Advanced Search

Please click here to take a brief survey

How Would You Talk with 100 000 People?
Hassan Masum, 5 Dec 05

Habitat Jam finished up yesterday - a facilitated online forum by UN-HABITAT, IBM, the Government of Canada, and others that aimed to bring together thousands of participants for productive discussions around six urban themes. It succeeded only partially in its ambitious goals, but that's the nature of experiments. But from its problems, we may be able to tease out some lessons and principles for future large-scale distributed meetings.

This is a hard problem. Forums like Slashdot can handle millions of users, yet lack a sense of focus and a real-time component. The more immersive technology behind Habitat Jam (which has been used for awhile by IBM) shows promise, but is clearly missing important elements: effective searches of previous posts; summaries of the "discussion to date" to avoid repetition on basic issues; a user-rating system for posts as they are read to identify the more interesting ones (aka, "collaborative filtering").

Then there are the even tougher questions: How can a sense of "discussion" be kept in a large threaded forum? Would it be better to have multiple smaller groups discussing the same themes, and sending their best ideas to a common second-level forum? How can inclusiveness and diversity be balanced with selectivity and high performance?

How would you talk with 100 000 people?

Bookmark and Share


"How would you talk with 100 000 people?"



I don't know. If it's about brainstorming and making decisions, the best is probably to sub-divide the group into smaller groups (based on themes?) and then have them vote on what ideas and topics they think are best, and do that process of democratic elimination until you have a manageable number of ideas, and then to discuss them with the whole in a number of ways (panels of experts, conferences, round tables, online forums, etc) and at the end have votes on final decisions, etc.

Posted by: Michael G. Richard on 5 Dec 05

I would discourage the "subdivide into many groups" solution. While tempting, it makes it increasingly hard to have group conversations and to take a sense of the gestalt.

Instead here are some suggestions:

1. Remember that in many things on and off line the ratio of listeners/readers to speakers is very high, frequently 100 to 1 or often 1000+ to 1. Take advantage of this fact to scale up.

2. Do not neglect the costly and labor intensive but ctitical role of editors, transcribers and consolidators - i.e. people who are collating the conversations and distilling them down to form a top level record (with the original "raw" data still available thanks to the power of modern technology)

3. When possible us quiet, anonymous (or at least semi-anonymous) means of collecting feedback from the vast numbers of people who are listening/reading but not talking/writing. By anonymous I mean allowing them to respond without knowing how others individually are replying (i.e. secret voting). Combined with a request for suggestions - i.e. more than just votes - this can be a great way to solicit input from non-active participants on their own terms.

4. Provide clear, learnable, consistent, and still simple structures to the conversation. This is vital and WILL bias the conversations, but it is crucial to success. By this I mean top level organization to answer "where" to hold a given conversation. Rarely is this easy or without complication but sufficient explanation should be provided to guide and direct participants. This means that you do not just say broadly something like "issues" and "solutions" as topics, but instead would pick topics that are conducive to holding the entire thread of a conversation.

To use the housing jams example perhaps you might pick top level topics such as:

Emergancy and Ad-hoc housing solutions
Permanent, green and affordable housing

But even that might clearly have overlaps and links back and forth.

To handle that, it would be ideal if the editors can transparently link to/include threads to make a coherent "story" out of conversations that began in different "spaces".

This might, for example, suggest that the structure you start with is NOT the structure you end with - or at least not the only structure.

Hope this is helpful,


Posted by: Shannon Clark on 5 Dec 05

I agree - I heard that lots of my friends were on line at the same time as I was but we couldn't see each other. Emotionally it's interesting to know who is jamming and be able to branch off and come back to the group. I felt like there were a lot of tangents that were important but were all mixed in with really focused ideas.

I really like web apps that use aflax and ajax to illustrate things so that you get a sense of all of the elements but that also give you the ability to quickly scan and see what's hot. An example of this at the more complex end can be seen in sites like 10x10 ( but even flckr and use simple ways of doing this by increasing font size or colour to make info more visually accessible.

Posted by: Britt on 5 Dec 05

These are really smart suggestions. I've led the development of IBM's jams, and our thinking tracks very closely to what Shannon is saying here. This is a hybrid medium -- not a miscellaneous chat, and not a controlled, linear progression. It's aimed at heterogeneous populations, not at communities (of interest or practice), so scale and diverstiy of participants are very much the point. And we do, in fact, use lots of behind-the-scenes facilitators and readers who, along with the text-mining tools, spend time analyzing the transcripts.

We've also had a semi-Slashdot-like rating system in all the jams we've done inside IBM. Typically, selected posts have been (a)distilled to a short statement, and then (b) marked as 'rateable' on the site. The criterion for evaluation has been very pragmatic: Could you - John Q. IBMer -- implement this idea today, without money or executive aegis? Our aim was to enable participants to find good ideas selected by the group without having to trawl through miles of threads. As many people noted in Habitat Jam, the volume of posts in one of these events is simply overwhelming -- and if you approach the live event with expectations shaped by traditional online community sites and forums, it'll seem chaotic and redundant. (Another way to avoid the tyranny of chronology in a jam is through search -- which wasn't bad in this jam, if you tried it out -- and the use of the aforementioned text-mining tools. Ours comes out of IBM Research, and also includes a 'conversation browser' that enables someone to look at the whole thing from a number of different angles -- names, topics, parts of the company, geographies, etc.)

The problem with our rating system had been that it fell victim to what James Surowiecki calls "information cascades" -- namely, the earliest ideas put up for rating inevitably wound up getting the most votes. For the last jam we did inside the company, we addressed this by deferring the ratings until after the jam was over. The contents of the jam were winnowed down to 191 ideas by the facilitators and subject-matter-experts, and about three weeks after the jam, we held a week-long rating period, in which IBMers could choose the ideas they thought were best. The highest-rated ideas were then assigned implementation teams and executive sponsors, and many have already been put into practice, while others are still in the works.

The UN-Habitat and Canadian govt didn't do that with Habitat Jam, because the 'punchline' of this jam will be the physical conference that takes place in Vancouver in June 2006. The jam's purpose was to shape the agenda for that event.

Anyway, I could go on about this, but I don't want to blather too much. I just wanted to tell you that we appreciate your interest in this new medium, and would welcome continued exchange on how to evolve it and make it better. Thanks again.

Posted by: Mike Wing on 5 Dec 05

As a so-called moderator on the jam, I found a few simple things extremely frustrating.

1. When I posted a comment, I had to navigate through several screens to get to a page where my comment would be displayed alongside all others.

2. There didn't seem to be any feature that would notify me anytime someone commented about one of my posts or commented on a post that interested me.

3. If I wanted to track comments on a post that I hadn't gone back to for an hour, I had to pick my way through all the new posts to find the one I wanted.

All of this made it hard to have anything that resembled dialogue. Though I very much want to keep things like the jam open and inclusive, it's important to foster coffee klatches too.

I must admit, though, that I have primitive dial-up internet access, and that this might have caused some of my problems.

By the way, I like the idea of a flexible format in which 'the structure you start with is not the structure you end with.' A number of jammers suggested that we keep the discussion open through an online habitat jam magazine. Perhaps this kind of flexible architecture could be worked out in that context.

Posted by: robert neuwirth on 7 Dec 05



MESSAGE (optional):

Search Worldchanging

Worldchanging Newsletter Get good news for a change —
Click here to sign up!


Website Design by Eben Design | Logo Design by Egg Hosting | Hosted by Amazon AWS | Problems with the site? Send email to tech /at/
Architecture for Humanity - all rights reserved except where otherwise indicated.

Find_us_on_facebook_badge.gif twitter-logo.jpg