At LIFT ’07 in Geneva earlier this month, I spoke about ‘Collective Intelligence for the Enterprise’ (Presentation as 4Mb PDF – warning: contains gratuitous use of Chameleon imagery).
LIFT was another great international gathering of social software thinkers, linkers and practitioners. Well done to Laurent & team for giving so much of themselves and their time for 3 days of thought-provoking interaction. The conference spawned some great coverage, lovely images and plans for world domination, as you might expect. My session was written up by some fine folks, including Bruno Giussani, Stephanie Booth, Pedro Custódio, Thomas Purves, Dieter Rappold, Colin Henderson, Mikel Maron, Maíra Carvalho, Patrick Ferran, Mark Kuznicki and Lisette from Macaw. They probably captured the key points, but if you would like my own text of the talk, then it is as follows:
We are wasting a lot of brain power in large organisations…
…primarily through poor use of ICT, which is still based on the factory model of knowledge gathering and sharing. In particular, the sequential mode of email and document processing seems to me especially wasteful of brain power.
Malcolm Gladwell’s book Blink provides a good account of how people use pattern matching to process a multitude of ambient signals and imperfect information when making instant, intuitive decisions. In the field of KM, writers such as Dave Snowden have popularised the notion of sense making in situations of complexity. It seems, people can and do take in lots more information than they are aware of, and they are very good at making sense of information fragments and sifting out actionable intelligence.
But in the office, most people still sit in front of email clients and workflow programs, ‘pushing the button‘. They are expected to feed the system, rather than the system feed them. Most IT systems use 1990’s software based on 1950’s management thinking, and feed users with an insufficient diversity of inputs to stimulate any kind of intuitive decision making.
We should feed our minds, not the machine
There is a lot of hope vested in so-called Enterprise 2.0, in other words the application of lightweight social tools to stimulate collaboration and creative thinking within organisations. Tim O’Reilly talks about “applications that harness network effects to get better the more people use them.” I think the opposite is in fact the case with many current enterprise IT systems.
Key elements of enterprise 2.0 include:
- Social tools (blog, wiki, tagging, bookmarking, IM, etc.)
- An ecosystem of data (RSS, microformats, APIs)
- Connected infrastructure (HTTP, URIs)
- Subscription and aggregation (feeds, people, places)
- Participatory culture (co-production, discussion, linking, collaboration)
During the past few years we have seen the first pilot projects inside large knowledge-based firms, and some of these are now reaching the stage of mass adoption. It is hard, and takes time, to engage people with new ways of working, but it is happening in some places you might perhaps not imagine. This will continue to be a challenge for enterprise 2.0, especially since much of the talent in the sector is focused on inventing new tools rather than making them work for second-wave adopters inside large, established organisations. Once we have some flow inside these networks, then we will start to see a richer array of data, ideas, information and practice being generated than ever before, so it is worth thinking for a moment what this is for.
From Participation to Collective Intelligence
The launch of the MIT Center for Collective Intelligence last year was an indication of the next frontier, building on the idea of ‘data as the next Intel inside’ to foresee applications that use signals and inputs from users, at scale, to create social affordances from mass participation such as Amazon, Google, Wikipedia, and perhaps also Digg, Delicious, Last.FM and some other Web 2.0 sites. MIT quote Google, Wikipedia and Innocentive as three examples of this idea.
I have been intrigued by the idea since spending some time with George Por, a broad-spectrum thinker who is required reading on the topic. But the problem is that, like much KM thinking, much CI writing is too esoteric and hypothetical. For example, the ‘see also’ section of the Collective Intelligence page on Wikipedia contains the following terms:
- Preference elicitation
- Hundredth Monkey
- Simulated reality
- Collective Effervescence
- Keeping up with the Joneses
In a very basic sense, these are NSFW. If somebody came into my office hawking Collective Effervesence or 100 monkeys, I would laugh … and I like monkeys! Are these concepts really not able to be expressed in simpler, more practical terms?
The kind of down to earth objectives we think about inside organisations are things like:
- better understanding, know how, awareness
- more effective collaboration between teams; and,
- improved decision making by individuals.
