Models/Voting Schemes/Structure/Theory for collaboration
May 29, 2011
Hello, I am wondering if the Co-Lab group will give us some insight on how they plan to optimize the “collective intelligence” of the users? I know there are a lot of theory and formalisms on this in the literature, and I remember discussing several issues I had, especially voting schemes and various collaboration schemes that seemed naïve from the last iteration of the contest. For my group, at least, we would be interested in working on that aspect in the back-end; if you plan to allow the community to help out (rules/algorithms etc.) with the evolution of the site.
May 30, 2011
Sorry logged into the old account, the above person is me.
Jun 3, 2011
I am extremely interested in this as well.
Jun 3, 2011
This is a great question, and the answer is complicated. And I should point out I only speak for myself here, so Rob & Tom's opinions may differ. I'll do my best, and hopefully the others can weigh in. I think you've raised three points, really: 1) theoretical grounding (esp. voting); 2) our plans for optimizing CI; and, 3) The degree to which we will involve the community in the back-end design of the site. I'll address all of these points. 1) You are quite right that our voting strategy is naïve, and we are aware of the landscape. In the short term though, it's more about what is pragmatic and makes sense given the size of the community than the theoretically optimal approach. Other strategies - say, preference ranking, or Salganick's pairwise voting scheme (http://www.allourideas.org/ ) - are more time-consuming or require more participants. And *anything* that is novel is going to raise the cost of involvement for users. So, we've been very careful to try to keep it as simple as possible for now, and we're relying on the experts to hopefully serve as a foil to some of the problems traditional majority voting creates. Honestly, we're not even sure that contests / competitions are always the right sort of thing - maybe there's a different sort of process entirely (any ideas?). 2) Clearly, there are lots of things to be done to improve the socio-technical design of the system, and we've limited resources. But I think we face two immediate problems. First, how do we break up the work to be done? This is the "problem decomposition" problem, and the community is only just beginning to grapple with it (see a relevant publication here: http://reports-archive.adm.cs.cmu.edu/anon/anon/usr/ftp/hcii/CMU-HCII-11-100.pdf ). Secondly, how do we scaffold the "social" side of the side so that people can manage the work to be done? For instance, currently it is very hard for individuals to find others whom would be willing / able to help out with specific aspects of proposal creation. I think these two problems need to be solved before anything else. And possibly a close third priority is to figure out just what exactly to do with models. 3) I think we welcome community co-creation, but I think our feeling is that we need to have more traction before we hand the reigns over and really start to explore the space, you know? That is *not* to say we aren't open to input right now - to the contrary, we're really excited to get your input & we want more (if there were 100 of you asking to get involved, my answer would be a lot different!) But, at least while we're tearing our hair out trying to get people to join in the effort, anything that distracts from community creation / enablement is probably going to get deprioritized. (Though, if you've got hands that can code, we could talk a little more). So, to sum it up, there's a lot to do, but we're reluctant to do much that doesn't have to do with building a community. My opinion is that this has to be the priority or we're simply not going to make it. But the door is wide open here - I hope we can continue this conversation a bit.
Jun 7, 2011
Hi Cleo 2010 and Pljom, I'm curious about what ideas you had for optimizing the collective intelligence aspect of the CoLab. You mentioned the voting and collaboration schemes. What do you have in mind?
Jun 9, 2011
Thanks, all, this is a very interesting conversation. I'd like to hear Cleo's and Pljom's ideas as well.
