Commons talk:Sound Logo Vote

Please direct future discussion to the main Sound Logo talk page on Meta-Wiki.

Voting process not accessible with a screen reader

It is impossible to reorder the entries when using a screen reader. KaraLG84 (talk) 12:14, 6 December 2022 (UTC)

Hi @KaraLG84, really sorry about that. While we look into what's technically possible, we would be more than happy to enter your preferences manually. Please email us at soundlogowikimedia.org at your earliest convenience before December 19. Thanks for your interest in the sound logo project and sorry about this glitch. MPourzaki (WMF) (talk) 15:14, 6 December 2022 (UTC)

I can confirm this is the case. I'm using JAWS and Kara is using NVDA, both extremely popular Windows screen readers, and it's kinda impossible. I can think of an idea for an alternative interface for us (which some sighted users might prefer) of a text box that only accepts exactly ten letters, each one being the first letter of the entry name like ABD ... etc. I'll email you my choices Mehrdad. Graham87 (talk) 15:18, 6 December 2022 (UTC)

Well-received @Graham87, appreciate it. MPourzaki (WMF) (talk) 00:09, 7 December 2022 (UTC)

I've just e-mailed my votes. KaraLG84 (talk) 14:31, 12 December 2022 (UTC)

2 entries violate the contest rules

the rules were clear: no more than 4 seconds. 2 entries are 5 seconds. Those are AC54 and PK62. EMGmus (talk) 16:52, 6 December 2022 (UTC)

As an initial screener for the sound logo contest, I can explain this one ... we were asked to be relatively lenient when applying this rule and if the sound was a bit over four seconds but met all the other criteria it was to be permissible. Also, we weren't supposed to count silence. Graham87 (talk) 17:10, 6 December 2022 (UTC)

There was nothing about "being lenient" or "not counting silence" in the rules. It's very difficult to create a sound to exactly to such spec and unfair that you chose to disregard your own criteria. In essence, two of the people who broke the publicly posted rules made it to voting. And somewhere, two who followed your rules did not. EMGmus (talk) 17:23, 6 December 2022 (UTC)

Hi @EMGmus, thanks for this message. I'm sorry if you felt the criteria were unfair. The sound logos were scored and reviewed based on the submission criteria and not length. That's the lenience Graham87 is mentioning. Early on we realized that some submissions would have a bit of silence before or after, and we accepted them as not everyone had access to professional audio editing tools. As you can hear, for example, AC54 and PK62 have silence afterwards and the actual sound logo is not longer than 4 seconds. Thanks for participating in our global contest. MPourzaki (WMF) (talk) 18:22, 6 December 2022 (UTC)

SUBMISSION CRITERIA - contrary to what you say, included length.

"Every original sound logo submission should be at least 1 second and no more than 4 seconds long."

See below.

https://soundlogo.wikimedia.org/#:~:text=Submission%20criteria,CC0%2Fpublic%20domain%20sounds).

You accepted two submission which did not meet that criteria.

There was nothing unfair about said criteria - what's unfair is accepting submissions that didn't follow your own guidelines. EMGmus (talk) 20:18, 6 December 2022 (UTC)

I hear you @EMGmus and the actual sound logo component of the said files is 4 seconds. MPourzaki (WMF) (talk) 23:57, 6 December 2022 (UTC)

The reverb tail goes clear to the end of the file - there is no actual silence. Intentionally or not - you are again telling an untruth here. The SUBMISSION criteria for both files was arbitrarily ignored - twice. In effect, this is an insult to every entrant who scrupulously followed the guidelines you set out in order to be eligible for voting. Perhaps the rules should have stated "4 seconds or thereabouts", or "4 seconds but we'll be lenient", or "don't count the reverb tail", but they didn't. They clearly state "at least 1 second and NO MORE than 4 seconds long." You have wasted the time of thousands of people who in good faith submitted work to you. I won't participate further in this charade. EMGmus (talk) 00:16, 7 December 2022 (UTC)

What you also need to know is that the player rounds UP. So AC54 is actually 4.1 secs (which I think is well within the margin's that Graham87's is speaking about, and PK62 is even 4.0 something something seconds. Their file description pages show this more clearly. —TheDJ (talk • contribs) 09:53, 7 December 2022 (UTC)

It’s nanoseconds indiscernible to most human ears. Tha logo that fits best wins.

What’s your intention to get 2 entries disqualified? Are you a finalist and feeling insecure? 2600:1017:B815:A220:2552:F8D4:DADA:44CE 15:29, 15 February 2023 (UTC)

My! Reading comprehension isn't your strong point clearly, so I'll repeat:

the rules were clear: no more than 4 seconds. 2 entries are 5 seconds.

and

There was nothing about "being lenient" or "not counting silence" in the rules. It's very difficult to create a sound to exactly to such spec and unfair that you chose to disregard your own criteria. In essence, two of the people who broke the publicly posted rules made it to voting. And somewhere, two who followed your rules did not.

SUBMISSION CRITERIA - contrary to what you say, included length.

"Every original sound logo submission should be at least 1 second and no more than 4 seconds long."

See below.

https://soundlogo.wikimedia.org/#:~:text=Submission%20criteria,CC0%2Fpublic%20domain%20sounds).

You accepted two submission which did not meet that criteria.

There was nothing unfair about said criteria - what's unfair is accepting submissions that didn't follow your own guidelines.

The reverb tail goes clear to the end of the file - there is no actual silence. Intentionally or not - you are again telling an untruth here. The SUBMISSION criteria for both files was arbitrarily ignored - twice. In effect, this is an insult to every entrant who scrupulously followed the guidelines you set out in order to be eligible for voting. Perhaps the rules should have stated "4 seconds or thereabouts", or "4 seconds but we'll be lenient", or "don't count the reverb tail", but they didn't. They clearly state "at least 1 second and NO MORE than 4 seconds long." You have wasted the time of thousands of people who in good faith submitted work to you.

OK? Let us know if you have any trouble with the more difficult words in the above or altrmately, find an adult to read it for you. 70.59.145.52 20:17, 15 February 2023 (UTC)

not be blocked on any project?

Woow - for most things Wikimedia you can vote and do other things with 1 or even if blocked on 2 projects but not on Sound Logo? That is a very interesting restriction, considering that new anonymous account person with 11 edits can. Let me wild guess, that this rule was done by someone who has authority to do (b)locks? --Zblace (talk) 17:40, 6 December 2022 (UTC)

Hey @Zblace, is this a personal concern or a general inquiry? If you find that this rule is impeding your participation, please email us and we can look into your individual case. MPourzaki (WMF) (talk) 18:23, 6 December 2022 (UTC)

@MPourzaki (WMF) both, but that is irrelevant. why should Sound Logo vote-right be harder then Board Elections in this respect? --Zblace (talk) 19:06, 6 December 2022 (UTC)

@MPourzaki (WMF) I've raised this off-wiki, but I'll do so here as well, and I did so on the same underlying argument - that having a stricter methodology than the BOT vote is bizarre and unwise. The rules should be specifically amended - an appreciable number of individuals will see the criteria and just leave - they won't turn onto the talk page and they won't know to ask for an exemption. Nosebagbear (talk) 19:15, 6 December 2022 (UTC)

Thanks @Zblace, thanks @Nosebagbear for raising this. I will discuss with the team. MPourzaki (WMF) (talk) 20:08, 6 December 2022 (UTC)

@Nosebagbear excellent points. What is even worse, this bad Vote rule decision is not implemented well technically, so I could vote despite my *(controversial) block on HR Wikipedia.

