Tag Archives: collective intelligence

The quest for collective intelligence: a research agenda

I am knee deep into the research work for opencare. I think I am learning new things on how to use collective intelligence in practice. This has far-reaching implications for my own work in Edgeryders, and beyond.  Far beyond, in fact. If we crack collective intelligence, we gain access to a new source of cognition. Forget my own work; this has profound implications for the future of our species. If you think that’s radical, go read the work of cultural evolution scholars, like Boyd, Richerson or Henrich. They think homo sapiens has started a major transition: evolutionary forces are pulling us towards a larger, more integrated “collective brain”. We are en route to becoming to primates what ants are to flies.

Collective intelligence is an elusive concept. It appeals to intuition, but it is hard to define and harder to measure and model. And yet, model it we must if we are to go forward. The good news is: I think I see a possible way. What follows is just a  back-of-the-envelope note, plotting a rough course for the next three years or so.

1. Data model: semantic social networks

I submit that the raw data of collective intelligence are in the form of semantic social networks. By this term I mean a way to represent human conversation. The representation is a social network, because it involves humans connected to each other by interactions. And it is semantic, because those interactions encode meaning.

2. Network science: it’s all in the links.

Collective intelligence is not additive: it’s interactional. We can only generate new insight when the information in my head comes into contact with the information in yours. So, what makes a collectivity more or less smart is the pattern of linking across its members. Network science is what allows a rigorous study of that linking, looking for the patterns of interaction which associate to the smartest behaviors.

3. Ethnography: harvesting smart outcomes

Suppose we accept that the hive mind can generate powerful insights and breakthroughs. How can we, individual human beings, lift them from the surrounding noise? Looking at what individual members of the community say and do would likely be fruitless. The problem is understanding how the group represents to itself the issue at hand; no individual you ask will be able to hold all the complexity in her head. We do have a discipline that specializes in this task: ethnography. Ethnographers are good at representing a collective point of view on something. Their skills are useful to understand just what the collective intelligence is saying.

4. “Shallow” text analytics: casting your net wider

Ethnography is like a surgical knife: super sharp and precise. But sometimes you what you need is a machete. As I write this, the opencare conversation consists of over 300,000 words, authored by 137 people. This is a very big study by ethnography standards, and these numbers are likely to double again. We are already pushing the envelope of what ethnographers can process.

So, the next step is giving them prosthetics. The natural tool is text analytics, a branch of data analysis centered on text-as-data. It comes in two flavors: shallow-and-robust and deep-and-ad-hoc. I like the shallow flavor best: it is intuitive and relatively easy to make into standard tools. When the time of your ethnographers is scarce and the raw data is abundant, you can use text analysis to find and discard contributions that are likely to be irrelevant or off topic.

5. Machine learning: weak AI for more cost-effective analysis

Beyond the simplest levels, text analytics uses a lot of machine learning techniques. It comes with the territory: human speech does not come easy to machines. At best, computers can evolve algorithms that mimic classification decisions made by skilled humans. A close cooperation between humans and machines just makes sense.

6. Agent-based modelling: understanding emergence by simulation

We do not yet have a strong intuition for how interacting individuals give rise to emergent collective intelligence. Agent-based models can help us build that intuition, as they have done in the past for other emergent phenomena. For example, Craig Reynolds’s Boids model explains flocking behaviour very well.

The above defines the “long game” research agenda for Edgeryders. And it’s already under way.

  • I am knee-deep in network science since 2009. We run real-time social network analysis on Edgeryders with Edgesense. We have developed an event format called Masters of Networks to spread the culture beyond the usual network nerds like myself. All good.
  • We collaborate with ethnographers since 2012. We have developed OpenEthnographer, our own tool to do in-database ethno coding I’d love to have a blanket agreement with an anthropology department: there is potential for groundbreaking methodological innovation in the discipline.
  • We are working with the University of Bordeaux to build a dashboard for semantic social network analysis.
  • I still need to learn a lot. I am studying agent-based modelling right now. Text analytics and machine learning are next, probably starting towards the end of 2016.

With that said, it’s early days. We are several breakthroughs short of a real mastery of collective intelligence. And without a lot of hard, thankless wrangling with the data, we will have no breakthrough at all. So… better get down to it. It is a super-interesting journey, and I am delighted and honoured to be along for the ride. I look forward to making whatever modest contribution I can.

