The conference website is at http://caae.phil.cmu.edu/style/Seminar.html. The agenda is at http://caae.phil.cmu.edu/style/agenda.html.
Later conferences in this series are linked from http://www.online-deliberation.net/ .
Another participant's notes are at: http://web.archive.org/web/20060208073556/rescomp.stanford.edu/~bshanks/cmu.pl/NotesByElisabethRichard
See http://caae.phil.cmu.edu/style/presentations/Shane.ppt for the ppt.
Peter M. Shane
Director,
Institute for the Study of Information Technology and Society
Both RD and DD propose to legitimate government by assuring full and fair consideration for everyone's interests; difference is in the mechanism :RD: Accountability of public officials :DD: Opportunity for each citizen to be heard in a meaningful way
Developing and Testing A High Telepresence Virtual Agora For Broad Citizen Participation
If IT innovations in democratic practice are going to enhance legitimacy (through inclusiveness and depth of consideration) and engage citizen attention, they must connected to actual decision making
Electronic Rulemaking represents significant use of IT, integrated with an official decision making process see www.regulation.gov
But:
Challenge = Absence of baseline info on information processing in RM
See http://caae.phil.cmu.edu/style/presentations/Cavalier.ppt for the ppt.
Robert Cavalier Carnegie Mellon
Diagram with 4 quadrants. Vertical axis goes from Madisonian at top to Majoritarian at bottom. Horizontal goes from Direct at left to Representative at right. "Nondeliberative" is in Direct-Madisonian (top-left), and "Deliberative" is in Representative-Majoritarian (bottom-right).
e.thePeople is a public forum for a new democracy conversation. Our technology promotes intelligent, diverse and deliberative discussions, both here and among a network of sites around the Internet.
TODO: link
TODO: add screenshot
TODO: add screenshot
Venn diagram: Two circles
Circle 1: Issue Discussion
Circle 2: Issue Representation
PICOLA/Deliberative Poll is in the intersection of the two diagrams.
Delibera software is outside everything with a questionmark.
Venn diagram: three circles
Circle 1: Immersion
Circle 2: Agency
Circle 3: Transformation
"User role" is in the center of the three diagrams
A few slides; looks cool; check 'em out!
TODO: post those pictures here
some deliberation/online deliberation methods:
EU is pouring tons of research into this :(one site from EU: http://www.eve.cnrs.fr)
In Peter Muhlberger's survey on why people talk politics
Pew survey: people who are online are more trusting of "the system" :(so, maybe online deliberation won't help draw more people into government, because those who felt most excluded before won't (be able to?) participate anyway)
not everyone is skilled in argumentation :there are other ways that people express themselves (i.e. telling a story, etc) (this argues for telepresence as opposed to text-based interaction)
Most online communities where discussion and deliberation takes place are self-selecting, that is, anyone can come along and sign up. Fishkin(sp)'s Deliberative Poll tries to combine the advantages of a (scientific)/(representative sample) poll with deliberation. In essence, the Deliberative Poll attempts to take a poll of what people think of the issue after they've deliberated about it with others. So, in a deliberative poll, a representative sample is chosen to participate.
example of a potential problem with self-selection: some poll was held to pick the most important man of the 20th century. The winner was Ataturk(sp), a Turkish leader. Turns out Turkish people had gotten real excited about the poll and got a lot of Turkish people and students to vote for Ataturk.
concern: if this is used to govern, make sure that the people not in the representative sample still have an avenue by which to contribute.
Unchat: text-based chat system (i think! todo). One person speaks at a time. There's a queue to speak, but everyone is also given K "shouts" which they can use to say something immediately (interrupt) (K is a configuration option).
Someone found it hard to manage.
comment (that i thought): a similar system without telepresence may take off exponentially b/c people could use it at present, without access to cameras and bandwidth (such a system would basically be an augmented chat client, i guess)
comment (i think i said this): in a deliberative poll (and in your system), can participants suggest reading materials for others to read? (the question was deferred, but I think in their system, at least, participants can send links)
robert cavalier's email should be rc2z@, not rc2c@ (?)
See http://caae.phil.cmu.edu/style/presentations/Muhlberger.ppt for the ppt.
(these slides are copyright by:)
Peter Muhlberger
Research Director For E-Democracy
InSITeS?