@MPourzaki (WMF) you should talk to your team if this was even well executed technically, also why are votes public, as this leaves space for manipulation... Personally I do not think you should use preferential vote system, if you wanna get people express aesthetic opinions in which option to like just 1 or 2 option should be possible to make. Anyway seems my preference is in the leading part with 25% of votes for it as a first option. --Zblace (talk) 23:59, 6 December 2022 (UTC)

Hey @Zblace, "if this was even well executed" ... why the tone, why the cynicism? Votes were made public particularly because some community members asked for it for legitimacy. Otherwise we could have a disputed winner because people wouldn't be able to verify for themselves, a Wikimedian preference :) In terms of being blocked on one vs. any project. @Nosebagbear do you have other instances were being blocked on just one project was acceptable? I'm not referring to Board elections, not a similar context. Would appreciate a frame of reference here. Thank you. Let's work on this together and improve it. MPourzaki (WMF) (talk) 16:44, 7 December 2022 (UTC)

@MPourzaki (WMF) pardon my non native English if the word 'even' sounds cynical, that was not my intention. Disproportion of aspiration with rigid access and not really executing that rigid rule for me sounds problematic (as does keeping votes public during vote process), but not to the point to insult...it is just bad choices and executions IMHO. --Zblace (talk) 21:30, 7 December 2022 (UTC)

@MPourzaki (WMF) - firstly, to clarify my position, I don't think the process has been handled poorly - other logo votes have been done publicly. I find it "odd", but not inherently flawed.

But to answer your specific question An extremely recent counterpart in the Wikifunctions/Abstract logo specifically used that basis. So that alone I hope would make that point. But I'd also like you to expand on your reasoning - obviously this is a different context to the BOT votes. But even if you do change the rules, I would like to know why this context would necessitate stricter eligibility requirements than the methodology for picking our Board members? Nosebagbear (talk) 23:03, 7 December 2022 (UTC)

@MPourzaki (WMF) Sorry for barging in but I also agree that’s an unreasonable restriction. Some wikis are (ok, at least one is) known for banning, say, political critics. Not be blocked on any project would indirectly serve political interests that might not be aligned with Wikimedia’s mandate. — Al12si (talk) 10:36, 8 December 2022 (UTC)

Thank you @Zblace, @Nosebagbear, and @Al12si for your input, guidance, care, and for your interest to participate. As promised, we discussed this internally and are happy to revert to movement precedent in this case. We have changed the ineligibility criteria to being blocked on "more than one project" instead of any. Thanks again. And Zblace, I'm also not a native English speaker and like you, I'm hoping for a great sound logo for the projects, maybe even better than non-terrible :-P MPourzaki (WMF) (talk) 17:21, 8 December 2022 (UTC)

Can't vote on mobile devices

I just run in a circle, clicking the "sound logo contest voting page" link on top of this page here and the "vote" button on the other page before I finally found the actual "start voting" button at the very bottom. Unfortunately the button doesn't do anything on mobile. Did I miss the explanation what to do instead?

Besides I'm more disappointed than expected. Most snippets really sound like somebody tried hard to copy a well known startup sound from an operating system or gaming console. I'm curious if you have a backup plan in case the final result can't be used? TMg (talk) 17:58, 6 December 2022 (UTC)

@TMg indeed - this was not super innovative, but I think 2 are good enough basis for fairly good result. Hopefully sound logos can evolve as did visual ones in next iterations. --Zblace (talk) 19:09, 6 December 2022 (UTC)

I will fix the mobile voting ASAP. Amir (talk) 21:11, 6 December 2022 (UTC)

Me neither... MiguelAlanCS (talk) 23:51, 6 December 2022 (UTC)

Hi TMg, hi MiguelAlanCS have you tried again since? It seems that only certain mobile devices have this issue. I just tried to vote on mobile and it worked for me. What operating system are you using? MPourzaki (WMF) (talk) 20:43, 7 December 2022 (UTC)

@MPourzaki (WMF) Brave is detecting the voting script/voting site as a tracker (I just hit the same problem on my desktop), thus blocking it by default. Many modern mobile browsers are similar, with aggressive integrated ad blockers. I’m suspecting this is the reason things are not working on mobile. — Al12si (talk) 10:40, 8 December 2022 (UTC)

The browser is just Chrome on Android. Clicking the vote button works now. I see a loading bar and then the list of tags I'm supposed to order. Unfortunately this doesn't work. No matter what I do (tap, hold and drag), they don't move. TMg (talk) 19:17, 13 December 2022 (UTC)

In case you wonder, I can vote on desktop, so this isn't blocking me personally. Unfortunately I'm unable to vote for another reason. I would like to pick my top 2, but nothing else. The rest is just so weird I really don't know how to order them. But it looks like the process doesn't allow this. TMg (talk) 19:33, 13 December 2022 (UTC)

I tried to vote from my mobile device and it mistakely sent what was on the screen before I could change it! It doent work! Orlysi (talk) 20:25, 19 December 2022 (UTC)

Orlysi, I see you've updated your vote since this comment, were you able to add your intended vote? Best, RAdimer-WMF (talk) 23:57, 19 December 2022 (UTC)

Ringtone of the century!

Kevin MacLeod's Patreon Ringtones

i like this one more! 🤣--RZuo (talk) 18:09, 6 December 2022 (UTC)

i'm curious about one thing. how's the vote done on this page? i've never seen this before.--RZuo (talk) 18:09, 6 December 2022 (UTC)

Hey @RZuo, instead of a simple majority wins approach, we wanted something more concrete, a way to see the popularity of the submissions and their relations to each other. We are using the Schulz method and asking voters to arrange their choices from most to least favourite. MPourzaki (WMF) (talk) 20:44, 6 December 2022 (UTC)

@RZuo, I believe the 2020 MediaWiki and the Abstract Wikipedia/Wikifunctions visual logo contests employed similar methodology. MPourzaki (WMF) (talk) 00:11, 7 December 2022 (UTC)

i'm curious about where the votes are stored and how the voting part is "embedded" on the page? RZuo (talk) 00:15, 7 December 2022 (UTC)

Personal opinion from someone that doesn't know music theory (ignore me)

• AC54: great, simple melody, but if heard many times can get annoying + originality is not high (I think I heard that progression many times)
• BY23: I like the concept but I cannot take the quiz show idea from my head. Also, while I like the idea of a "child laughing" to make Wiki fun, the fact that it is the last sound heard make me feel it is not serious enough (a child's play)
• DS71: Great concept of "opening to knowledge" (probably the most original one), but I think bad execution- the knocking on the door is used sometimes to prank other people and can get annoying if heard multiple times with headsets. It should be subtler.
• FM76: Nice, but too simple. I think an ascending arpeggio is ok, but it has to give me something else than just that. I like the ending.
• GX13: This is great, but the mixing is not right, the samples for the typewriter sound wrong to me (I don't like the ending).
• JW08: The idea is good, but not enough differentiation between voices. The ending is confusing to me.
• OZ85: This is the most original to me, but what it evokes on me is "people talking", then moving on by clicking away, which I don't like as a concept
• PK62: I like this one, it reminds me of the videogame civilization, when discovering a new technology. My only issues is that it lacks warmth to be heard many times.
• UN03: Short and joining typewriter with traditional strings, I like it, but I wonder if it is too "much"
• VQ97: This one separates the sound effect (book pages being read) with and ascending tone, and the melody. I like the concept (we just heard an article so we close the book), but the melody at the end needs more work (maybe integrating it more with the end of the first part and make it shorter), but it is very distinguishable

The reason that I wanted to write this is that some of the sounds have a great idea but poor execution, while others have a poor idea but great execution, and just a vote doesn't give the whole context. --jynus (talk) 08:44, 7 December 2022 (UTC)

Hi @Jynus, thanks so much for taking the time to share your reflection here. We wanted to make the participation in the voting relatively easy and accessible, and as always, we welcome and appreciate a more detailed feedback that captures the individual nuances so well. Thanks for participating. MPourzaki (WMF) (talk) 14:40, 7 December 2022 (UTC)

You almost begin to sound like Fred Armisen in that Portlandia sketch with that repeated criticism of the endings . Daniel Case (talk) 06:26, 18 December 2022 (UTC)

Personal opinion from User:Fuzheado

I have some opinions on the candidates that may be useful for folks.