Late last year, we did some in-depth analysis about how to save money and improve effectiveness in a knowledge-intensive environment, and we realised just how achievable some of the basic features of CI were becoming, through intelligent use of social tools. The goal is to bring some of the network effects and participation we have seen on the internet to the networks of companies and organisations.
On the net, we have seen sites like Digg and Wikipedia achieve levels of participation and ‘flow’ that companies rarely achieve, as evidenced by tools like Digg Swarm. But the next phase will surely involve the penetration of these tools, ways of working and ideas into more specific networks that share greater common purpose and culture. Digg, like Slashdot before it, reflects the native culture and norms of its al
pha geek origins. The dream of some kind of universal collective intelligence is just that: a dream. Most real-world social networks and communities are diverse and ‘local’ in some sense. Achieving consensus, or even peaceful co-existence within such large-scale, transient and largely anonymised networks, is perhaps unrealistic. Instead, one of the key characteristics of Web 2.0 is the model of networked individualism, where things are aggregated upwards from below rather than organised from above.
You don’t need the scale of Wikipedia (recently approaching 50,000 active users per month) to do something more focused and specific. Organisations with hundreds of people can leverage their scale to achieve positive network effects; with thousands, they can achieve much more. With tens of thousands of people in large corporates, the potential benefits are staggering, but user engagement is a challenge. Actionable CI exists within more tightly bounded communities and networks than the big public examples.
The enterprise opportunity for basic forms of collective intelligence is huge. We should be able to ask our aggregator:
- Show me the FT articles senior Directors think are important this morning
- Show me what our R&D people are reading about mobile technology
- Which internal blog posts do our clients find most interesting?
- Who is the most popular writer about corporate tax law in the firm?
Far too much time and effort is wasted by many people reading the same thing without sharing their ‘markup’, and it is hard to find out who is doing what, which causes duplication of effort. Laborious publishing to internal systems makes content hard to find again, and offers no pay-back to the user. Many companies still create high-cost conventional internal and external publications rather than simpler, cheaper, conversational versions of the same thing.
There are potential cost savings through doing existing work in a smarter way, but more importantly, social tools in the enterprise ca create a multiplier effect on productivity. Better peripheral vision – knowing what is going on around you – can reduce duplication of effort, and support deeper, more conversational client relationships. In a time when commoditisation is creeping up the food chain, providing high-end service like this is key to client retention.
CI need not be rocket surgery
The basic process of social reading, writing and filtering is quite simple. Individuals, groups and divisions act as funnels:
- 100 items suggested by your social newsreader today
- 10 items important enough to be linked and tagged
- 1 item gets blogged in full
Social reading and filtering drives relevance, so that others can share what you blog, link to or read. If lots of people do this, then within a given network, a useful flow of information, links and analysis is generated to support peoples’ work. Collaborative filtering of the resulting content based on social networks, tags, sources and attention data means that over time, information begins to find you, not vice versa.
Dion Hinchcliffe recently provided 5 useful suggestions for starting points:
- Be The Hub of A Hard To Recreate Data Source (eBay)
- Seek Collective Intelligence (Google)
- Trigger Large-Scale Network Effects (Katrinalist)
- Provide A Folksonomy (Del.icio.us)
- Create a Reverse Intelligence Filter (Memeorandum)
I would add some concrete steps towards implementing some of the basic plumbing to start creating CI within the enterprise:
- Expose feeds (RSS/ATOM) everywhere for everything
- Feed library management with attention data baked in
- Simple filtering tools: social newsreaders, a good recommendation engine, social bookmarking and blogs
- Clipstream tool to share collections and remixes
- Social search, driven by attention data & link authority
But finally, it is impossible to overstate the importance of engagement and context in making these kinds of application work inside companies. Software is not enough. To reach second wave adopters, we need to create ‘situated’ applications that are mapped to existing practice in order to make them relevant & contextual, and which look and feel like they belong. Engaging people with new ways of working is not easy, but it is not impossible – most people will buy into change if the value proposition is clear and immediate.