will give you the percentage with a 0.1% precision, so that you can track your progress, and see what is the best strategy. – gvo Apr 01 '15 at 16:44
  • @gvo: There is no bot; it is just some other user. Haha I don't think noting the progress bar value gives much information though since we don't even know the relation between it and the markers. – user21820 Apr 02 '15 at 01:33
  • @user21820 : A user which does only know how to click on upvote and downvote, and which doesn't know what to do when the button is not here (flag or restart vote)? Very strange user. Or it is (was) a bot. Having the value is precisely useful to identify the relation between it and the markers, you can't do an algorithm when you don't know how to formulate the problem. – gvo Apr 02 '15 at 08:30
  • @gvo: Not all users may realize that their actions have deterministic results. They may think that they are supposed to select what they do on the site usually! The problem with using the progress bar to gauge efficiency is that its growth seems to be according to another strange nonlinear dependence on the markers. Anyway I didn't play anymore since my last edit so I don't know haha.. – user21820 Apr 02 '15 at 12:37

0 Answers0