Photo credit: jbdodane on flickr.com CC-BY-NC

Networks, swarms, policy: travels across the weird, dark landscape of 21st century policy making

I used to be an economist. Then in the two-thousands, I started to read about complexity science. I chased an intuition telling me networks are important (it was 2009, I still remember the epiphany when I saw the network analysis of interactions in Kublai) and started to study them. I was – still am – looking for a sort of Holy Grail: design and build online communities that can deploy collective intelligence to attack problems too complex for individuals (even very smart ones) or small groups to crack. To burrow deeper into the issue I had to re-learn some linear algebra and probability theory; and that unlocked paths entirely new to me, dark passageways across computational biology and experimental psychology.

The landscape got really strange, a far cry from the orderly, well-lit architecture of standard economics. Quite dangerous too: it’s full of philosophical traps (if it is really collective intelligence, will we individuals be able to recognize it? Would that not be like a neuron trying to understand the brain?) and even moral dilemmas (it is possible that the well-being of a system implies sacrificing its components, just like a species evolves killing off its weakest members: what happens if the system is society and we are its components? Do we sacrifice the whole or the parts?).

But here’s the craziest thing: I am not the only one wandering in this place, wherever it is. In the world of public policies, where I have worked for years, with every passing month I recognise new fellow travellers. I find myself talking of esoteric stuff like evolving networks, smart swarms, online ethnographies, variability engines. I feel like a sixteenth-century alchemist: we do stuff, it seems to work, we am not quite sure why but it works too well to be just random luck. We feel on the verge of an important discovery, something like the seventeenth-century scientific revolution. This weird, dark world is behind my talk at Personal Democracy Forum, held a month ago in Rome. If you want a taste, the video is below, both in the English original (left audio channel) and in Italian translation (right).

(Dedicated to Giulio Quaggiotto)

The Edgeryders conversation network in December 2012

Farming online conversations: assessing moderators impact with panel data econometrics (long)

Online communities have been my Swiss army knife for the best part of a decade. I tend to throw an online community at every problem I face – and that’s not limited to work: even my wedding party was organized that way.

We all know online communities have interesting properties, but how much can we control them? Can we “farm” them, growing one around each problem we are interested in? How expensive is this likely to be? These questions are relevant for my work, because we can’t use them as tools, as I tend to do, without some degree of control on them. I simply need to know if my intuition of online communities as general purpose collective intelligence tools is grounded, or if I am just delusional. As part of obsessing on this problem, two years ago I started a semi-structured research project, which is supposed to become a Ph.D. thesis. Some of the results are now in: the executive summary is that there seems to be evidence for our ability to artificially grow an online conversation about a specific problem.

Data source: the Council of Europe’s Edgeryders

I am using data from a project called Edgeryders. It was meant as a “distributed think tank”, an attempt to grow and harvest a large-scale online conversation around the task of building a proposal for reform European youth policy. The project was led by the Council of Europe and co-funded by the European Commission; it took place between late 2011 and the end of 2012. I was its director, as I worked at the Council of Europe at the time.

Edgeryders was based on an online interactive platform; its database if the source of all data described in this post. It worked like this: my team would ask it research questions grouped by topics (topics were broad issues like employment and income generations, education and learning etc.). Anyone was free to join and provide answers in the form of blog posts. All such blog posts were commentable for validation. By the end of the exercise, it had about 1200 registered users. Most were “lurkers” who never wrote anything on the platform. Those that did contribute were 260; collectively they wrote 500 posts and over 4000 comments. This material was analyzed by ethnographers, who used it to construct a publication on youth issues as seen from the youth themselves. The researchers obviously found Edgeryders material relevant for the task at hand. This, however, does not per se prove that we could really “farm” an online conversation around the issue of youth policy: maybe we simply intercepted a need to discuss the issue, and our efforts to grow and steer the conversation were in vain, or even counterproductive.

A small team of moderators devoted part of their time to encouraging users to share and discuss their point of view using positive reinforcement. This was a policy in the strict sense of the word: they were paid and instructed to engage users, especially first-time ones (“Hey, X, this is really interesting!”) ; ask them questions conducive to extracting issue-relevant information from the conversation; and connect them with other users with similar interests or approaches. Sometimes this would happen spontaneously, as the members of the Edgeryders community engaged one another; but when this did not happen by itself within hours of somebody writing a new post, the team of moderators was tasked with “breaking the ice.”