Background: I've done graphic/identity design in the past for various entities. The logo I designed for English Wikipedia was chosen for commemorating 5 million articles and has been used as a template for many other Wikipedia editions. en:Wikipedia:Requests_for_comment/5_millionth_article_logo. While I've never worked with sound design for audio logos, I have extensive experience as an audio/video editor for journalism and creative storytelling. I also have a musical background in orchestral music. So I find this process really interesting in bridging different spaces and skillsets.

TL;DR: To me, there are only two strong candidates for the audio logo. I'd rank them as a clear #1 and #2, as they are the only ones that tell a strong story, both along the lines of, "Search... discovery... aha!":

#1 - VQ97 - Page turning sound, ending in five strong notes, almost like a power-up sound in a video game. The five notes at the end sync perfectly with saying "Wi-Ki-Pee-Dee-Ah." This candidate also has the ability to be tightened up so that it is even shorter, or the five-note riff can be used on its own. The nice thing is that there are different possibilities.

#2 - PK62 - Bassy drumbeats, then four strong rising notes (though it may sound like three). As with number 1, there is a story, almost like we are journeying, and then discover an answer, which is highlighted by the final high note, like a light bulb being illuminated.

These two are the strongest as they are simple, iconic, and delight the listener. Many of the other candidates have sounds that can be mistaken as environmental sounds (people talking, knocking, keyboard clicks), which can grate on folks not just upon first listen but certainly when played 100 more times on a voice assistant or mobile device. Without disparaging the creators or the sounds, some of the other candidates are too busy for the intended use - too many textures, too many instruments, or too much complexity. That doesn't make them bad sounds! But for this particular use as a stinger/logo, they are nonideal. (If there is interest, I can post a full evaluation matrix of all the sounds I jotted down.).

As a comment on the voting system - I wish I could vote for two, and only two, and indicate that I would not find the others suitable.

Hope this is useful. Thanks for all the hard work by the creators, judges, and intermediaries. - Fuzheado (talk) 20:16, 7 December 2022 (UTC)

Thanks for this great feedback and reflection @Fuzheado, appreciate it. I'll share my favourites once the voting has closed :) I noticed that you haven't submitted your vote officially. Realistically, beyond the top 3 choices, there wouldn't be much relevant statistical significance for our contest. Can I encourage you to choose a third and cast your vote as well? Thanks again for taking the time to share your opinion. MPourzaki (WMF) (talk) 21:30, 7 December 2022 (UTC)

I'd be interested in your evaluation! I actually took the time to collect some entries from the random submission batches that I felt had some aesthetic potential (unfortunately they don't seem to be available individually here on Commons, so I could only reference them as timestamps in the batch audio files). This process is fascinating to me as well. Waldyrious (talk) 20:13, 9 December 2022 (UTC)

Which voting method will be used with the rankings provided?

There are many possible ways to tally ranked ballots, from various Condorcet methods, Borda count to instant-runoff. Which method is going to be used here? This is not made clear.

I really hope it's not IRV, as it's mathematically not a method for finding consensus, but only useful in competitive conflict resolution! I believe using such a method would be a disservice to the project. — LucasVB | LucasVB_Wikipedia | Talk 01:03, 8 December 2022 (UTC)

Hi @LucasVB, we are using the Schulze method. What do you think? We'll add a reference to the main vote page to make it clear. Thanks for flagging that. MPourzaki (WMF) (talk) 02:21, 8 December 2022 (UTC)

Great! Thank you. Schulze is a great choice, as it's a robust Condorcet method, therefore favoring the consensus. — LucasVB | LucasVB_Wikipedia | Talk 02:45, 8 December 2022 (UTC)

Voting not possible on some translations

I tried pressing the Start Voting button again and discovered what problem I actually ran into (my ad blocker is off right now, otherwise voting doesn’t work at all): The voting button doesn’t work on some translated pages. It works fine in English, and for some reason also in French; but it doesn’t work in Japanese, Cantonese, Malay, or Farsi. I have not checked the other translations.

The Submit Vote button also doesn’t look translatable — not that it’s a big issue given the Start Voting button doesn’t even work for most translations...

ETA: Given the fact that voting isn’t working on translated pages, I propose that voting should be paused until all translations are ready, then we make sure voting works on all translated pages, and only then should voting resume. — Al12si (talk) 13:00, 8 December 2022 (UTC)

Thanks a lot @Al12si. Should be fixed now. MPourzaki (WMF) (talk) 23:37, 8 December 2022 (UTC)

Voting errors

*[[User:Jdlrobson]]: OZ85 - UN03 -  - VQ97 - GX13 - AC54 - FM76 - BY23 - PK62 - DS71 - JW08

*[[User:Babureslem]]: UN03 - AC54 - FM76 - DS71 - BY23 - ÖZ85 - JW08 - PK62 - VQ97 - GX13

*[[User:Hassan Hassoon]]: FM76 - UN03 - JW08 - DS71 - VQ97 - GX13 - PK62 - AC54 - BY23 - أوز 85

Nardog (talk) 14:22, 8 December 2022 (UTC)

Thanks for monitoring the vote page @Nardog, more than 550 votes on just day 3. Don't worry, we'll do an extensive clean up of the votes and results, as well as a check for voter eligibility before we announce the results. Appreciate your diligence. MPourzaki (WMF) (talk) 17:09, 8 December 2022 (UTC)

Bug on revotes

The JS seems to handle well voting again (e.g. maybe someone clicks twice submit or changes opinion), but it breaks the formatting. E.g.: https://commons.wikimedia.org/w/index.php?title=Commons:Sound_Logo_Vote/Votes&diff=next&oldid=713515877 The issue seems it could be easyly solved by an extra \n at the end? --jynus (talk) 21:58, 8 December 2022 (UTC) Edit: CC @Ladsgroup: , who solved it manually before I did. --jynus (talk) 22:00, 8 December 2022 (UTC)

@Jynus I think I fixed the revote bug in Special:Diff/713724636 Amir (talk) 16:42, 9 December 2022 (UTC)

terrible

Besides I'm more disappointed than expected. Most snippets really sound like somebody tried hard to copy a well known startup sound from an operating system or gaming console. I'm curious if you have a backup plan in case the final result can't be used? TMg (Diskussion) 17:58, 6 December 2022 (UTC) Yes, it's my feeling, too. And according to my perception, most of the tones are much too high, especially too abrasive, too sharp, and too aggressive. Is this supposed to be an alarm? Tozina (talk) 22:23, 8 December 2022 (UTC)

Where to vote?