In an experiment, the PrisonersDilemma? game was played by two groups. In one group, it was introduced as the "wall street game", and in the other group, it was introduced as the "community building game". In the second group there was more cooperation. Conclusion: context greatly affects the cooperation and community-mindedness of a group.
People who had a website were wondering what attracts and drives people away from that site, so they had a poll on the site to see which things people most liked or disliked. But, the people who answered the poll were predominantly the heavy users of the site. Muhlberger pointerd out that it is difficult to tell what drives people away from the site if you have data only about who stayed. He suggested instead targeting the poll at those who had recently signed up (as they were most likely to leave).
A few parts of a definition of when people in a deliberation have "equal standing" are given. One of them was "each person has a willingness to discuss with any reasonable person".
Muhlberger talked about the idea of a web congress with three parts:
(to be tested): give people anonimity for awhile at the beginning in order to raise issues, then enforce non-anonimity later on. For instance, in the web congress, anonimity in the issues congress but not later on.
Social science research shows that people are reluctant to say information which is unique to them. That is, if they are the only person who knows something, they will often not say it even if it is very important to the group's decision. This tendency should be countered if possible.
idea: have people at the beginning say "i think this" (rather than waiting to see what other people are talking about and maybe not saying anything)
my comment (didn't say): but what about unique information about a subtopic of the conversation?
Social scientists talk about a virtuous spiral (positive feedback loop) between three variables in a group: political knowledge, political action, and political sophistication. If the group has a lot of any one of these, the other two tend to develop, but if it has none of them, none tend to increase.
Some ideas (see notes from first social science breakout session for more):
:bayle: after the conference I looked at this on Google (I skimmed this and then skimmed the first 3 pages of this. It seems to me that one can have a good discussion/deliberation without high integrative complexity. For example, in the example with handcrafted goods in the second link, it seems to me that the first or second statements about why handcrafted goods are expensive (the second one is "Handcrafted furniture is expensive in part because there are few skilled artisans and in part because most people do not have the good taste to appreciate high quality work.") would be more helpful to a deliberation than the later ones (for instance, the example of score 7 is not very concise and brings in so many factors that I think that discussion about a problem would be hindered rather than helped by this level of complexity).
(see social science session for discussion of this book)
Vincent Price is a big man in the social science of this stuff.
See http://caae.phil.cmu.edu/style/presentations/Easterling.ppt for the ppt
Technology Decisions and Tradeoffs in Designing an On-Line Deliberative Environment
Stuart Easterling
Software Engineer, InSITeS?
Carnegie Mellon University .
Problem: How best to securely and seamlessly integrate client-side Flash movies with a J2EE back end?
:Allows for future swapability of client-side technology
Problem: How best to capture and store both quantitative and content-based social scientific data?
clickstream data user inactivity (timeouts) survey answers survey response times content ratings emoticon use discussion participation poll results postings and chat logs etc
comment by me (unsaid): The Delibera demo looks really fun.
How to "help society think bigger thoughts"? :(see also Kagan, "In over our heads") TODO: lookup ISBN
example of use of mess maps:
made a mess map of community problems. traced the cause of increased numbers of mental health patients in jails (something like the county jails were currently the biggest providers of mental health services) back to increased numbers of homeless people on the streets, back to the departure of droves of experienced caseworkers, back to increased paperwork that was driving away caseworkers (who are people persons), back to the conjunction of new federal medicade paperwork requirements with a lack of computer programmers that year due to Y2k.
one committee opted to turn in only a mess map instead of a 60-pape report; town councilmembers liked that
Other things like mess maps:
Horn has a book "Visual language". It is written with visual language (cartoons/diagrams rather than paragraphs of text).
Vincent Price's self-introduction: they put people online and just asked a few questions to start out and then let people chat (w/o structure), and it looked like good deliberation. So, he recommends not prestructuring deliberation too much. Also, feels that trying to control too much may actually make the group harder to control (parallels to SoftSecurity?, and specifically AvoidTemptation?)
Studies of small groups have shown that deliberation tends to cause polarization. That is, when a group discusses something, it's median opinion tends to become stronger in the direction of the median opinion at the beginning of the discusssion.
