Yizzy Wu

Method In order to gather and analyze the information, I printed out a poster with a classic trolling argument under a Youtube video that talked about LGBT and Marriage Equality issues in June 2015.

*see the video titled: #ProudToLove - Celebrating Marriage Equality and LGBT Pride Month. https://www.youtube.com/watch?v=WSiehK2asbI.

I choose this issue as a sample for experiments because it is a topic that is being widely debated and apparently controversial in different viewpoints. Also, the communication under the video evidently shows the pattern of what I am looking for, which aggressively filled with trolling and bullying language.

The testers who engaged in this experiment were asked to read the comments and to annotate their feelings reacting to the comments.

Content The instructions that I created for the testers are as the following:

READ through the comments to see what people were arguing about.

HIGHLIGHT the words or sentences that you consider inappropriate or whether makes you uncomfortable from a mental aspect.

WRITE DOWN the reason why you think it causes feelings in a negative way.

RATE them each from the right side of the comments.

WRITE DOWN what you think according to those comments as a spectator.

Also, if you have a similar experience of being in one of the respondents, feel free to SHARE your opinion about your feelings underneath.

Let me briefly explain the incident that took place. There were four people in this conversation. The first one, “Announcer A" was expressing her own feelings and perspectives reacting to a Youtube video that talked about LGBT and Marriage Equality issues. And three other people, “Respondent B, C ,D" are the respondents to her comments and they reacted and expressed differently. “Respondent B and D shared a similar interest and aspect, while “Respondent C" is the one who started to provoke the others and used profanity and offensive language to spread their strong venom, and tried to make others become emotional and arouse anger.

The incident started with “Respondent B” first thumbed up to A’s statement, and elaborated his own thoughts. Then C came in and dropped the bomb. “And you haven't killed yourself yet? Or maybe you're waiting on the AIDs to end it.” said by C. The conversation went terribly vindictive since then. The following visualized graph presents the relationships of this conversation.

Interview Before testers (annotators) begin their trial, I asked them several questions to understand their possible characteristics as the users.

Generally, do you care about social issues? 

Do you usually leave comments or react to the post / video on social media? 

Do you ever read the comments of the post on the social media or news press? Have you noticed the trolling activities online?

How do you usually feel about the conversations with hate and malicious meanings?

Outcome / Findings There are three roles (characters) that I found from the experimentation, whom could be the target audiences of the project. They can be generally defined as the following:

/Participants/ 1. The one who always involved in the arguments, either like to share opinions (or making statements) and persuade others or who just can’t resist the rage to fight back.

2. People who like to provoke, assault, and troll the particular ones.

These people may have risks of having the mental illness causing by over engaging the arguments or fail to hold a healthy attitude during or after the conversation.

*Their behaviors and actions are included in a pattern from the consideration.

Poking with mean & Sensitive wordings: people tend to be mostly sensitive with the negative words rather than the good one.

Correction & Persuasion: people have the imputes to correct other’s errors such as typing errors, wrong terminology, and misunderstanding viewpoints.

Repetition & Preference of words usage: people are subconsciously (or unintentionally) reuse the sentences or words they accustomed to say in their daily lives. like a habit.

/Spectators/ People who aware issues and the abusers on some level, but never get involved and take part in the arguments. They always remain silent. However, they could be in some levels emotionally affected by the fight, and potentially become the participants if the conversation triggers a point of their anxiety.

/Outsiders/ People who rarely care about social issues, no comments or public reactions. Though it is also important of how they think about these arguments / abusers, since they do active on the Internet. Moreover, a mission should be declared to stimulate people’s sense of obligation of knowing and defending cyberabuse.

The most well known controversial subject, topics or areas are politics, religion, philosophy, parenting and sex. History is similarly controversial.

Other prominent areas are economics, science, finances, culture, education, the military, society, celebrities, organizations, the media, age, gender, and race.

NEXTDOOR.COM AND RACIAL PROFILING

Nextdoor is a social network for the neighborhood. It was created to provide a handy service and friendly environment for people to share information and support each other in their daily lives. However, it had become a place for racial profiling.