The button at the bottom doesn't work. Is any plain link available? Prüm (talk) 19:28, 9 December 2022 (UTC) The error in my browser reads "NS_ERROR_DOM_QUOTA_REACHED". What now? --Prüm (talk) 20:18, 9 December 2022 (UTC)

Hmm, we haven't seen that error before, sorry about that. We'll look into it @Prüm. Try refreshing your browser or using a different one. If it's just not working out for you, no worries, you can email us your preferences at soundlogowikimedia.org. All the best. MPourzaki (WMF) (talk) 21:32, 9 December 2022 (UTC)

Thanks, I voted by email. --Prüm (talk) 22:07, 9 December 2022 (UTC)

Some unexpected things

1. I read the propositions and listened to the sounds, and found and ranked two of them that I both liked and felt suitable for the task; about others I had no preferrence.
2. I read through about Shulze metod of voting, and thought such choice as I described above (for me firtst one preferred, next second ranked, all other about equal), should work with Shulze metodfo.
3. for a while click on voting button was followed by message that I should log in or create an account, but I already was logged in.
4. now I could vote, but not as I wanted (to express two or more samples are equivalent to me)

I relistened all samples several times and found I can live with all of them, liked one better, but that choosing a up to 4 second sound for a representation of all human knowledge that we can stand listening, is futile. A hint to a symbol of all human knowledge could be a symphony or maybe a song (with lyrics of an excellent poet...). We could pack a lot of data in a 4 seconds sound, but my ears couldn't decipher that apropriately. There could also be a 4 seconds sound that is an established symbol for that, that would be interpreted in the mind of listeners as the feeling and/or associarion for all human knowledge, but if such exists I am not aware of it, and probably is also not universally known and accepted as such.

So I decided not to cast my vote in this case.

Well... I found no choice offered to interrupt or quit the voting process. I am becoming interested (I used to design, make and troubleshoot SW) to see what happens if I log off or shut down the computer (hopefully nothing catastrophic). I'll post this and look for an answer here in an hour or three... Marjan Tomki SI (talk) 17:08, 10 December 2022 (UTC)

Hi there @Marjan Tomki SI, thanks for writing to us and for sharing your reflection here. Indeed there is so much that could be packed into 4 seconds and at the same time, balancing that with a sound that is international, not disruptive, and easy on the ears is not a simple task. I don't see a vote registered for your username. If you would like to cast your preferences, we would welcome it. You can do it through the voting interface or by emailing us at soundlogowikimedia.org. Thanks again. MPourzaki (WMF) (talk) 15:44, 12 December 2022 (UTC)

PK62 natürlicher Trommelschlag

Es gibt in der Liste ein sehr gutes Soundlogo (PK62) und neun Logos, die ich für ungeeignet halte. Jetzt soll ich mir die neun schlechten Sounds x-mal anhören, um eine Reihenfolge zu bestimmen? Diese Abstimmungsmethode führt dazu, dass man entweder gar nicht abstimmt oder unnötig viel Zeit auf die Abstimmung verwendet. Besser wäre es gewesen, wenn man Punkte vergeben kann oder drei Favoriten benennt. Kolja21 (talk) 02:26, 12 December 2022 (UTC)

Hi @Kolja21, danke, dass Sie uns geschrieben haben. Ich verwende einen automatischen Übersetzer, entschuldigen Sie eventuelle Fehler. PK62 ist eine gute Wahl. Die Menschen haben sehr unterschiedliche Meinungen und die Schulze-Methode wird uns einen klaren Hinweis darauf geben können, wie die Menschen ihre Präferenzen eingeordnet haben. Wie ich sehe, konnten Sie Ihre Stimme abgeben, vielen Dank für Ihre Teilnahme. MPourzaki (WMF) (talk) 15:47, 12 December 2022 (UTC)

Man kann einfach all die Einträge, die einem gleichgültig sind, in irgendeine Reihenfolge bringen—das bietet dir genausoviel Ausdruckskraft wie in einem FPTP-System Theanswertolifetheuniverseandeverything (talk) 23:06, 12 December 2022 (UTC)

Disappointed by voting process

To its detriment, this vote is not the Schulze method. It is missing several of the voting options, most notably that voters can't keep candidates unranked. When a voter doesn't rank all candidates, then this is interpreted as if this voter (i) strictly prefers all ranked to all unranked candidates, and (ii) is indifferent among all unranked candidates. Requiring that voters must rank all ten entries is an undue burden and will probably discourage people from voting when they aren't able to just rank the two or three entries they like and ignore the rest. Regards, Danbloch (talk) 03:19, 13 December 2022 (UTC)

Hi @Danbloch, I know that some people have put in a lot of time and listened to all 10 submissions multiple times and ranked them. At the same time, I know that others have only ranked their top favourite few. It is the patterns of people's preferences that yield pathways that determine a winner. It is quite a strong method to determine a singular winner that has been used in the movement and specifically for logo contests before. I will ask my colleagues with a more in-depth grasp of the methodology to respond here. MPourzaki (WMF) (talk) 02:19, 15 December 2022 (UTC)

My point is that the system doesn't allow people to only rank their top favorite few. The only way to cast a vote is to rank all ten. Thanks, Danbloch (talk) 02:29, 15 December 2022 (UTC)

I understand. We thought that it would give all finalists a more equal chance when asking people to listen to and rank all of them. MPourzaki (WMF) (talk) 02:44, 15 December 2022 (UTC)

Wow... this is bad.

The voting set-up, this "move the bottons" arrangement, does not work for me. Even if it did, as noted above, the requirememt to rank all 10 is just apita. We should be able to just submit our top choice (mine would be AC54) or top 3 or top 5, and even note one we absolutely don't like. But all that aside, the voting method, which isn't very good, doesn't work anyway, due to... a technical glitch? Something else? Anyway, it's unfortunate that this is process that will select our jingle/theme/bite/effect whichever. It would be better to start again with a more effective selection process, and a selection set-up that is easy to use and actually works (for everyone). IMHO - wolf 08:09, 13 December 2022 (UTC)

Agreed. There's only a couple I like, the rest sound like annoying mobile phone ringtones, so I'm forced to order them in order of least annpyance. The POTY voting system would have been far better. Voice of Clam 08:54, 13 December 2022 (UTC)

I also had trouble trying to "move the buttons." (In fact, I wasn't able to do so, and nearly cast a vote for one of the worst sounds because the page somehow determined that I had voted for it. Fortunately, I didn't hit submit, and was able to back myself out of the page by hitting the back button on my computer before I recorded a vote I didn't intend to make. My concern with this vote's design is that Wikimedia won't end up with a final decision that represents the vast majority of Wikipedians and Commons users because most users will either accidentally vote for the wrong sound logo, or will give up in frustration when they weren't able to "move the buttons." 47thPennVols (talk) 09:40, 13 December 2022 (UTC)

Hey 47thPennVols, so far we've only been aware of some OOUI difficulties with Android devices, and have included a Special:EmailUser link under the how to vote section so that affected users can send their votes in via email. Could you let me know about the issues you faced with the voting script, so that we can try and fix it? I'll also note that a vote isn't final; going back to the voting script after having already voted will update the original vote. Best, RAdimer-WMF (talk) 02:26, 15 December 2022 (UTC)

The static ranking doesn't reflect what I think much either; yes it reflects the order - but no it doesn't reflect the magnitude. A 0-10 point value on each would have been better. — xaosflux ^Talk 16:14, 13 December 2022 (UTC)

• This voting process does not allow voting for only one entry, so I will not be submitting a vote, with apologies to PK62. I do not wish to support other entries or to be required to do so as a condition of participating in this process. Beccaynr (talk) 23:58, 14 December 2022 (UTC)

• Hi! We're using the Schulze method for calculating the results, which has been used in past Wikimedia votes (including the MediaWiki logo vote) and does certainly work quite well; I'm not sure of the technical glitch you're referring to. It picks the Condorcet winner if there is one, and has a rather intuitive and statistically optimal method of determining the winner if no Condorcet winner exists.