In more detail, take some group and before they talk, poll everyone about their opinion on an issue, with "0" being a neutral stance, "-10" and "10" being the extremes of one side or another. Now let the group discuss the issue. Afterwards, poll the group again. You will often find that the median opinion has become more extreme in the direction of the original group bias. For example, let's say the median opinion of the group is 3 to begin with. Maybe it will be 5 after the discussion. The studies that deal with this use issues where the quanitification of opinion is not a problem; either there is a binary decision to be made, or the decision itself is quantifiable (how much funding for some budget, for example). (TODO: diagram)
Perhaps this will be referred to as "inter-group" polarization to distinguish it from the system of "intra-group" polarization where after a discussion some people may be more strongly in one direction whereas other people end up more strongly in another (statistically, an increase in varience of the distribution of opinions).
Sometimes something else happens; an increase in group varience, but a common tendency. For example, if a group had two members, A and B, and A started out at 5 and B started out at -5, and then after a discussion A was at 3 and B was at -9. Intra-group polarization (varience) has increased, but there is a trend in the direction in which the group members opinions has shifted (downwards). (is this desirable? does the direction of the trend indicate the 'right' decision to make?) (TODO: diagram)
Obviously, polarization is troubling; for example, if the conservatives and liberals hang out in small groups of their own kind a lot, one would expect their views to become more polarized, making it harder to acheive consensus across party lines.
Price: In small groups, a push to consensus paradoxically leads to more (intra-group) polarization. This effect is greatest when a binary decision must be made, for example, on a jury. Perhaps the push towards consensus in such a situation actually makes people go into a goal mentality where they feel that they know which decision they want, and they try to acheive that decision.
Horn: Steve Rosel (sp?) & Daniel Yankelovich
Price: We want consensus, but sort of by definition the study of "public opinion" is the study of issues for which public opinion conflicts. If there is a consensus we call it "culture" rather than "public opinion" and don't study it in this field.
Republic.com claims that online discussion is the worst mechanism in terms of polarization, because
However, H. and Jennifer Stromer-Galley both found evidence that this doesn't happen (because self-selecting groups weren't so homogenous after all)
What are desirable outcomes from deliberation (and how do we empirically test for them?)
Note: these may apply at the level of the whole discussion group, or maybe at the level of the factions or subgroups of parties which feel similarly about the issue)
It is accepted wisdom that people flame a lot online, and more than in real life, but we're not so sure. Some participants did not find evidence of increase flaming.
One example. Someone told a participant (of this breakout session) about an online soap opera discussion board where no one flames. This claim was met with disbelief, and the claimant had not visited the site since '94 or something. But when the participant checked the site today, there was no flaming. Conclusion: social norms can create a community with or without flaming.
Asynch con:
Asynch pros:
Advantages of telepresence over meeting in person:
Pros:
(I've never seen this before, but Chris Lampe and Jennifer Stromer-Galley had. Chris was kind enough to explain it to me, but I fear I've gotten it wrong)
Two axes: synch vs. asynch and distant vs. local. (or maybe the second one is "collated vs. not"?)
examples of what belongs in each quadrant: :synch, local: meeting :asynch, local: file cabinet, whiteboard, memo :synch, distant: online chat :asynch, distant: email
(where does a discussion board or a wiki go? i'd say asynch, distant)
(if you wanna do (collated vs. not), i guess you might say: :synch, collated: whiteboard at meeting :asynch, collated: wiki, file cabinet :synch, uncollated: chat :asynch, uncollated: discussion board, email
although this is what i would say, not what Chris said was on the standard chart)
Price: problems w/ deliberative polls
The three groups from the three different breakout sessions had reported summaries of what they had discussed back to the plenary session
People don't get excited about participating until a key decision is being made
goals - effecting government or developing democratic capacity
One question we were trying to answer: how to spend $200,000 to do something useful in this area?
(at some point around here we talked about threading as a mechanism for scaling discussion, and I mentioned wikis as another alternative)
(since this was my group, I tried not to duplicate notes that I have above) ... Why would you ever want to have synchronous, electronic (but local) conversations (as opposed to just meeting in person)?
(How to discuss complex issues?)
:either personal or automatic
:uses: report to policymaker help for facilitator standardize mental models between participants
:(note: I think a wiki would be a great summarizer tool along the lines of what we talked about)
Muhlberger: Research says that in any decision, people will consider only three things.
Muhlberger: Unique information sharing (see above); some potential reasons:
Lampe: When does complexity occur? What size?
