Wikiask:Discussions

We can use this page (and the corresponding talk page) for discussions about Wikiask.

Alternatively, Telegram (https://t.me/wikiask) is a good place for live discussions.

What is WikiAsk wanting to be, in terms of content?

I haven't edited very much yet, but I realize that I started out with a belief that the kind of question-answer pairs WikiAsk wants to address was clear, while now I don't feel so sure. What I got from introductory discussions includes:

• that WikiAsk would address question-answer topics that are somewhat complex, i.e. that couldn't be answered by a single number (e.g. how tall is a certain basketball player)
• that WikiAsk would create content that is "free" by some definition. This is at least partly addressed by WikiAsk's choice to adopt the CC-BY-SA (am i getting that right?) license.

But what is the goal for type of content to develop?

• Is it to address questions that people often have? Then can we work from some list of most common questions people have, like what are the most common Google searches?
• Is it to address certain topic areas?
  • Could it try to tackle topics that are new and in the current news? E.g. the war in Ukraine, or election denial, or other topics, possibly controversial ones, that other forums aren't dealing with very well? Maybe WikiAsk could have an advantage somehow, in topic areas that are murky, where terms aren't well defined and complexities are not yet settled out yet?
  • Could it try to address certain field areas? E.g. cultivating expertise and going systematically through a given subject area, generating good questions- and-answers that aren't much covered, or are not well-covered, by existing forums?

I'm just not sure how I can fit in, and what it is that I can be part of. --Doncram (talk) 00:01, 27 October 2022 (UTC)

Bill and others, I am seriously asking here, like what is the elevator speech for this enterprise? What did you say to me that at least worked at the time. What do you actually say, in the elevator? --Doncram (talk) 09:55, 29 October 2022 (UTC)

Is Wikiask about the most important questions?

I haven't heard it asserted yet that Wikiask tackles the toughest questions or the most important questions, or the like. But it could take on a role like that, and if it did, that might be easy to explain. At least as one part of what Wikiask does. And this would be very helpful in recruiting, how has it been said, "power contributors" or something like that.

"Most important" could mean "most frequently searched for", but I don't mean that. I suppose I would like to see a list of the changing "most frequently searched for, day by day, or month by month, or year by year, for Google or any other big search system. And maybe some such source could be found, and maybe Wikiask could usefully focus on those. But i expect those are temporary things, like searching about Princess Diana. Actually, in Wikipedia's {{w:Signpost}} newsletter, there has been a running feature on what are the top most read articles. Which if i recall correctly are about pop culture things, temporary things.

By the most important questions, I mean really big perennial questions, including:

In the science/math domains

Some of the most important questions to address could be:

• Why is the sky blue?

Never have I ever gotten a satisfactory answer, although I do recall teachers providing a reading that supposedly answered it, which maybe they would read aloud and conclude by saying, "there you have it", "it has been explained".

• How many irrational numbers are there, relative to the number of rational numbers? (the answer is LOTS MORE, and it can be explained and understood. This is the concept of countable infinity vs. uncountable infinity, maybe it should be asked using those terms instead.)
• What did Einstein figure out, and how? / What is special relativity? / What is general relativity?
• What does it mean that something has been statistically proven? [Weirdness about the the null hypothesis, and one alternative hypothesis. How about a null vs. all alternative hypotheses?]
• What is the scientific method?
• What is "incompleteness"? Around 1900 there was a famous list of 100 unsolved great mathematical hypotheses. Many have since been solved, e.g. the three-color map problem (which can be simply explained). Many have been disproved. But some of them turned out to be, well, impossible to say. Meaning it was _proven_ they could not be proven true, and it was also proven they could not be proven false. What the heck does that mean?

in social domains

• In social domains, there arise many questions because of our natural inability to make "interpersonal utility comparisons" or the like.
  • Like how do we know which of two patients is in greater pain? Which of two persons is happier. Etc.

In a medical setting, people are asked to rate the pain they are experiencing on a 1 to 10 scale. Even if the answers given truthfully by each person, without any intention to game any decision system which is going to use that data, it is simply impossible to say that a "3" on my scale is the same as a "3" on yours.

• On the other hand, I think it is asserted that people taste less as they get older. How would you know?