We do have a modification to the ballot, requiring all 10 be ranked in preferential order. Technically, the Schulze method can be calculated with a ballot that allows only a few candidates be ranked, or for multiple candidates to be given the same ranking. However, as with the MediaWiki logo vote, in the interest of ease-of-use for the tool and voting, and avoiding confusion with the process, the script presents all 10 in a randomized order and prompts voters to re-order it.

Though not everyone will reorder all 10, the initial order is randomized, and the true preferences of the voting population are accurately quantified through the Schulze method. After the vote, we also intend to post a pairwise preferences matrix and margin of victory matrix that can help visualize the magnitude of results.

I hope this helps ease some of your concerns, let me know if there's anything I can clarify or help with :-) RAdimer-WMF (talk) 02:47, 15 December 2022 (UTC)

I feel you do your best, but for a small group (as I expect you are) it is IMO hard to catch all the unexpected situations (some listed on this talk page), and our input shall hopefully help you refine the metod for next cases.

Info about randomized initial sequence (and I'll vote twice to verify the initial lists are not the same ;-), where - if I move my preferences to the head of the list and let the rest pseudo-random - will help, but only if the number of so "equally voted" is big enough'.

I intend to think a bit about how big the number of such votes would need to be, and if I had the time (and a piece of suitable SW available) I'd try a simulation to see about that. If someone does try this I'd be glad if (s)he let me know about results.

I actually met an essentially similar solution (in 1981 and about) with AERO (Automatic Error Recovery Operations) Hungarian program package for removing obvious errors from 1981 Yugoslavia census data. The teory was that if an obvious error was spotted (e.g. a 10 years old female works as a medical doctor, has appropriate education, and has 3 children, Age variable is suspicious). SW could vary that variable and check to see if the "corrected" combination is found in (possibly relevant regional soubgroup of) population, "corrected" age would be substituted by program. Estimate then was that with big enough population and small enough number of errors of each type, such correction could be done by program (and monitored by humans only on statistics level, which it were), and the induced error would be small enough and with added speed of getting the results (compared to time needed e.g. for error corrections with reinterview of the person in question etc.) would be justifiable.

Here I have no data to base any estimation how many votes with two (or more choices) estimating equal preferrence (which I think I saw the original Shulze method supports), and for next cases it would be good to reconsider those things, but for this voting what you propose seems good enough. Marjan Tomki SI (talk) 02:15, 19 December 2022 (UTC)

This is an interesting question which myself and Ladsgroup talked about in preparation for the vote. I initially had similar concerns, despite this method and ballot having been used for the MediaWiki logo vote. Interesting to hear about the Yugoslavian census info...that is a similar-ish problem in principle, but the data and end results are fundamentally different. We're talking about identifying the signal (population preferences) in the noise (remnants of initial randomized order) rather than handling clear discrepancies, and this method provides a great way to do so. Making any "corrections" to this vote data would be practically guaranteed to be a net negative in terms of vote efficiency, in capturing population preferences. The Schulze method uses pairwise preferences and margins of victory in its calculation...i.e., considers the amount of people who voted candidate A over candidate B, and candidate B over candidate A, and so on. With the roughly 2,000 votes we have and the initial presented order being random...the remnants of initial random orders will tend to cancel each other out; it is not a problem. Population preferences (the signal) will rise to the top out of the increasingly miniscule noise. Hope this helps, RAdimer-WMF (talk) 18:18, 19 December 2022 (UTC)

Yes, Shulze method uses ranking principle, but if I recall correctly, it explicitly allows the voter to assign two or more choices same rank (as if at sport two or more runners had same time - within precision of measuring equipment - are asigned same rank), or not rank part of the choices, and meaning of such vote is defined. Here and now, this possibility is not available (and MPourzaki (WMF) explained his/your rationale for that).

Above, You explained that if the voter moves only his preferred choices to the start of the list and leaves others as they were, it will be like they are equivalent, because of random preanged list offered to the voter. But that (pseudo)equivalence is only true if: 1. prearanged list are truely random, and 2. if number of voters using this choice is large enough.

Whenever a voter can't vote with two choices asigned same rank when (s)he wants to, you have a "correction" of the vote to something voter didn't vote.

So the example I gave with Census data correction was a similar problem: here we couldn't vote as we wanted (it seems you actualy modified Shulze method), at 1981 Yu census the interviewee was usually not asked for correct answer in case of detected logical error.

In both cases, assumptions that using random data would give relevant result would need to be monitored, and the "corrected" data dropped, and process amended and retried if previous result was found flawed. For Slovenia and Croatia part of Census data I know that it was done so, because I had to rerun preparation and correction runs several times with parameters modified by subject matter specialists for parts of both republics (and keep all the results if external inspection and verification was necessary).

Whenever one uses similar mechanism, both points 1. (randomness) and 2. (numerus) should be monitored. It could be tested in a test run to see if statistic of the results with randomized "same" ranking for several choices really come out close to geting same rank, and if so what is the necesary minimum numerus. In this case IMO it is number of voters which wanted to assign same rank to more than one choice, not number of all votes, so if "With the roughly 2,000 votes we have" 2000 is expected number all votes, 2000 would be the correct numerus only if every voter had his choices partially unranked.

I think this is a good oportunity to (re)test your voting procedures (and I wish you good luck and success with that). Marjan Tomki SI (talk) 01:43, 20 December 2022 (UTC)

Yep, the ballot is modified from that maximally permitted by the Schulze method. The method of result calculation (Schulze) was not modified.

Good points regarding randomness. The prearranged list is effectively truly random. I think your point is that you would need a large enough number of voters not changing a given set of ranking orders (i.e., first, second, third) for the randomness to be maximally effective in minimizing noise. So 2,000 wouldn't be the most relevant number; what's relevant is how many voters chose to not rank each possible set of ranking orders.

However, remember that the orders are relevant only in the pairwise comparisons of people who ranked A over B, and B over A. For the people who did not manually rank either A or B, on average 50% of the time A will be randomly ranked over B, and B over A, in the prearranged list shown to them. The same for every other two candidates.