Modelling strategy

To assess the impact of the policy on the conversation, we proceed as follows. First, we model the Edgeryders conversation as a social network, whose nodes are active users and whose edges are comments. The network is directed: user i is connected to user j if i has commented some content (one or more posts or comments) written by j (comments are threaded in Edgeryders, so users can comment another comment). This was done by

  • using Datasource on the (Drupal) Edgeryders platform to extract relevant information from the database in JSON form;
  • writing a Python script to read the data and build the Edgeryders network, in conjunction with Tulip for graph analysis.

With dynamic network analysis still in its infancy, the main challenge to address this question was how to capture the time dimension of the data. I adopted the approach of “slicing” the course of the project into 57 one-week time periods, ranging from late October 2011 to December 2012. For each period I extracted the corresponding subgraph, using the following conventions:

  • “edges don’t die”. Let user i comment user j‘s content for the first time in period t. The edge from i to j appears in all graphs described from period t onwards. The interpretation of this convention is that, by interacting, i and j transform their relationships (among other changes, i is giving j a more or less explicit permission to interact with him or her) and this transformation is permanent in the context of the project (about one year).
  • nodes appear from the period in which they first create their Edgeryders account, even if they will only write their first contribution in subsequent periods. Nodes corresponding to not-yet active users will of course show up in the network as singletons, whereas nodes corresponding to users that will never become active are simply dropped from the network.

The final subgraph has 260 nodes and 4041 edges, many of them parallel to each other(same source and destination, different dates).

Once I had the slices, I needed a model of individual user behaviour that I could run against the data. The natural thing to do, with network data, would be to stay with a network approach, and estimate

 \forall i,j \in N P(e_i,j,t)|S_t

Where N is the number of nodes P(e_i,j,t) is the probability of user i forming an edge e_i,j at time t and S_t denotes the state of the system in terms of non-network variables at t. Unfortunately, this is not computationally viable. So, I collapsed network information into a vector of variables attached to each user, and allow it to vary over time. This transforms the problem into one of estimating

 A_i,t = f(A_{j \ne i, t}, EgoNetwork_{i,t} ,GlobalNetwork_{i,t} )

where:

  • A_i,t is the activity of user i at time t
  • A_{j \ne i, t} is the activity of other users at t
  • EgoNetwork_{i,t} is a vector of ego network variables (for example in-degree and clustering coefficient)
  • GlobalNetwork_{i,t} is a vector of global network variables (like density and modularity).

Once ego- and global network metrics have been computed for each time slice, this problem is tractable by panel data statistical techniques. Incidentally, notice that the relatively small size of each time slice – only one week – was chosen to capture the signal of the many lagged variables I had to use to avoid endogeneity issues. This, however, has the flip side of making most users inactive at most periods (13710 observations out of 14820 take value zero). This makes the dependent variable almost, but not quite, binary (taking value “nothing” in most cases and “something” in the rest). Therefore, the estimate was computed using a negative binomial model with fixed effects. This model estimates mostly the effect of regressors on activation (the probability of users to become active), but – unlike fully binary models like logits – it also uses the extra information encoded in users writing more than a post or comment in a given period.

An even more fundamental flip side is that this “flattening” of the network into a vector loses key information about the identity of who is connected with whom. In this first iteration, I kept track of whether inbound and outbound comments for each user come from moderators or non-moderators. Other refinements can be added.

Results

Preliminary data exploration shows evidence of structural change in early April 2012 (period 21). This corresponds to the date when a major Edgeryders conference, to be held in June the same year, was announced. People involved in the project confirm that this announcement was a game-changer in the project, as it made the Edgeryders proposition to its community much clearer and more credible – we referred to it as “addressing the what’s-in-it-for-you question”. We promised to cover travel expenses for about 50 contributors to the platform to get together in Strasbourg and flesh out the policy document that the Council of Europe would then advocate with respect to the European Commission and its own member states. We could spot structural differences across three subsets of the data: the first one describing activity in the interval before period 21; the second one describing activity of “old” users (who became active before period 21) after period 21; and the third one describing activity of users who became active in periods 21 and later.

The results of our negative binomial estimation for the whole dataset and for each of the three subsets are visible below.

To a first approximation, these results hold two lessons.

  1. Policy works. In all subsets as well as in the whole dataset receiving comments by moderators as well as by non-moderator community members has a positive and strongly significant impact (p-value < 0.01). This result is unambiguous.
  2. The network’s shape influences activity. All models show strongly significant influence of some variables capturing the shape of users’ ego networks as well as the global network. This is nontrivial, because some of these variables, like modularity, cannot be perceived directly, even by the most attentive users, without access to the database and network analysis software. However, these results are (still) ambiguous, and not consistent across subsets of data.

