2015 May San Francisco

Register here!

Saturday, May 16, 2015 at 9:00 AM | Sunday, May 17, 2015 at 5:00 PM (PDT)

Nonprofit Tech Center

Universe of Topics

Education

• Educating "Beneficiaries
• how do we document success and lessons for others to learn?
• Getting literacy / competency distributed
• Educating elites/power holders

Scale Up/Out

• Framework for effective change, distributing solutions
• Integration into "real world" systems/institutions
• Scope insensitivity (maybe hyperbolic discounting, too) ("my contribution doesn't matter")

Homeostasis

• Error correction (can it ever happen online?
• Glorifying learning in public
• individual rights balanced with community health
• Solidarity with critique
• "Free Speech"

(Potential) Platform

• Prosocial UX: how the medium shapes the message
• Front end features
• back end support
• how do you design a social platform to discourage conflict?
• ethical platform/medium experimentation

Norms

• Social norms in hyper-permeable spaces (Durkheim's anoemie)
• Can we change social norms?
• How do social scripts break down on internet?
• How proposal traits get exploited (e.g.: first-round defectors love the norm of "always cooperate on the first turn")
• "listening period" for n00bs in online space
• Familial dynamics (parent/child, sibling/sibling)
• Empathy generation

Structure/Organization

• who does enforcement?
• Bridge figures/translators
• closed groups which deter racist/sexist/etc patterns
• Transactional analysis (Games People Play)

Discussion Topics

What is a social script, anyway?

Kind of important to the whole idea of shifting them, in this discussion we covered what social scripts are to us.

What is currently going wrong?

Using a facilitation method picked up from Tim Davies, we arced thru what we know, how that makes us feel, what our gut tells us, and what that means we might do.

Components and workarounds of dog piles

Individuals bringing their historical experience to bear on one example of a long-standing issue, without maintaining the sustained attention necessary to actually change the issue.

Are virtues scalable?

We discussed how communication-targeting virtues and tactics for improving small-scale interactions could be reinterpreted as strategies for larger social environments, with a focus on preventing or interrupting large-scale hostility phenomena without intruding on individual expression.

---

Small-scale, limited-time (eg one-off Prisoner's Dilemma):

• "virtues" as codified by algorithms are not very useful because they rely on a timeline for effect
• prisoner cannot affect whether to be punished, only whether to reduce punishment by defecting if it happens
• most rational choice is to defect (for a value of rational with purely individual and no social considerations) regardless of opponent's choice

Small-scale, repeated interactions (eg iterated PD): see http://www-personal.umich.edu/~axe/Axelrod_Evol_of_Coop_excerpts.pdf or http://journals.plos.org/plosone/article?id=10.1371/journal.pone.0080814#s2

• "virtues" (niceness, forgiveness) become more effective by swaying future interactions
• player can choose to risk now for reward later, or vice versa; more choices, but muddier connection between decision and consequence
• most rational choice depends on opponent's strategy
• iterative learning becomes possible

Large-scale, repeated interactions (eg societal iterated PD) see http://amirrorclear.net/academic/papers/sipd.pdf and http://amirrorclear.net/academic/ideas/dilemma/index.html#OrdBlair

• "virtues" start making major differences in propagation or elimination of strategy-populations
• individual player has very little control over consequences, only choices
• most rational choice depends on large datasets like relative population of exploiters vs naive players
• social learning (and enforcement) becomes possible

The social strategies we use intentionally, particularly those we enshrine as character values, are often geared toward the lower end of this scale -- one-off interactions with a single other person, repeated interactions with another person or small group. Strategies best applied at the upper end of the scale, toward whole communities or societies, tend to be categorized in an practicality-agnostic framework such as morality or religion, sometimes more just-so-story than ethical reasoning, and without rigorous examination outside special fields of study. The typical person's strategies for dealing with social media, for example, are not designed for the scale and scope of that environment and don't have understandings of its muting and amplifying effects built in. In some fields where such strategies are examined specifically on grander scales, the lack of consideration of the shaping effect of current media may prevent direct translation of lessons learned into engineerable practices.

Game theory offers one way to examine the scalability of these virtues: What is the effect of forgiveness on an entire society? (It's great for the society but possibly bad for the individual.) What is the effect of loyalty on an entire society? (It can be a survival mechanism for the individual, but crushes diversity in the society.)

One virtue the conversation gravitated toward was that of self-reflection, particularly a pause for self-reflection in the middle of conflict-exacerbating or hostility-reinforcing patterns.

• carrying forward the emotion of one negative encounter into another introduces hostility "noise"
• negative emotions can provide energy to redirect toward positive goals and shouldn't be dismissed lightly

Reflexive/intuitive thought can tend more toward the former, increasing conflict more than producing solutions. Measured, effortful thought can help emphasize the latter. However, it's the reflexive/intuitive system that determines what's worthy of effortful thought. How can people be encouraged to think about their anger, hurt, or other negative emotions and how to direct them productively without transferring them to the next person or situation that comes along at random, and without taking out frustration with an entire pattern on a single example?

Detailing components of communication

We laid out various aspects of how communication and community occur. This will feed into how to build a pro-social platform.

Projects

• Communication Karma Score : Matt suggested a way to get credit and signaling for being a good conversational partner, for using logical fallacies, etc
• IntrospectionBot : just like a desktop reminder to occasionally stretch or go for a walk, this would occasionally prompt a moment of self-reflection. We hope this would de-escalate
• Visualizing solidarity : when outrage happens online, it's often because people are dog piling on an exemplar of a historical issue, rather than using the moment of attention to direct energy towards understanding and addressing the historical issue.
• We discussed (and hopefully contributed to) TQ's Pro-Social Platform, and associated Parameters of Interaction as related to Sands and Ethan's framework for social networks.
• Quiz to comment : to ensure a commenter is both not a robot AND has the context of the entry, a captcha of sorts would be necessary to indicate having read the full entry before commenting.
• Aggregated comments : rather than dealing with every negative (or positive!) comment which occurs, instead aggregate with "there are 47 other comments similar to this one"

Things for further perusal

• Squelching Plugins : mute anything which matches {predicate}
• Individual accessibility settings : we're ok with Out Of Office responses in emails, why can't we signal that we are stepping away from social media?
• Data-driven introspection for communities and individuals based on data generated from interactions
• Hedging versus accountability
• Saving face
• Different intents of education (providing data) vs correction (proscribing behavior)
• Providing labeling tools for airing grievances
• Handshake before mutually beneficial argument
• spaces for relaxed, civil group chat ("hangouts")
• Sentence structure (conversational style) / word choice (filter bubble) machine learning as curated by an individual
• Introductory / recovery circles or levels
• 9.5 Theses toward Internet Reformation: An Anti-Manifesto
• Monkeys and electrified ladder. GR Stevenson
• Targeted ads on Twitter. Look at how cheap and easy you've made it to put terrible things into people's timelines.