Talk:CI2019List(commons,act5.e)

@RMaung (WMF) Your survey is MUCH MUCH too long: more two hours filling the forms, and still not half-finished (you said it should take 15-25 minutes, this is definitely FALSE) ! Those that have much more difficulties than me to read will take even more time: your questions are really very long, too complex, difficult to read correctly. And the survey is only in English.

At almost each page I see the cursor coming back to a lower completion level. How did you structure it ? Really this should be several distinct surveys and we could fill them easier. Too many questions are repetitive.

And this survey is definitely NOT neutral, severely biased (with very evident omissions instroduced on purpose in the proposals, and multiple ways to pose the same question to get the same expected biased answer).

I've not been able to finish it, this survey never ends (and the good sign is that questions are not numbered, we don't really see the number of questions.

I don't know what you can conclude with it: an half-filled survey can only give biased results based only on the first replies.

The results may be only usable for English Wikipedia, and once again biased for Americans only and in favor of a small community of users.

Or it will be completed only randomly by most users giving non decisive results.

Plase reconcentrate such survey on more precise topics, publish several surveys if you wish and allow users to select which one to fill in.

This will also allow some surveys to be translated, notably those that are more important for some linguistic communities, and will open more correctly the point of views you'll receive.

For now I consider this survey a complete failure. verdy_p (talk) 02:49, 10 September 2019 (UTC)

@verdy_p I am sorry you are having a negative experience with the survey! The estimated time is taken from pilot testing of the survey. The survey is actually available in twelve languages other than English (French, Portuguese, Arabic, Italian, Chinese, Japanese, German, Dutch, Spanish, Polish, Russian, and Ukranian). You can change the survey language in a drop-down menu at the top-right of any browser page. We are hoping each year to make the survey shorter, offer it in more languages, and improve the experience of taking it. Thank you very much for your feedback and the time you spent taking the survey! RMaung (WMF) (talk) 15:25, 10 September 2019 (UTC)

Note that the time to complete it is not depending on the language used; I'm fine with reading English (even if many users are not). But there really are too many options, to many repetitive questions asked again and again with subtle differences, that should have been grouped logically by topic and most probably splitted into multiple shorter surveys that we can take one by one and really complete each in a few minutes (and then get back later to fill in the other surveys in the collection if we need or want to).

I've past another hour on it, and no significant progress: I'm still far from the end (no sign it will ever finish in a predictable time). You probably used really too many "optional" sections that are activated one by one depending on the former questions, such thing should be limited so that this does not multiply each time the number of combinations, I think I'm at a point where I'm lost in a factorial-growing or exponential number of combinations that your survey generates). It's impossible to isolate which reply(ies) caused that behavior. Most probably this is caused by the number of checkboxes I ticked in prior questions, and all these checkboxes are implicitly multiplying to hundreds or thousands questions. Above some theshold, don't expect that people will resist the option to just click the minimum or select replies randomly with lots of errors. The results will then be very insignificant if people jsut come to the conclusion that they will speed up and click randomly some questions.

There's a clear sign that this survey is too long: its translation is also very long to complete: any survey should not include more than about 50 questions if these are just radiobuttons with 2-5 options and a single freeform reply (and this really requires 15-20 minutes as stated): you're MUCH above this threshold, which means you should have restructured it and published several smaller surveys linked from a page where they would be listed once activated progressively. Due to its size, the survey was clearly not tested, or just superficially with fast random clicks/replies of at most one box in questions with multiple replies. If you want people to fill in some conditional subsurveys, don't force them to reply them immediately, just add these in their current survey profile (but as you say you do not "track" those that reply, I think you have problems to manage sessions and respect the privacy statements: to do that at start of replying to a survey you may propose to post an anonymized link to the survey state to track its completion and allow people to finish them in several steps on different days. If you cannot do that (no user profile), you have no choice: split surveys with clear topics and people will remember which one they have replied want to reply. verdy_p (talk) 16:10, 10 September 2019 (UTC)

Also note that the unique ID you post on user talk pages is not anonymous. This ID should be used only once, and you should propose at start of the survey to post a new secret anonymized link to start the survey, so that we can return to it. I hope that the unique ID posted in our talk page is not used to identify users and associate their replies and that you have considered privacy when granting access to people managing the survey results (which should probably receive only replies under separate unique IDs, when they will process and aggregate the replies for statistic analysis. But I know that some admins have seen my replies (even if I did not name any one) and made their own interpretations (there's lot to comment about the roles of "admins" and their abuse of privacy, and lack of protection offered by the WMF. As well, all demonstrates they have too often used private data collected and kept illegally on third party sites, and reposted what they found in public areas (without permission and without even notifying users from which they collected and republished their info without consent, and without right of rectification).

The WMF is NOT respecting the RGPD in its projects and still has work to do to improve the security and enforce the privacy statements with contractual requirements for admins and effective sanctions plus active corrections when needed to remove all publication of forbidden data collected illegally. In fact, this is a comment you can take into account: contributing on Wikis is even more dangerous than contributing on commercial sites which are liable and want to protect their own images to gain and keep reasonable trust. Surveys are critical and probably the WMF should use the service of qualified companies that will perform that under strict contractual agreements (including penalty fees if they fail) and direct liability (from the company directly, not just their individual workers: it's the responsability of the company to ensure users and compensate them). I know this can have a cost, so you should completely separate the roles of peoples, not allowing them to mix some privileges for various admins rights: this will ensure that no one can work alone and do what he wants, and that there will be a team to supervize work made by separate teams, and a separate ombudsman who will also not have access to private data but will only formulate statements only based on what is given to them (temporarily) by each party that have accepted their independant ruling. Wikimedia clearly lacks such organization that would work on all wikis without a few overpowered people. No one in Wikimedia should have universal access (not even the WMF president: this can be warrantied by contracts for all delegations of limited powers). verdy_p (talk) 16:32, 10 September 2019 (UTC)