Population preferences very clearly manifest themselves through voters manually changing the initial random ranking, and the magnitude of that can be viewed in a margin of victory matrix, which will be published with our voting results in mid January at the latest. Best, RAdimer-WMF (talk) 15:57, 21 December 2022 (UTC)

I agree. I listened to all the sounds, but only two I liked, and among them only one is my favorite. Why do I have to order 10? Also, I can't vote from my mobile phone, I'll have to go find my PC to be able to vote. When will the Wikimedia Foundation do a vote using a normal system? Because the one used here is neither “trustworthy, reliable, open and accessible for all”. Ivanics (talk) 15:14, 19 December 2022 (UTC)

A LOT of staff

Looks like ~5% of all votes are from WMF employees so far, that seems a bit surprising as we rarely get that much staff involved in things open to community input. Side note, an editor notice (Watchlist notice) was activated on enwiki today, so may attract more editors from that project. — xaosflux ^Talk 16:19, 13 December 2022 (UTC)

Hey @Xaosflux, thanks for inviting more participation into the vote, this is great. Could editors that are not interested in the notice, turn them off? I think the greater participation from staff is because this is a fun project, the topic is accessible, and it's relatively easy to participate. MPourzaki (WMF) (talk) 02:34, 15 December 2022 (UTC)

Yes, there is a dismiss link for the watchlist notice. AntiCompositeNumber (talk) 02:37, 15 December 2022 (UTC)

^that. The "watchlist notices" are fairly unintrusive, as they only show when an editor is actually looking at w:en:Special:Watchlist. — xaosflux ^Talk 11:54, 15 December 2022 (UTC)

Thanks @Xaosflux and @AntiCompositeNumber. We were considering mass messaging village pumps, but thought that would be a bit much. What do you think? MPourzaki (WMF) (talk) 13:32, 15 December 2022 (UTC)

@MPourzaki (WMF) I don't think that's a great idea; but perhaps manually messaging some VP's of large wiki's. Alternatively, potentially increasing the existing CN for Wikipedia's to more languages. I'm not sure if WMF has had pushback from non en-* for prior banners? The impression settings are low, getting ja, ru, fr, de, es, it, zh, fa, and pt in there could help. — xaosflux ^Talk 14:44, 15 December 2022 (UTC)

I'm not a big fan of mass messages either. Will do a round of manual messages to some VPs this weekend as a closing reminder. We already pinged es, pt and id. There is a whole conversation happening on the Kurier already, which is great. Increasing the CN reach is great, but I would not want to post in English on those other languages. –MPourzaki (WMF) (talk) 19:33, 15 December 2022 (UTC)

Abstain

This vote does not allow abstained votes at all. Saying that all options are bad is perfectly valid. So, I am making an public vote to abstain, unless I am allowed to make that choice through normal voting process. Snævar (talk) 20:03, 14 December 2022 (UTC)

I agree. The idea is crazy: waste some readers' time with a jingle; more noise pollution. And I can't hear any of them anyway. So I wanted to register a protest vote. I clicked just in case it was possible; I hope that didn't count as a vote for whatever order I was presented with. Yngvadottir (talk) 02:38, 19 December 2022 (UTC)

Translation to Slovene

Hello! I have translated the page to Slovene (Commons:Sound Logo Vote/sl), but the button does not seem to work. It works with other languages. Could you please check if it works for you and what could be wrong? --TadejM ^(t/p) 04:39, 15 December 2022 (UTC)

Done Amir (talk) 07:27, 15 December 2022 (UTC)

Thank you. --TadejM ^(t/p) 10:12, 15 December 2022 (UTC)

Only 10 sound votes

I don't like that I have to strictly vote for 10 sound votes. The vote = all or nothing. Dušan Kreheľ (talk) 19:09, 18 December 2022 (UTC), --Dušan Kreheľ (talk) 19:16, 18 December 2022 (UTC)

Why is there no vote "against all"?

Subj. Lesless (talk) 16:52, 17 December 2022 (UTC)

Hi @Lesless, I think you know well that in wiki practice people usually post their comments chronologically by the date where a new comment goes to the bottom of the page rather than inserting it up top. Do you have other examples in Wikimedia contexts and logo contests where there is an "against all" vote? We weren't aware of such precedent. MPourzaki (WMF) (talk) 21:04, 17 December 2022 (UTC)

Always: w:ru:Википедия:Форум/Архив/Общий/2020/12#Против всех, w:ru:Википедия:Форум/Архив/Общий/2021/12#Против всех, regulations: "включающее варианты логотипа и вариант «против всех»". Lesless (talk) 07:12, 18 December 2022 (UTC)

Hello Lesless! In planning the vote, we reviewed available voting methods and tools, and went with the script and method used for the MediaWiki vote. That being the Schulze method with a modified ballot (all candidates are preferentially ranked by each voter). There is no "against all" because this is not a support/neutral/oppose vote, nor is it a single vote plurality method as in the linked holiday logo votes. There is no "status quo" option for the sound logo. I hope this helps, let me know if there's anything else I can clarify. Best, RAdimer-WMF (talk) 01:31, 19 December 2022 (UTC)

See above. We should have been given the option of rejecting all the choices, or the entire concept. Yngvadottir (talk) 02:40, 19 December 2022 (UTC)

The Commons search for Wikipedia could have been a contender

I posted a similar thought on a related page but it belongs here. Essentially I agree.

In the recent Kazakhstan presidential election there was an option to vote for "against all" and this did better than all candidates except the inevitable incumbent. Having listened to the selection here with my household, we agreed that we really needed this option in this election too. It's a poor state of affairs to be less democratic than Kazakhstan.

As almost anything would be better, I just tried searching Commons for audio files using the keyword "Wikipedia". Hit #3 (right) seems as good as any of the candidates and more distinctive than most.

Andrew (talk) 19:19, 19 December 2022 (UTC)

Maybe the WMF could dig into its deep pockets and pay a professional composer to write its ditty? As I suggested elsewhere, maybe Mary Halvorson would adapt something from "892 teeth"? SashiRolls (talk) 01:42, 21 December 2022 (UTC)

Explanation for my votes

Since about 90% of the discussion here has been about the voting system (mostly negative), I have decided to add to the 10% that's about the actual sounds, in the hope that anyone looking for guidance on that will find some addition to what Andrew and Jynus have offered above. You can consider my opinions in your own vote, or not.

First, I must say, I like all the sounds, all of which were likely created by people who had some more skill in composing and creating them and/or time to create them than I did. So everything I say I say with respect and acknowledgement to the effort involved.

I decided while considering my votes all week that my criteria would be, how does this stack up against some of the best-known sound logos (or at least best-known to me, a middle-aged American). I was guided by this video, which runs through the ten best where the video creator was able to find out and attribute the creator of the sound. These include:

• the AT&T sound,
• the Michelin sound,
• the very recognizable Mac bootup sound,
• the Brian Eno Windows 95 bootup sound (hearing it for the first time in a very long time brings back a lot of memories)
• the iPhone Marimba ring (Someone should find a way to adapt it into a song about people who are too lazy to set their phone ringtone to something other than the manufacturer's default yet still think they're cool for having those phones (when, really, the real reason is that they're afraid that if they do choose an individualized ringtone for themselves, the people to whom it matters that they know the phonebearer has that particular brand of phone won't know they have those phones )
• the Intel sound
• the short staccato high-octave five-note piano lick that T-Mobile uses (and do they offer that as a ringtone? Not that I want it—I have an Android, and I have proudly distinguished myself from the herd by setting it up so it uses a 25-second burst of Kraftwerk's "The Telephone Call" (from about 0:15 to 0:40 in that video) whenever my family, or really anyone who's been trying to reach me about my car's extended warranty, calls. But that would still be a cool ringtone, and I really would mind hearing it less in public than the Marimba ...)
• the classical Nokia piano logo (less familiar here in the US these days)
• the 20th-Century Fox fanfare that any regular Rocky Horror viewer inevitably can't resist following up with by intoning "A long, long time ago in a galaxy far, far away ..." immediately afterwards.