Kelly: Just below 10 people
Muhlberger: Group of size 12 is about the limits of a "small group"
Lampe: (used slashdot as an example, suggested distributed moderation as a way to scale small group discussion)
Lampe: I question the underlying assumption that every deliberation must be successful
Muhlberger: Even without external structure imposed by a moderator, groups self-organize, too; groups start out brainstorming, then later analyze
Maybe this is harder with people with very hostile/different views
Lampe; Price said he did hot-topic deliberations online and it went fine
Lampe: Also, if the time-frame is really long, people may have time to self-organize and thereby reduce hostility
Muhlberger: And maybe less time pressure leads to less hostility
Kelly: As moderators, we try to structure w/o suppressing
We try to help minority views get heard, and try to tone down views which have been unfairly entitled (for instance, views which were overly/unquestioningly accepted either because of the status of the speaker or because of recent world events)
(also, Kelly described his scenario-planning system, which sounds cool but I wasn't able to write it down fast enough)
Muhlberger: Problem with rating systems: minority viewpoints?
Lampe: In slashdot, there's lots of people on either side of an issue. So I think minority viewpoints are given a fair airing.
Kelly: One heavy-handed faciliation technique that we sometimes use: Secretly recruit some people who are on side A to pretend to be the opposing side B of an issue for awhile, i.e. argue as if they are on side B, with no one but the person themself and the moderator knowing that they are pretending.
Muhlberger: Some people don't believe that politics are debatable (i.e. that there is one clearly right answer, their side). Stages that such a person goes through when talking to someone with a different point of view:
Muhlberger: Many such people believe that politics are a private matter. I don't understand how anyone can think this. I mean, the very essence of politics is disucssion of conflicting points of view.
? (I don't know the name of the woman who said this): You are in stage (3) with respect to those people, then? :)
Hamlett: Explain it to them again, louder! (not so funny in print I guess but this and the last comment were really funny at the time; here, Hamlett was sarcastically telling Muhlberger to just talk louder at the private politics people until they agreed with him)
Hamlett: We'd better refute Hibbing very soon or it'll become the accepted consensus view.
Hamlett: very soon. Example: We got a provision into the U.S. house version of a nanotech bill which required consensus conferencees/deliberation. However, in the course of getting that provision into it, some research was mis-cited to argue against the provision. So, soon Hibbett will be cited in such discussions.
(various people): Hibbett's book sucks! It's polemic, not social science.
(someone): There's also a bias towards representative democracy, as opposed to direct democracy, in the field of social science today.
Lampe: I firmly believe in microcontribution (that is, people should be able to make a small contribution to decision-making on an issue, rather than being forced to contribute nothing or to pledge tons of time to it.
:Described when he and others used the slashdot system to vote comments up to policy makers. Everyone entered their opinion/policy suggestions on an issue, and then voted on which comments they liked, as per slashdot. The results were much like the results of a slashdot discussion. Policymakers were presented with the top 5 comments. (One disadvantage; the top 5 comments were not necessarily the most lucid presentations of their point; there was definitely the usual winner-take-all effect where the first comments to be posted had a huge advantage)
Muhlberger: From Price's #s, it seems that about 8% of his original sample shows up at a given session.
:When I sent something to a bunch of students asking if the'd like to come and discuss politics (no $ offered), about 40% said yes!
Motivations to contribute
Muhlberger: We want to let little groups form their own discussions.
?Problems?
Davies: I've observed that deliberation breaks down when you put together people or groups without shared assumptions on deliberation norms
Hamlett, ?: meta-level
?: need a starting process to see what goals, etc are in common
?: Economist article on Fishkin (TODO: link) (he brought together environmental groups and petrochemical groups, (in a deliberative poll framework??? TODO: check)
Davies, Muhlberger: stating discussion norms at the beginning makes a big difference (see also PrisonersDilemma? studies mentioned in Muhlberger's plenary lecture where people are given different contexts)
Kelly: There's a lot of relevant moderation techniques. We like techniques that provide structure and the beginning and then fade out.
:it's good to allow competitive games (outside the subject area of the deliberation), like who's diagram is better or something like that.
:also, it's nice to switch people between groups later on (I'm not sure exactly what this note meant, but I assume it was in a context where there were multiple subgroups debating the same issue)
Muhlberger: Can you just give a list of criteria for a "rational person" and get everyone to agree to them in the beginning? (I think he gave examples of some such criteria, too)
?: What these are is a set of criteria for a rational process, not a rational person
?: When we moderate discussions, we get everyone's name and email and introduce them, and they become friendly and civil. I feel that that is sufficient without explicitly talking about rationality.
