5
\$\begingroup\$

In contest for Weekly Featured Image of September 5th, two top entries were tied with equal upvote counts (when including late votes). Given the number of quality of images in last weeks, I suspect some week we might run into a tie again.

Currently, the rules of contest say nothing on handling ties. From chat transcript, it seems that in this case, winning image was chosen randomly, not disclosing what was "randomness" based on - a coin-flip, random number generator, or personal preference?

Since the image should be chosen by community, not by randomness, perhaps we should have a rule on how to resolve ties in case two or more images have the same number of upvotes at voting deadline.

What rule would be fair and feasible to check against?

\$\endgroup\$
5
  • \$\begingroup\$ If the community voted close enough for a tie, I don't think it matters how random the randomness was, IMO. But we do need to come up with something. Normally, I've been saving a vote to cast at the end (but before the official time) to avoid the issue, but I was on holiday. \$\endgroup\$
    – rfusca
    Commented Sep 6, 2011 at 16:02
  • \$\begingroup\$ Using personal preference as source of randomness would give the moderator an extra vote, thus not making it community chosen any more. \$\endgroup\$
    – Imre
    Commented Sep 6, 2011 at 16:04
  • 2
    \$\begingroup\$ Well, don't forget the moderator is part of the community and is entitled to their vote. I really think we're making a mountain out of a molehill here in regards to the current situation...nothing was malicious. A tie indicates that the community is equally distributed between the two... We can come up with a solid solution to a tie, but lets not make this into some bigger issue than it is. \$\endgroup\$
    – rfusca
    Commented Sep 6, 2011 at 16:09
  • \$\begingroup\$ @rfusca of course the moderator can use his votes during the voting period, just like everybody else; I was talking about possibility to choose to accept a tied entry he might personally prefer, thus giving unique privilege of extra "voting" after voting period. I'm not saying there was anything malicious going on - I believe there wasn't, but I do think we could have a more transparent way of handling such cases in future and I'm seeking if the community feels that way too. If you have a good solution, please feel free to post an answer. \$\endgroup\$
    – Imre
    Commented Sep 6, 2011 at 16:28
  • \$\begingroup\$ As far as randomness goes, its however StackExchange chooses to sort same-rank answers. As far as I understand, StackExchange uses a SQL Server backend, and SQL Server does not make any effort to produce deterministic results in the absence of any additional qualifying ORDER BY factors. We basically get the order that SQL Server gives us, and that tends to change from one page load to the next. That said, from what @rfusca said, when he accepted the winning entry, there was no tie. \$\endgroup\$
    – jrista Mod
    Commented Sep 6, 2011 at 20:28

4 Answers 4

Reset to default
6
\$\begingroup\$

I'd prefer, for simplicity sake, that a mod just save a vote to act as a tie breaker. If they go to put it up and there's a tie, they simply cast their vote for the one they like. That way there's nothing random and its still ultimately the community that decides (since the mod is part of the community). Nothing complex and nothing that should change voting or submitting habits.

\$\endgroup\$
3
  • \$\begingroup\$ If the moderators agree that they can be decisive voters between eventual top entries in exchange of giving up early voting, then this could be a solution indeed. I know you're not going to like it, but an early vote has more weight since people tend to like more what other people seem to like (Cialdini's "Influence" explains it nicely), so early voters have the power of giving others hints on what should reach the top at all. \$\endgroup\$
    – Imre
    Commented Sep 6, 2011 at 19:24
  • \$\begingroup\$ @Imre - Sure, I'm well aware of the 'follow the crowd' syndrome. But its not something to promote if we can help it. Natural Early voting may have more influence, but we're not officially giving it more weight. \$\endgroup\$
    – rfusca
    Commented Sep 6, 2011 at 19:45
  • 2
    \$\begingroup\$ As far as a mod reserving their vote, I can certainly do that. Its not a big deal for me. As for making the contest more fair from a voting standpoint, we have some ideas for that, such as having two phases: submission phase and voting phase. Start one thread a week early to "preload" the content...votes on any entries in this thread are meaningless. Start a second thread, a mod adds all the submitted entries from the submission phase (including links to the profile of each submitter) and close the topic. Answers in closed topics can still be voted on, but no more answers can be added. \$\endgroup\$
    – jrista Mod
    Commented Sep 6, 2011 at 20:33
6
\$\begingroup\$

I'm a little embarrassed to see what a hot potato this has become after my photo was pipped to the post this week. Let me add a few comments of my own:

  1. Let's not forget it's just a fun competition on a website: no prize money involved, just a little glow of pride for the winner. Let's keep it fun!

  2. Thanks to @rfusca and others for doing all this in their spare time. I used to run a big Flickr group and it was great fun until people started arguing and complaining about all sorts of trivialities, and seemed to assume it was my full-time job to mediate. So I have a lot of admiration for people who get stuck in and organise stuff for others. It can be a thankless task!

  3. Personally I'd be happy with the moderator casting a deciding vote, or going on the time of the last vote, or tossing a coin - whatever's easiest for the mods.

\$\endgroup\$
2
  • \$\begingroup\$ Thanks :) I agree, we need to not blow anything out of proportion. \$\endgroup\$
    – rfusca
    Commented Sep 7, 2011 at 0:09
  • \$\begingroup\$ As a side note, don't forget there is a Photo.SE Flickr group for any entries :) \$\endgroup\$
    – Rowland Shaw Mod
    Commented Sep 7, 2011 at 18:54
2
\$\begingroup\$

My own idea:

In case of a tie, the image that received its last vote earlier, is declared winner.

The time of receiving votes on a post can be checked from author's reputation page.

This follows the logic used in auctions: an earlier bid is a stronger bid and has to be trumped, not merely called, to win. Similarly, a vote cast early shows slightly stronger support than a (hesitated?) vote cast later, and must be trumped to win.

UPDATE: since a user's reputation history is no longer visible on Meta, this solution is no longer possible.

\$\endgroup\$
4
  • \$\begingroup\$ I'd rather not encouraging early voting like that - it may create an issue around early entries that we currently don't have. \$\endgroup\$
    – rfusca
    Commented Sep 6, 2011 at 15:59
  • \$\begingroup\$ @rfusca edited to make it more clear - the last vote received would count. Entries added in the last couple of days are already known to have low chances of winning. \$\endgroup\$
    – Imre
    Commented Sep 6, 2011 at 16:13
  • \$\begingroup\$ I understood the last vote. I just don't think the time frame a vote was placed in should change the weight that it holds. \$\endgroup\$
    – rfusca
    Commented Sep 6, 2011 at 16:20
  • \$\begingroup\$ This was the solution I thought of too. And +1 for the poker analogy! :) \$\endgroup\$
    – Mark Whitaker
    Commented Sep 6, 2011 at 20:40
0
\$\begingroup\$

I think we should simply declare them both winners and run them for the next two weeks, with the order selected at random or by Imre's method.

The next contest can be either open for voting longer, delayed in start, or run as normal but with a week delay before the winner is shown. (That would then allow a vacation week or something later sometime.)

\$\endgroup\$
1
  • \$\begingroup\$ Well, a tie either means the top images are both great, or they are both equally uninspiring. Just because we lucked out this time with two great ones doesn't mean we will always want to draw the results out for two weeks ;-) \$\endgroup\$
    – Sean
    Commented Sep 7, 2011 at 3:25

You must log in to answer this question.

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