To these classics I would add the NBC chimes, the Lucasfilm Deep Note, McDonald's "I'm lovin' it" five-note melody, the elegantly simple HBO sound logo, and (since it's fresh in our minds still) the Qatar World Cup sound logo, that five-note phrase on some Middle Eastern instrument (an oud?). Americans my age might also have a soft spot in their hearts for the WGBH logo, a/k/a "Prelude to Zoom", whose '70s analog-synth sounds should sound quaint but are actually pretty retro cool (at least if you listen to classic rock stations and regularly hear Steve Miller's "Fly Like An Eagle", with that similar spacey repeat of "Space Intro" at the end (sounding like a medley of all possible public-TV sound logos of that era)) and lastly the percussive and cool CBS Special logo, very seasonable since it was always the last thing you saw before A Charlie Brown Christmas or Rudolph, on that one night they showed them during the holiday season back then.

The point of this exercise is to reacquaint the ear and the mind with sound logos that have worked for their brands, for at least a decade or so, even after those brands may have gone defunct. After reviewing them, I decided to consider first and foremost in my voting what could do the same for us.

Here are my choices, from tenth to first:

• JW08: The combining voices sound great (and I sort of had the same idea) ... but at the same time too much like the HBO and Mac logos. We need to be distinctive.
• UN03: Interesting ideas, but feels like it's trying to do too much. But definitely short like a lot of the good ones above.
• OZ85: Well ... on the one hand, this one is definitely distinctive; it sounds like none of the other logos I've mentioned above, or even any other candidate. But here this may be too much of a good thing. My main concern is that it will turn out within weeks of everyone falling in love with this one and adopting it that the voices are reading off some sexual proposition or ethnic slur in some language, or dialect thereof, so obscure that we don't even have a Wikipedia in it yet (but, be assured, after something like this happening, we will). Or we'll get accused of cultural appropriation or something like that because it might sound too much like some sort of obscure cultural thing to some people, even if it isn't exactly.
• VQ97: I agree that that five-note phrase goes nicely with "Wi-ki-me-di-a!" But then why not actually have voices/a voice singing it? Also it starts off with another swell similar to the HBO/Mac sound.
• AC54: An interesting idea, and I get it, but I'm not sure this melody takes full advantage of the concept (And BTW, to get in one comment about the voting process, why not perhaps set up the page to present the sounds in random order, not only when first opened but anytime it's refreshed, so there's no advantage from being first alphabetically? Just asking ...)
• PK62: Both Andrew and Jynus liked it. I like what's going on with the drumbeat, but the transition to the kalimba seems a little abrupt.
• FM76: If we have to do something similar to the HBO/Mac logo, this is a better way to do it, as it to me evokes the idea behind it without sounding too much like it.
• BY23: I think here, the composer managed to execute their idea. I do agree that the laughing child may make people think this is more appropriate to a cable network showing children's programming.
• DS71: A more consistent whole that, like the best sound logos currently out there, gets in, makes its point and then gets out.
• GX13: The one that sounds to me like it would fit in best with the classic sound logos I listed above. Gets the whole idea across within the four seconds and then gets out.

That's my take, for what it might be worth to anyone. Daniel Case (talk) 06:44, 19 December 2022 (UTC)

Hi @Daniel Case, thank you for taking the time to reflect on the submissions, for doing background research like a true Wikipedian, and for sharing your thoughts here. Appreciate it. Thanks for participating. MPourzaki (WMF) (talk) 17:02, 19 December 2022 (UTC)

Manipulating with Schulze

VQ in 1st over time

Dear all, in another organization, I have worked with Schulze quite a lot. I would like to point out that the Schulze method allows for manipulation of the result in ones favor if the other votes are recorded publicly. This is clearly the case here.

I have chosen to point this out only after the voting procedure is over, such that it may be taken into account for future votes. -- MGChecker (talk) 01:54, 20 December 2022 (UTC)

MGChecker, I'd be interested to learn what you mean by manipulation of the result, and why it depends on the votes being recorded publicly. I'm not aware of any statistically overpowered, or otherwise detrimental to efficiency, strategic voting methods in a Schulze method calculation. Best, RAdimer-WMF (talk) 00:17, 21 December 2022 (UTC)

With access to the prior votes you can calculate all the current beatpaths, then strategically make votes to alter them to promote a candidate; a band of strategic voters registering late votes could strategically vote to alter them. The mathematical proofs for this are a bit much to go in to here. Not exclusive to Schulze: a voter or group of colluding voters could for example disingenuously vote for their own second choice candidate as their first ranked one, if doing so would allow it to win overall in the case where neither their 1st or 2nd choice would have otherwise won. — xaosflux ^Talk 00:53, 21 December 2022 (UTC)

Interesting...hm, I'm not sure how any vote not ranking your true first choice as first would benefit your true first choice, in this method.

A given pairwise preferences calculation is indifferent to the numerical rankings on the ballot relative to candidates outside the specific pairwise comparison: it reflects the number of voters who voted candidate A over candidate B independent of the relative placement of those candidates in each individual's ballot. Ranking your true first choice second would register as a +1 to the net number of people who prefer your second choice candidate to your first choice, and would not be helpful in manipulating results.

What definitely is possible is to rank your favorite first and then the opposite of the current order...e.g., let's say your favorite was currently #5 in a Schulze ranking, you'd vote for #5, #10, #9, #8, and so on. This would still only count as a 1 number change for each pairwise matchup: it would improve the situation of your favorite candidate relative to each other candidate by 1, and reduce the magnitude of the win from candidate N to N-1 by 1. In other words...this doesn't directly improve the situation of your true favorite candidate, but reduces the margin of victory for every candidate except yours by one. This type of strategic voting is alluded to in Schulze's 2018 paper (p.322), but only in the context of Schulze proportional STV, because that is where this type of manipulation is relevant to determination of winner(s).

If there is no Condorcet winner, i.e., where beatpath calculation is actually relevant, I'm not sure how manipulation could occur more efficiently than the above idea, where a voter optimally decreases the magnitude of opponents' paths by 1 and increases their favorite candidate's by 1. It would still require a lot of voters to do this to make a dent in the result, because it doesn't do much, and it can be argued that it's an optimal reflection of preferences.

Best, RAdimer-WMF (talk) 03:41, 21 December 2022 (UTC)

Hm, rethinking it...depending on how the beatpaths are organized, and which candidates they pass through, there can be more efficient methods of minimizing the strength of an opponent's strongest paths and maximizing a manipulative voter's favorite candidate's, beyond simply the inverse of the current Schulze order. This would need to occur very late in voting, and would require some interesting calculations to ensure that a vote isn't inadvertently harming their favorite candidate's best paths. However, the effect of an individual voter is still heavily limited, and would require a lot of votes to make any change in results. I'd be interested to see if there's discussion of this in one of Schulze's papers, or similar. RAdimer-WMF (talk) 04:03, 21 December 2022 (UTC)

Agree, I came across a few papers that all required a coalition of late strategic voters. Real-time access to current votes is more likely to lead to traditional sacrificing of your own 1st position candidate in favor of your 2nd position candidate in attempt to have 2 beat 3-n. — xaosflux ^Talk 11:31, 21 December 2022 (UTC)

Indeed, though most of what I've been able to find focus on whether calculating an optimal voting strategy is possible, not identifying the weight of that voting strategy relative to a normal voter's ballot. In order for a manipulative voter to have an actual effect on a not-currently-winning candidate, it seems that:

1. There cannot be a Condorcet winner, otherwise the optimal strategy would simply be what I mentioned above with ranking your favorite first and then the inverse of the current Schulze ranking, which has minimal effect above a normal vote.
2. It needs to be late-stage, with public votes, to calculate the optimal strategy.
3. The true population favorite needs to be close in ranking (and lowest path strengths) to the manipulative population favorite.
4. Manipulative voters still need a lot of voters to make a change, and need to optimally calculate which paths to weaken. I've been unable to find a study that discusses the difference in effect on the results between using an optimal manipulative strategy and getting the same number of accounts to simply rank some candidate first.