Muhlberger: Yeah, sometimes maybe you want people to share narratives; as we said, some people may not be good at rational argument. They form idea but aren't consciously aware of how. But, if you come to a conclusion intuitively, it's hard to explain + persuade someone else
Kelly: Encourage rationality and explicitness, but have tools (drawing, drama, etc) for people better with other modes. Use these other-mode tools, and ultimately move towards words as an end product. The end product may be largely rational but may contain components in other modes.
?: One objective is ... (???)
Lampe: Iron (???)
Davies: Back to the rational process vs. rational person. I think the rational process will be a wedge that will lead into buy-in to a personal identity later. There mere act of making a concession or agreement as part of entering a group strenghens one's identification with the group. Example: At a buddhist retreat everyone put on monk's robes at the beginning. (this argument reminds me of MeatBall:UseRealNamesGauntlet?)
Muhlberger: Classical economics assumes fixed personal preferences. Deliberation, by contrast, is about transforming preferences
Muhlberger, Kelly: Some might try to use deliberation to get to know your enemy. Kelly: In fact, sometimes they advertise our business like that. I was uncomfortable with that, but now I'm ethically OK with it b/c more often the participants become more open to the other side and change their own points of views (not to the other side's point of view, but to something new).
Muhlberger: Some studies showed no increase in empathy by deliberation. But maybe that's different from "respect for your point of view".
Davies: Example of the benefits of intrinsic rewards (as opposed to extrinsic rewards). Intrinsic rewards are when you do something because of some internal motivation, extrinsic rewards are when you have a direct external motivator (like money or a job assignment). The example: a study had two groups of kids. One group played with magic markers, while the other group had a contest for best picture drawn with magic markers. After this period, both groups were allowed to do what they wanted; magic markers were still available should they wish to draw. The kids in the contest group were uninterested in playing with magic markers at this point, but the play group still played with them.
Muhlberger: Interestingly, educational theory social scientists find in their studies that extrinsic motivation decreases motivation (in various ways), but business school people find that it increases motivation (in various ways).
Muhlberger: How to use social processes for "division of labor" within large, complex discussions?
Kelly: Sub-committees
Kelly: (described some group "tipping points", i.e. trust, etc. Sorry, I couldn't write fast enough here)
http://www.it-democracy.com/ncp3/tutorial.jsp
Vivarto
Netconference Plus
personal note: I think vivarto is AWESOME!!! This system should be used ASAP.
Note: vivarto allowed you to delegate your voting (transitive voting). They have a fancy name for this that implies that they made it up, but it sounds just like MeatBall:LiquidDemocracy? to me.
A political party somewhere in the netherlands actually agreed to vote the way that their members told them too via Vivarto. There was a study ... (couldn't write fast enough) (TODO: ask about the study)
Some sites involved with e-democracy:
TODO: add/send link to candidate summary site
Idea: build an argument map collectively (personal idea: this could easily be implemented in a wiki)
in it.comm (is this a magazine or journal or something?) :the political issue that least engaged people was the budget (which is of course one of the most critical issues)! (but they were excited by buildings), so maybe fun is a big factor
TODO: post on MeatBall? about instrinsic motivation and agency
Richard Scwader's Thinking through culture
On respecting someone's point of view (vs respecting the person) :How can an anti-abortion person respect abortion, i.e. murder? Also, if you start out by taking the bible as literal truth, how could you respect someone who even puts the dividing line in a different place?)
Deliberation may be able to escape Arrow's Theorem by changing people's preferences
TODO: point to MeatBall?. MeatBall? discusses social science issues on a theoretical level (but does not often use empirical scientific studies). It also has a lot of practitioner heuristics for moderators.
Deliberations need a space for meta-discussion
TODO: finish other mess map page. look up ISBN for Horn's book.
Hamlett's recommendation for an intro to (bad) small group discussion dynamics: Mandelbury's Chapter in "Review of Micropolitics", Tael: at Princeton.
from the social science literature, seems there is a ~130/140 limit to a ____ (true?) community where everyone knows each others
it takes about 20 years for a new academic discipline to start getting established