Jun 12, 2011
For collaboration in Climate Co-Lab there are community aspects that intuitively come to mind: • Depending on the topic, actively interested individuals will lie somewhere on the spectrums below: o Those with strong specific vision and goals vs. those with general topic interest and willing to participate. o Those who are model builders vs. those who are more generalists o Other qualities • Conversations will occur within a triangle of: o Somewhat random free flowing dialog o collaboration towards a goal; and o a collection of strong direct or indirect (e.g., prior referenced work) individual contributions. • We are dealing with problems that may be of a wicked nature (I like the definition of wicked as a problem so complex that it can only be defined by its solution) that involve macro, micro and connectivity aspects which are best solved together. As for people it will help to know who we are and perhaps in our profiles we can check that we each show what we are able to offer and want to do? As for conversations, I think there is a point at which an online conversation can be too thread laden to be sustained -and then any value created may be lost (not to mention the barrier to entry by fresh blood). On this front I have been thinking in terms of one or several persons volunteering (by vote button?) to summarize a conversation once it has reached a certain critical thread size or time duration. One might also consider voice conversations as part of the process as online dialog can only do so much. The summary would include at the top one or more brief bullet points that are searchable (and in a more ideal world linked to a comprehensive ontology to increase usability) and if the topic was important enough the summarizer(s) could help feed the findings into a related conversation just starting. This would increase flow and cohesion of the whole. That this would work is unproven, but I have been considering these ideas for some time. At the collaboration level there is a goals aspect that must be solved, and then the question becomes how to lead toward the specific or fuzzy goal; spring-boarding off work already done. I would guess given Joshua’s investment and the people here the goal would be something complex but manageable, building in stages. The way I sometimes solve complex business problems (I am an actuary, alas) is by having a crude overall vision with subset blobs for the places that have not been figured out yet. There is usually something I know has to be gotten right for the whole thing to work and I first focus on that, getting others involved towards a shared vision and solution for that keystone, then more team-like we tackle the other blobs developing a collective vision for the whole. As example for a current project, I knew who needed to be involved based on the goals, and began by collaboratively expanding a precursor data capture and use system already delivering value. One key issue was getting more power out of the information flow and this was solved through taxonomy development and mapping -with the taxonomy considering all the relevant granularity and new uses we were aware of, but parking some issues we knew we were not going to be tackling soon. We then built the more technical parts around the taxonomy. This data and taxonomic focus proved a powerful tool and we ended repeating this process for other parts of the project. After each major ‘blob’ was mapped out this way, we found that where the independently created taxonomies overlapped new insights were to be gained, and connectivity naturally developed: and we ended up with a very powerful, comprehensive and intuitive representation of the nearly wicked problem we were tackling. Our collaboration would work to get some of the foundation blobs once we visualize them to a good state so that tangible value is created even if the whole thing is left unfinished, and then start building the rest in a part simultaneous, part phased fashion, but trying to create discrete value along the way. I suspect data and taxonomy will provide a very helpful bridge and focus, but what other problem solving examples can our group share? So how to begin? • Inventory (data, taxonomy and dialog?), goals exploration, tearing apart what we already collectively have, then create? A CoLab scrapheap challenge? • Passion led dialog, kind of like what currently exits? • Would an initial World Café or Open space dialog be helpful? • Other tacks? What are the answers to mission, key resources, key activities, key partners, customers, and etc.?
Jun 28, 2011
Thanks Chris! I wonder how much of what you described can be done already using what we have already onsite? The dialogue could be here, using these discussion pages? Taxonomy and mapping could be with debates - think of them as a landscape of possibility where we can position and reposition new and existing options? From these debates and dialogues (and alongside them) we can make collective visions and solutions as proposals for what to actually do? Here's a copy of a previous comment about what I' enjoying in the CoLab. https://www.climatecolab.org/web/guest/discussion#discussion%3DpageType%3ATHREAD%2CthreadId%3A7203 The CoLab's aim of harnessing collective intelligence has immense potential to stimulate change as it opens up the question of what to do about the climate in important ways: • Open to collaboration between climate experts and non-experts. • Open to a variety of approaches and understandings. • Open to collaborations between approaches. • Open to challenging assumptions so we can move from collective (herd) thinking to collective (re)thinking. Also here's the link to the Summer 2010 article about Thomas Malone and the Center for Collective Intelligence. "How can people and computers be connected so that collectively they act more intelligently than any person, group, or computer has ever done before?" http://spectrum.mit.edu/articles/normal/collective-brainpower/
Apr 20, 2015
Hey, I just read comment 7 and thought that is quite correct, and I am afraid colab has it "backwards". They try suck suck people into a "broken system", which is up hill for them. Then when they have people that can work in this environment they want to change it. And all the people have to re-adjust. Sounds like a nucleation/bootstrapping issue to me. I'd want a system where I can piggy-back on an old-timer who seems to have a similar alignment, and perhaps leave comments along the way for others doing the same. Also I'd want to be able to drag existing groups into some collaboration system with me, and not be member of n seperate ones (COlab, not SEPlab). That would need some sort of shared interface or migration help I assume, so you'll have to keep this architecture anyway - but you need to provide a better soil if you want to grow big ideas I am afraid. Unless you target a "text-box species" - but then you get skinny "text box results".
Apr 29, 2015
In some cases where there are proposals that fit together with each other like jigsaw pieces, the judges could set combining these proposals as an requirement for proceeding to the next round. This could make one strong propsal out of some proposal missing important aspects that the other proposals would bring. In other cases the judges could just hint of the benefits of combining the proposals.
Apr 30, 2015
Categorization of contests should be grouped further; land use, climate change awareness, youth action could be grouped. In fact climate change awareness and youth action programs can be grouped with many other categories. Such combined/partly integrated groups would add more meaning to overall theme groups. Combining Proposals is one way to add more meaning. Could another be grouping of contests as one proposal may fit into more than one contest theme? Both approaches are feasible and could even be combined!
May 5, 2015
Seems like this could be an interesting topic for a contest.
ADD YOUR COMMENT
You must be logged into your account to post a comment.