The platform of Nextdoor presented a free-tag posting feature for people to classify their reports to their communities, which was a valid approach as a concern of usability. Though the “Crime & Safety” forum has grown into an alternative path farther away than they expected, according to the report from Fusion.net in March in 2015. 

The article reported by Pendarvis Harshaw from Fusion.net, articulating a typical case about the warning messages and comments that happened on Nextdoor.com in East Oakland’s Ivy Hill neighborhood on a Saturday evening, 2015. A white girl named Meredith Ahlberg invited her friends to her house for a party, while she noticed her phone with notifications claiming there were “suspicious African American” rang their doorbells asking for a wrong address, and the words used from the poster comments were “Scary sketchy” and “call OPD?” whereas these messages were targeted the friends that Ahlberg invited that night, and according to her words in the interview, they are “ridiculously handsome and stylish.” By the time she saw the posts, her friends had found the correct address, and Ahlberg was looking right at the ‘thin, young, black man.’ As Ahlberg reported back to the discussions for clarification, the comments responded were “That’s a relief,” reacting to this false alarm.

“While Nextdoor’s ability to assist in crime-spotting has been celebrated as its “killer feature” by tech pundits, the app is also facilitating some of the same racial profiling we see playing out in cities across the country.” said by Pendarvis Harshaw in the article. That is, either the conveniences that Nextdoor has bridged the gaps among neighbors or it arouse the potential racialism to communities of the living area. Other discussions from the local bloggers in  another Oakland neighborhood were posting as “Racism quietly flourishes in San Leandro,” or “Nextdoor: In case your Facebook feed isn’t racist enough,” wrote by a woman in Wisconsin as a post title on Tumblr.

MODIFICATION

A following report from Fusion wrote by Kashmir Hill associates with the previous discussions about the racial issue on Nextdoor networks talks about the actions that the company had been trying to improve their profiling system. Accompanying NPR news press wrote a relevant article about how Nexdoor tried to cope with the situation.

And here are the methods they did as an annotation, testing by inside and outside the company and undertake to tag and rate the suspicious racial profiling.

Crop from Fusion.net

Crop from Ｗeird.com

A pilot project using algorithms to check for racially charged terms has been brought out. The modification they are attempting to make is to add a section of “detailed information”. Although people can still freely reporting their “suspicious aspects,” the new form that the users will see when they mention race in a post may request them to involve more specific information such as hair colors and clothes wearing.

An algorithm works as a detector to spot checks the terms and length used in the posts that contain the suspicious activity for racial profiling. The system may presume to lack meaningful details and block the draft post if the post has inadequate or unspecified contents.

AFTER EFFECTS

The company then tried to analyze the variation and the user experience from their change to the system. 

Crop from Fusion.net

While it is hard to define which post is racial or not, they have collected thousands of posts by the end of the summer in 2015. “According to the categorization done by the manual reviewers, there was a 75% reduction in racial profiling posts by people using the new version of the site, compared to those in the control group who were still using the original version of the site. So the company decided to roll out the new version to all of its users.” excerpt from Fusion. On the other hand, Tolia says they’ve also seen a 50 percent increase in abandonment—a lot of the people initiating Crime and Safety reports aren’t finishing them. 

“We tried to create decision points, and to get people to stop and think as they’re observing people cut down on implicit bias,” a quote from the interview (Fusion) said by Nirav Tolia, Nextdoor’s CEO. Comparing to other social networks such as Facebook, Twitter, and Snapchat who don't have any frictions at all the process of posting, Nextdoor is going on the opposite and unusual path. While other companies don't modify their product to respond to the complaints of bullying, hate speech, revenge porn or other cyber abuse activities, Nextdoor was taking their first step that reacts to the existing problem.

FRICTION AS A DESIGN PRINCIPLE

Article from NPR and Wierd both mentioned the principle of friction in user experience. 

Since Nextdoor claims that Ninety-nine percent of Nextdoor’s racial profiling is inadvertent, based on Fusion’s story and the research, whereas their interface did nothing to stop them, the company spent 6 months to build one that could. And they operate the design in a reverse direction by breaking a conventional rule of design principle. 

