Khoj App and Community UI Working Group
Oct 5th, 2020 Defining Questions and Strategies to find Answers
Denny and Rishabh attempted to answer two questions :
- What doubts do we have with respect to Khoj
- What are simple prototypes we can create to answer those doubts. A mural board for the discussion can be found here
- Given that in the next two months we can make do with just 4,5 contributors, its not very important to understand contributor motivation and designing software for them as of now. We should be able to move ahead by defining a simple Contributor Guideline.
- We should be comfortable defining what is out of scope for us. As long as we can define how do we respond to out of scope queries, we don't have to worry about losing users at this stage.
- The domains that need some more thinking are :
- Existential Questions - Agreeing on what the purpose of the app is.
- Understanding the User - Checking out the existing literature and also doing our own interviews to understand Ruhi Maasi more. Asking specific questions to understand nuances about her digital behaviour and technical abilities.
- Discovering new Users - Finding the right users that match our persona and talking to existing groups that already work in that demographic to get validation for our ideas and promote our app in.
- Novel Approaches - Besides the basic app functionality, what are some novel ways we can engage the users in the Khoj project. Each domain is described in more detail below.
The main question to answer about Khoj is that Is Ruhi Maasi deriving value from the service. The quickest way to observe this might be to start engaging her on Whatsapp. Can we send them a daily digest on whatsapp to remind of the service and field requests from them via whatsapp to see if they even care for this.
Lets do the basic and read existing literature on this. Rishab mentioned Anirudha Joshi's work on User Usage Model of Emergent Users. Lets also start talking to the Ruhi Maasis around us to understand their motivations better and their experience with technology.
We should start reaching out to existing groups who work with Ruhi Maasi Demographic and see what their response is like.
Denny talked about creating games inspired by the aesthetic of misinformation. For instance takking the common format of having a picture of a public figure and a quote next to them, if we paired a true or false button next to it, that would make users start questioning the format itself and hopefully this will transfer to how they consume similar media on other channels.
We need to start breaking down these tasks into doable chunks and assign timelines to them. To begin with, we should set up a simple whatsapp number to start interacting with Ruhi Maasis and a few contributors can start accessing the community UI to answer queries. We should also setup a minimal first draft of the contributor guideline to assist the contributors.
- Ground App Rishabh talked about this weekly news dispatch called Blind Spot thats run by this app. The email's content is supposed to surface articles that you might have missed due to your political leanings. So they have titles like "If you are left leaning, you might have missed these articles"
- Rishab mentioned if we should start documenting the broader themes from these discussions into a document that defines Khoj, its logistical, moral, ethical concerns etc and where we as a team stand on it.
- We were amused by the idea that a motivation for a contributor to join our platform would be that they might be too scared to get involved in fights inside their family group but with our platform they can help bust misinformation in other family groups.
- This broad idea that is Khoj a sense making system for a regular person.
Sept 7th, 2020
- Do people realise that content they encounter might be false?
- Do people care enough to submit questions to a third party to verify?
- If so, what are their expectations?
- Do we need them to begin with? Coud we focus on a fact checking black box first? To prove that this can be done. One benefit of doing the experiment this way is that then question depositors can be assured of answers to exist at all.
- Is it possible to crowdsource facts? (How efficient/qualified are 'professional' factcheckers?)
- If so, is it only because of the current structure of social media tht it is not done by itself?
- Counter example: Wikipedia. It does not have an impeccable record but it is pretty good.
- Do the policies make it happen?
- Stack overflow has high quality answers. Is this because programming is much more black and white?
- On the other hand there are things like Yahoo answers etc which have terrible standards
- Reddit is pretty much a mixed bag but still manages to have quality content show up
- What kind of facts can be checked by the fact checkers?
- If so, how does one go about it?
- annotations : requires fact check, in which case forward it to fact checkers requires googling domain tagging requires sensitive approach - translate original writing in one language to another
- link trusted sources
- ^ Adding attachments-done by experienced people in the field. How high are the standards we're looking for?
- summarize investigation - add consistency to the tone of the community
- vote on quality of responses
- Can we begin experimentation using MTurk? (Might be a budget constraint but easier to iterate than with a community)
- Why would the doers do it?
- Altruism
- Recognition
- Developing critical skills/media literacy
- Its fun(Game elements)
- Show how knowledgeable they are (Based on the BBC report) - Social Status
- That is partly why people upload on apps like Public, even wikipedia to a degree
- Also true for domain experts
- Protect their families