Interestingly, I'd intuitively assume that the requirement, in this Sound Logo vote, that every voter rank all 10 candidates makes the optimal manipulative strategy under a non-Condorcet situation far more difficult to calculate, as every vote effects every pairwise comparison, and is far more susceptible to becoming useless if more non-manipulative votes come in. RAdimer-WMF (talk) 15:36, 21 December 2022 (UTC)

Agree, I'm not seeing that anything explicit was done to try to manipulate beatpaths here; I did not look in to extreme edge cases that would also require getting 1 or more winners disqualified. — xaosflux ^Talk 17:11, 21 December 2022 (UTC)

• In this election, one candidate has a large lead, but consistently had that large lead as a function of overall votes by time throughout, see chart. — xaosflux ^Talk 11:32, 21 December 2022 (UTC)

Unofficial poll results

While it makes sense that it may take months to do something in production, surely the results of the poll can be published? I suppose since it is public someone from the community could just crunch the numbers but I'm assuming the organizers already have a program set up for this? — xaosflux ^Talk 11:58, 20 December 2022 (UTC)

...and that somebody is me :D — xaosflux ^Talk 12:28, 20 December 2022 (UTC)

• Using the online calculator at https://www.cs.angelo.edu/~rlegrand/rbvote/calc.html suggests a winner of.... — xaosflux ^Talk 12:23, 20 December 2022 (UTC)

Comparisons of strongest beatpaths for each pair of candidates

AC>BY 1435783 839822 BY>FM>DS>AC AC>DS 1260537 985609 DS>FM>PK>UN>GX>AC AC>FM 1265200 985609 FM>PK>UN>GX>AC AC>GX 1091560 1043651 GX>AC AC>JW 1595641 839822 JW>BY>FM>DS>AC AC>OZ 1624689 794780 OZ>JW>BY>FM>DS>AC AC>PK 1160507 985609 PK>UN>GX>AC AC>UN 1108066 1043651 UN>GX>AC AC>VQ 874371 1260840 VQ>AC BY>FM>DS 839822 1325623 DS>BY BY>FM 839822 1295389 FM>BY BY>FM>DS>GX 839822 1434258 GX>BY BY>JW 1283469 851742 JW>BY BY>OZ 1465377 794780 OZ>JW>BY BY>FM>PK 839822 1327213 PK>BY BY>FM>UN 839822 1413720 UN>BY BY>FM>DS>AC>VQ 839822 1555719 VQ>BY DS>FM 1068021 1067190 FM>DS DS>FM>PK>UN>GX 985609 1205473 GX>DS DS>JW 1485644 839822 JW>BY>FM>DS DS>OZ 1583419 794780 OZ>JW>BY>FM>DS DS>FM>PK 1030102 1149229 PK>DS DS>FM>PK>UN 985609 1196998 UN>DS DS>AC>VQ 874371 1395803 VQ>DS FM>PK>UN>GX 985609 1232199 GX>FM FM>JW 1519812 839822 JW>BY>FM FM>OZ 1553112 794780 OZ>JW>BY>FM FM>PK 1030102 1105109 PK>FM FM>PK>UN 985609 1196663 UN>FM FM>DS>AC>VQ 874371 1394648 VQ>FM GX>JW 1574591 839822 JW>BY>FM>DS>GX GX>OZ 1590467 794780 OZ>JW>BY>FM>DS>GX GX>PK 1156562 985609 PK>UN>GX GX>UN 1077504 1057707 UN>GX GX>AC>VQ 874371 1312158 VQ>GX JW>OZ 1340431 794780 OZ>JW JW>BY>FM>PK 839822 1555613 PK>JW JW>BY>FM>UN 839822 1553472 UN>JW JW>BY>FM>DS>AC>VQ 839822 1695578 VQ>JW OZ>JW>BY>FM>PK 794780 1646657 PK>OZ OZ>JW>BY>FM>UN 794780 1601824 UN>OZ OZ>JW>BY>FM>DS>AC>VQ 794780 1748835 VQ>OZ PK>UN 985609 1149602 UN>PK PK>AC>VQ 874371 1342351 VQ>PK UN>AC>VQ 874371 1293141 VQ>UN

• That tool output is: VQ is the only candidate to lose no beatpath comparisons and so wins the election outright..

This is completely unofficial and could be completely wrong. No scrutineering of voter eligibility was performed. Input was from Commons:Sound Logo Vote/Votes/raw, the "#" is the ballot number, taken in order. — xaosflux ^Talk 12:26, 20 December 2022 (UTC)

Same results using this calculator: http://www.ericgorr.net/condorcet/condorcetCompute.php — xaosflux ^Talk 12:45, 20 December 2022 (UTC)

In looking at only "1st place" votes, the results were:

• VQ:523
• AC:315
• PK:240
• UN:196
• DS:174
• GX:178
• FM:159
• BY:121
• OZ:101
• JW:58

Also placing VQ in the lead by a significant majority. — xaosflux ^Talk 20:09, 20 December 2022 (UTC)

Order of winners, should all higher ordered winners be disqualified: VQ, AC, GX, UN, PK, DS, FM, BY, JW, OZ — xaosflux ^Talk 20:18, 20 December 2022 (UTC)

Awesome. We have a script for this, the same that we used in the MediaWiki logo vote, that makes it a bit easier to calculate (and easier to identify magnitude). Given both that it'll take some time to re-verify eligibility of each voter and vote integrity, and that most of the WMF is on break, voting statistics will be published mid-January (at the latest). To preemptively note, there will be rounds of legal due diligence before the contest winner is ratified and announced, which is technically distinct from the winner of this election. If the top-voted sound logo does not pass the legal checks, the second-placed logo will be, and so on. We're hoping to announce the final contest winner between the end of February and April 2023. Best, RAdimer-WMF (talk) 00:51, 21 December 2022 (UTC)

Community voting statistics

Hi all! I've uploaded the Schulze results, and some other relevant information about the vote and voters, to Commons:Sound Logo Vote/statistics. Please note that this is the result of community voting, which is not necessarily the official sound logo contest winner, as there are rounds of legal due diligence before the contest winner is ratified and announced. Let me know if there's any questions about the stats page. Best, RAdimer-WMF (talk) 03:30, 13 January 2023 (UTC)

Pinging users who have expressed interest in these results: CharredShorthand, Xaosflux. RAdimer-WMF (talk) 03:31, 13 January 2023 (UTC)

What if a username changed after they sent their vote?

In rare/uncommon cases, usernames change. Someone may forget to fix it after it changes. Should we have recorded the user IDs? 154.20.213.253 00:00, 29 March 2023 (UTC)