It has been a widespread discussion of reducing frictions in the user experience. Most of the arguments may support the idea of that friction could damage a user experience because it may increase steps and slows down interactions, however, the situation which Nextdoor had faced might drive them into a contrary course. 

In order to moderate the “unintentional” racism, creating frictions to the user interface can be useful. Comparing to the old interface (a standard white box as the single entry point), which left users an easy way filling out and post up, the new interface takes the lead with guidances by using checkpoints and informative confirmations to help users to evaluate their content of their reports. The options include crime, suspicious activity, or other. 

Crop from Nextdoor.com

To anatomize the procedure of the profiling, I went through those steps and would like share my experience here. Let’s assume if you click “suspicious activity”, a list of tips will appear on screen, including one reminding you to focus on behavior, not appearance; and it notifies that racial profiling is expressly prohibited as well. Then you will be given up to three steps by filling particular information before finish posting. 

First, you may be asked about the incident or situation that you want to report, there is also a checkbox of informing whether you already report to the police or not. The interesting segment of this part is if you type in something like “AA (American African) man checking car door handles…” and hit Next, it will appear a red line says "Please remove descriptions of any people involved and add them in step 2.” The system can automatically sensing the possible racism in your words. 

Then you may move on to step 2, specifying the details of the person or vehicle. In the person section, it provides 8 fields for particular inputs, such as hair, clothes, shoes, and other basics like age, sex race, or other details. Noticing there is only one field that does with race, whereas it is the most critical part in the section. According to my own experience from testing, if I fill out boxes like age or sex, I can easily pass through this step to the next, but if I include the “race” field in my report, I may be required to include more specifications or the algorithm may forbid me from advancing. 

Finally, you may be asked to provide an overall description (more like a summary). By the constant efforts from those steps that the users go through, the final report may present a more profound and helpful look from the original reports (the time when people use the old version of the interface). 

A relevant discussion seems interesting to look at. According to Weird’s article, a UX expert Samuel Hulick compares Nextdoor’s redesign to Civil Comments, a peer-review platform that requires any user who wants to comment to first review three other comments. 

“It’s the online equivalent of taking ten deep breaths before picking a fight.” 

On the other hand, Weird mentioned that Nextdoor suffers from the tricky balance between kind reminding and strict admonishment. However, Hulick still appreciate the efforts of the company. “Anytime you’re filling out a form online and it admonishes you for not matching up with its own criteria, that’s a universal usability problem,” Hulick says. 

“But this was an admirable decision. The intention behind it seems to be a humanitarian one.” 

Excerpt from Weird.com.

Although the new form apparently did cut down the instances of racial profiling, the company knows that it won’t end racism. “But the company is convinced it will reduce the amount of bias aired on its platform, and thus contribute to its mission to create and foster close-knit communities.” “It’s a mission that tech companies are increasingly embracing.” Although it seems to be a stiff sailing to meet the ultimate goal, but Nextdoor feel a moral and business obligation to be part of the solution. 

SOURCE

Fusion: How Nextdoor Reduced Racist Posts by 75%.

Fusion:Nextdoor, the Social Network for Neighbors, Is Becoming a Home for Racial Profiling.

NPR: Social Network Nextdoor Moves To Block Racial Profiling Online.

Taiwan has the problems among immigrants. They separated into different groups based on where they came from and when they move to the island. Generally, most of them emigrated from Mainland China except some were moved before 1945 (called "islanders") and some were after ("Mainlanders", who moved to Taiwan after Japan resigned from WWII). The slightly difference is the time. The ethnic disintegration and opposition between two groups as well as the people from Mainland China never stop, based on the disagreement of cultural distinctions and how they see the country differently. To me, I stand at an awkward position since my grandfather is an islander and maternal grandfather is the Mainlander, which made me a hybrid of them both. So when it comes to a fight, usually I don't know where to stand, because I understand the persistence between both sides.

Take another case as an example, LGBTs. This issue is not only attracting concerns in the US but also intensely debate in Taiwan. likewise, I still in a weird place. Unfortunately, I grew up in a religious Christian family while my best friend decided herself as a lesbian when she was 15. I was happy for she finding her true self, but I have to lesson to my parents and elders from the church saying that what my friend did was sin.