• What is love, and how does one know if one is in love?"
• How does one know there is a God? There are multiple good answers.
• Why do people like music?
• What is the culture of a people/tribe/nation?
• Which bidding or voting system is most fair? / In which bidding or voting system do people have proper incentives to tell the truth? / Which system cannot be gamed? Pose in current terms: is the proportional voting system now used in Alaska and Maine(?) and wherever else better? [There is no such system, no system which cannot be gamed. I think it can be explained/understood why a system with truth-telling would be best, and I think but am not sure it can be explained in fairly simple terms why there cannot be such a system. So under what circumstances should you _not_ tell the truth, not give your true ranking of preferences for multiple candidates?]
• What is the fundamental theorem of calculus, simply put?
• Go through academic subject areas, or at least all the science ones, and consider what are mankind's greatest accomplishments? [Answers could be strategically given that answer, instead, what are the most important accomplishments that can be understood? Answers could be in terms of survey results, on what do researchers/teachers in the field think are the most important accomplishments. Answers could be, for a college-educated person, what is the farthest thing reached in undergraduate curricula.]
• In theoretical physics, it is arguably general relativity. I dunno from tensors.
• In finance, it could be the Capital Asset Pricing Model, or the Black-Scholes Option valuation formula.
• What does chaos theory mean, or imply?

There must exist lists of what are the greatest, most important questions, or the most impossible to answer, etc. What are all of the questions on some such list: answer all of them, and say that Wikiask does that.

Is wikiask different because it will actually give the best answers for those important questions?

Certainly in many areas Wikipedia is unintelligible. In general its articles in statistics are. Could Wikiask specialize in doing better, in Q&A form, better than the understanding which one can get from reading the corresponding Wikipedia article? Go through, systematically, some index of topics/articles to generate a good list of Qs, and then invite/recruit specific people to participate in the interesting discussion that is going on, about how to best answer this specific question.

One example would be (where Wikipedia is unintelligible):

• What does it mean when a study comes out saying that factor A doubles your risk of cancer? [It absolutely does not mean that your risk of cancer is doubled. Instead, it means a kind of weird thing that the researchers effectively agree among themselves is the sequence of words that will be uttered to express that weird thing to one another. It is code. Two good answers can be given on this question, though: one answer is what that does mean, another answer is what could/should be said, instead (which would involve some programming to calculate). Journalists do not understand it. Journalists are constrained, or they accept constraints, in some ways that are arguably legitimate. But they should not end up saying completely false things, which they are often doing, while they would choose not to, if they could.]

--Doncram (talk) 10:22, 29 October 2022 (UTC)

• There's another statistical question, which I can't remember what it is right now, where there is great difficulty in explaining/understanding it, where it simply a matter of no one having produced the right 3-D picture that is needed. I then didn't have the right graphical software to produce it easily, and I wouldn't have gotten "credit" in any way, but I knew what was needed, and i recall that it was an important, general thing. I could find/reconstruct this. --Doncram (talk) 10:22, 29 October 2022 (UTC)

Is wikiask different because it will be braver, or bolder, than Wikipedia, meaning it will give good working answers, where Wikipedia falls down?

Could Wikiask target the blind spots where Wikipedia fails, in practice? This would include concepts in current events, where there is not yet any body of academic research that has developed answers, or even figured out what an answer can be. In practice Wikipedia turns out to be conservative, and coverage/answers that anyone suggests, just get deleted. One recent, possibly still current thing is covering what is election denial. What is meant when a candidate says they disbelieve the result of a past, settled election. Or that they will disbelieve the result of a future election, if it doesn't turn out as they prefer. Wikipedians are tripping over this due to how their processes work, and the presence of political division. [It's obvious that one thing meant is to signal, to convey some messages that are not stated, like dog-whistling. The election deniers are not saying out loud "of course I don't believe that election was stolen, what I really mean is to express solidarity with Trump". They are not saying "I dunno why this kind of statement is what I have to say, but I understand it works, tactically" Some of them probably do believe what they say. Currently/recently, it has been impossible for Wikipedia to untangle this situation. And not commenting, not providing anything, is fine and good.]

