Here are some quick spontaneous notes & thoughts from my notes after an event “What Next for Collective Intelligence” hosted by UCL’s department of Science, Technology, Engineering, and Public Policy (STEaPP).

The thrust of the event was how to better use “collective intelligence” to inform decisions - so more sophisticated public consultations and deliberations, better use of scaled-up citizens’ juries, technology to draw out areas of consensus and disagreement, etc. etc.  Some examples of the tech ideas associated with this space can be found in the Civic Tech Field Guide. 

For those who more normally read this website for tech regulation thoughts - I see the problems of tech governance as reflecting wider problems of governance, including overly centralised information, insufficiently innovative analysis and decisionmaking tools, and increasing layers of “monitory democracy” (John Keane), or more vividly bureaucratic “sediment” or “sludge” (Cass Sunstein, Jennifer Pahlka, Ezra Klein), which complexify decisionmaking (see a critique of this by Sam Freedman). I am increasingly interested in new ways to use tech to govern tech. Particularly given AI developments, which seem good - if risky - ways to engage with the complexity of regulating.  

Hence following discussions like this (and also participating in schemes like Newspeak House, the Prosocial Design Network, etc).  My main interests here are (1) innovating governments/regulators and relatedly (2) how do we get this stuff to go beyond a few small pockets of innovative practice. My experience of government (like other orgs) is that it’s hard to shift critical numbers of people to change how things are done, even if existing methods (e.g. normal meetings, traditional methods of consultation) are clearly bad and disliked.  As host Geoff Mulgan said, a lot of the methods discussed in the event “are currently pretty marginal”; he thinks they may become common sense, I think they need a push.  

So from that mindset a few points which stood out to me:

• Joshua Becker: “Collective Intelligence” is already being implemented a lot (markets, elections, Parliaments, etc.).  It’s not a case of more, it’s a case of being deliberate to define and solve problems.
• Kathy Peach: People need to be bought in by trusting the messenger.  Existing institutions (like governments) may need better collective intelligence, have tools to do so, but be the wrong messenger so get little engagement. Martin Lewis’ Small Ideas Initiative is a good example of the opposite.
• Niccolo Pescetelli:  The barriers are not the tech, but human attention, institutional will, and money.  He thought new institutions might be a better way to overcome resistance than modifying old institutions (I feel old institutions would still resist; he felt that a “market” of competitive possibilities was worth considering.  That’s a bigger discussion I think!)
• Miriam Levin: Parliament and government are “collective intelligence”, but are failing citizens on some hard and polarising topics. Demos is doing alternatives which more directly engage citizens on some of these polarising topics; participants manage to agree on outputs, and report their changing minds during deliberation (e.g. deliberations on immigration, the Waves project). 

I then had a good conversation with Edward Saperia, Dean of Newspeak House (if you’re around London, have a look at the events calendar).  I started with the question: Why do the majority of people stick to traditional methods of doing stuff (meetings, sharing info) even when they're clearly inferior. Even people in this space - even this session was still basically a traditional series of short lectures and a Q&A, as are a fair few things I've seen at other spaces focused on deliberation and innovation. What hope to get innovative tools widely adopted?

We discussed that we need to distinguish between (i) people are encountering them and not adopting them because they're not convinced (which could be for many reasons - the tools aren't actually good, they have high barriers to entry, people stick with the familiar, etc.), or (ii) people aren't encountering them. I started as strongly (ii), Edward pulled me more towards (i) after the discussion.

I posed the question as "name a recent tool for Collective Intelligence which has achieved widespread use".  (My example was Pol.is, which I would say is one of the longer-standing and more famous tools in this field, but I wouldn’t say has become a super well-known tool outside it).  Edward pushed back with "Google Docs".  I said yes true, but examples like that are mostly because big tech companies can make us interact with their products. Then Edward noted Slack*, Doodle, Zoom**... I also thought of Miro, Trello, Slido***, etc.  So yes, there are plenty - maybe a lot in corporate settings, at least initially, but not entirely.

On reflection I think I was focused too much on thinking Collective Intelligence tools must be innovative and different. Some certainly lean on that, including as part of their own self-presentation. Whereas Google Docs feels like a normal document, or Zoom like a meeting, until you start to appreciate the myriad ways in which they can really change how we do things.  So maybe there’s room for innovations which don’t look like innovations, and the old app marketing trope “it’s like X, but for Y” (“it’s like Uber, but for Cats”). Kathy presented one tool - Zeitgeist - as like Zoom, but for deliberative discussions.  Edward has made a thing called docs.vote, which looks quite like Google Docs but with built-in rules around how proposals can be made, vetoed, etc.  Miro is basically post-it notes but digital, Trello is a to-do list… If we want to normalise innovations in Collective Intelligence, maybe they need to seem less innovative and more normal.

But I’m still left with the question of how, in the context of governing, such normalisation scales up beyond “making specific interactions and exercises more productive” (as useful as that is) when it’s all still embedded within traditional structures of governing and regulating.  When, if at all, will we see legislation actually being regularly made using innovative deliberation tools, rather than Arguments in Important Rooms? When will the “normal” public consultation no longer be either a set questionnaire or a free text box for any thoughts?  

There are also questions of who finds it good to have the massive proliferation of tools we see in places like the Civic Tech Field Guide, versus a few clear “winners”.  Some groups are good at picking the appropriate tool from a range of options for their needs; it seems harder to scale that approach up to e.g. a whole government department, without constantly having “wait, do we need to learn another thing…”.

But, in the meantime, it’s still good if a few more civil servants, MPs, etc. are having better meetings; more public groups are deliberating more effectively; and more intelligence being absorbed and acted on better. So next time you’re running a meeting or doing something else involving group decisionmaking, maybe just have a look if there’s a tool to make that better.  If you’re successful in convincing everyone else of that, even better.

* Having said this, I distinctly remember trying to instigate Slack usage in the DCMS Data Policy Directorate in 2019. I failed, as people didn't want a new tool to use; but it did maybe prep some of us to move more effectively to Slack-like Google chat functions when they began rolling out.** Zoom you might say "it was Covid! We need videoconferencing tech!". Well, true - but some of the design, like the side-chat for text comments, was not necessary but also substantially changed meeting practices. Edward was pointing out the untapped potential of Zoom apps and plugins.

*** I actually still much prefer to do Slido-style based digital Q&A - people write questions during the presentations, others can read and vote on those questions, host chooses questions to ask based on that - even in-person, if the audience is large. Written questions are usually more concise and precise than verbal ones, and hosts can choose questions rationally rather than randomly. Honestly, the randomness and frequent rambling of hands-up questions is one of my least favourite parts of in-person events, second only to hosts leaving no actual time for audience Q&A, which is where most of the value of such events comes from IMO. As such, I think a Slido based approach massively changes the usefulness of such events. There is, of course, a trade-off that some people are following a digital Q&A side chat rather than fully focusing on presentations; I think that’s a fine trade-off for a better Q&A.  Would I propose this approach for any sort of panel I was running? Yes.  Would I anticipate unwillingness and push-back? Also, yes.