Implications

Result 1 has clear implications for online community managers running collective intelligence exercises. It says that user activity propagates across the conversation network, with each user receiving an impulse from in-neighbours and retransmitting it with its own input added (with some probability) to its out-neighbours. This is consistent with this paper by Nathan Hodas and Kristina Lerman, who find that contagion models explain well the spreading of information across online social networks after accounting for meme visibility as constructed by the designers of the online social networks themselves. The overall picture is reminescent of neural networks, a topic that I know little; it might be worth exploring this similarity further, as neural networks are in a sense a natural modelling choice for collectively intelligent organizational arrangements.

Contagion in a random directed network. Starting from the bright red node in the center (the moderator), the signal travels across the network.

Contagion in a random directed network. Starting from the bright red node in the center (the moderator), the signal travels across the network.

In a policy perspective, the result is saying that you can indeed “farm” an online conversation  by deploying one or more moderators/animators to interact with users. It works like this: a moderator engages directly some users; this increases the probability that they will become active; if they become active, this increases the probability that these first users out-neighbours will also become active and so on. This confirms anecdotal evidence: for example, as Shirky relates (source), photo sharing web 2.0 company Flickr deployed its own employees as the first users of the website, so that, when unpaid users came online, they found a lively conversation going on. Caterina Fake, Flickr’s CEO, reportedly remarked “You have to great the first ten thousand users personally”.

For online community managers, such a policy has two attractive properties.

  1. Firstly, there is a sort of multiplier effect: moderators activate users, whose activity in turn activates other users – remember, the coefficients on both comments received by moderators and comments received by non-moderators is strongly significant.
  2. Secondly, it does not matter which users in particular are targeted by moderators initially, because the activation signal propagates across the network according to some contagion model (video). To a first approximation, propagation will be limited only by the size of the strongly connected component that the signal starts in (though of course the size and topology of the giant component is itself endogenous to user activity; by becoming active, users can choose to activate new edges). Most real-life social networks tend to evolve towards a topology that features a giant component gathering a large share of all participants. In December 2012, the Edgeryders conversation network featured a giant strongly connected component that gathered more than 50% of active users. 100% of active users were part of a weakly connected giant component.

So, the model’s implications for the community managers wishing to “farm” an online conversation around a specific topic can be translated into two simple rules:

  1. Deploy moderators to engage users. This policy increases activity.
  2. Try to connect users with other users. This policy increases connectivity and, critically, the size of the network’s giant component. Moderators in Edgeryders were instructed to do just that, trying to match new users to existing ones on the basis of their first contributions (“Hey, Anna, this is really interesting! Something similar, but with a very different slant, has been proposed by Bob at this link…”).

The more activity and connectivity there are in the network, the less the conversation needs moderator effort to keep going. This is not so intuitive, and indeed online community managers tend to think it is the number of users, not the number of connections among them, that drive a conversation towards being self-sustaining. Again, if true this is excellent news for people who, like me, deploy online networks to attack problems by a collective intelligence approach, because it implies that you can get nearly self-sustaining (hence cheap to maintain) conversations even with relatively small networks, in the hundreds of low thousands of active users. This is because the potential connections in a graph grow quadratically with respect to the number of its nodes, so for a given level of performance, moderators can reduce costs both by attracting new users and by connecting existing users to each other.

What to monitor?

In the context of the CATALYST project, I am involved in an effort to build easy-to-install social network analysis software for common CMSs. Based on the above discussion, what should this software monitor? I would suggest:

  1. a visualization mode that makes it easy to tell the activity of moderators from that of non-moderators.
  2. the number of strongly connected components,  making sure that moderators engage each one.
  3. the size of the largest strongly connected component, both absolute and relative to the conversation network, as an indicator of conversation self-sustainability.

What do you think?

I wish to acknowledge the role of the Council of Europe in launching the project and firing me to lead it; of the University of Alicante, the INSITE project and IUAV for supporting my research; and to Giovanni Ponti, Luigi Di Prinzio, Guy Melançon, Benjamin Renoust and Raffaele Miniaci for their invaluable help and generosity. Financial support from the Spanish Ministerio de Economía y Competitividad (ECO2012-34928) is gratefully acknowledged.