WikiAsk could monitor what are the questions coming up in wikipedia, where Wikipedia processes lead to failure to say anything. And jump in with pretty good answers that can certainly be given. Pretty Good Privacy (PGP) was indeed quite good, i believe it was really good, way better than people hearing of it assumed. Aiming for areas where "Pretty Good" is the best that can be done, could be a strategy. It does not need to be revealed/explained that is what Wikiask is doing. It might be best to avoid publicly saying the term "pretty good", but behind the scenes be working on identifying and delivering on those areas. --Doncram (talk) 09:55, 29 October 2022 (UTC)

My current vision of what Wikiask should be

• Answer difficult complex questions, such as the ones you cited, Doncram
  • Filling all the content gaps that Wikipedia has, yes
  • Also answering things about current events, where there are no satisfactory answers online
• In a way that is very simple to read, with:
  • Short answer
  • Headers that are full sentences
  • Bullet points
  • Images
• Highest-quality possible
  • The internet already has a ton of crap content, so we always need to be the most detailed and thorough to stand out
  • In the model of Wikipedia, Wikihow, and Fandom

-- Bill (Wikiask) (talk) 10:17, 29 October 2022 (UTC)

Ideas for questions

Doncram, I have started a spreadsheet with a list of potential good questions; please check here: https://docs.google.com/spreadsheets/d/11aA6z2Qi4tM-499XhYvwc-Q3GiGrmWD4Rz6fvSwAL5Y/edit?usp=sharing

My method:

1) Found the most popular pages on Wikipedia: see the "1 - Topics" subsheet and this http://wikirank-2022.di.unimi.it/?search=&type=pageviews&pageSize=10

2) For each topic, think about very interesting questions that Wikiask could address. Also, check Google's People also asked section for more ideas.

3) I've found that there are some question structures that could work for any subject -- see below. So, if we need to add say 10 thousand questions, we could just take 10 structures and combine them with the 1000 most popular topics. However, for now, I'd rather write questions manually, to keep the quality high.

I am compiling them in this document, and will upload them when we have the new version of Wikiask ready, in a few weeks. By the way, you can check out a preview at dev.wikiask.org.

Question structures that could work for topics that are people

• What is the daily routine of XXX?
• Which heroes did XXX have?
• Was XXX a good leader?
• How does one contact XXX?
• What motivates XXX?
• How did XXX make his money?

Question structures that could work for most topics

• What are some unanswered questions about XXX?
• What are interesting facts about XXX?
• How has XXX changed over time?
• What are the best things about XXX?
• What are the worst things about XXX?
• How will things likely change for XXX?
• What are the most surreal facts about XXX?
• What are XXX's strengths?
• What are XXX's weaknesses?
• What is XXX best known for?
• How famous is XXX in the world?

-- Bill (Wikiask) (talk) 10:17, 29 October 2022 (UTC)

I like the approaches to generating questions systematically. Maybe a set of generic questions could be produced that works pretty well for an "expert" in a field area, to apply to their field area. Including what were the most surprising-to-you things that you encountered in your study of this area. What are the biggest accomplishments in this area. And more along lines of above question sets. --Doncram (talk) 10:54, 29 October 2022 (UTC)

logic problems

(I dunno if this kind of stuff should be said out in public. This is effectively private right now, and in practice maybe this stuff can be hidden permanently, later.)

But, are there traps which could be anticipated, where Wikiask could not be sustained, relating to competition and license. Perhaps there are ways to survive that can be worked out, if things were sliding towards a given trap situation though.

What if random person takes full copy and posts it all, but with ads. This certainly happens with Wikipedia; there are multiple copycat sites. This seems to be not too much of a problem though, I think, as Google systematically favors Wikipedia in search results, over those copies, and perhaps it completely rejects all content of those sites. And the public largely understands, chooses to go to Wikipedia, but certainly not always.

But what if: what if another Q&A competitor, just harvested, each day, what Wikiask has added, and added it to theirs? Theirs would always be larger/"better"? Could/would Google give preference to Wikiask as a whole? Could/would Google keep track of "first provider" for each question and always put that first in results (i doubt that it could)? Would wikiask have advantage because it doesn't have the junky stuff that the competitor also has? Would wikiask forever be preferred because it runs less ads, forever, because it can afford to while competitors cannot?

"Copying and trying to present it so it looks different" happens i think with any public dataset, including the GNIS (the U.S. Geographic Names Info System?), including the NRIS dataset listing all the historic sites that are registered on the U.S. National Register of Historic Sites. They have some success.

What urgency would there be to come out with Spanish, French, Mandarin, Cantonese translations, can these be done at low cost. What do competitors do.

Wikiask contributors don't have to know about and be reviewing Wikiask's strategic plan. But how are some potential Wikiaskers to know/believe that Wikiask will persist? Or put the other way, how deal with doubts some potential contributers will have, about whether their contributions will persist. There could be several ways to address/answer, without the whole plan being everybody's business. My last comment for a while. --Doncram (talk) 11:23, 29 October 2022 (UTC)