Could a social medium be designed for leveraging collective intelligence, rather than entertainment and advertising?
We currently have social networking media, but we do not yet have social intelligence media. Existing social media is effective at building networks based upon familiarity or common interests to deliver simple diversions (for the users) or paid, targeted promotions (for the sponsors). Conversely, that other staple of digital culture, the Wikipedia, is neither social nor intelligent, and is apt to represent the collective trivia of the internet and the prejudice of nerds. I dream, perhaps idly, of something more than this – an online communication and knowledge aggregation tool that could not be ideologically dominated, and that might allow our intellectual resources to be effectively pooled.
How would we go about designing a social networking tool for leveraging intelligence? We would first have to avoid the obvious pitfalls. My purpose here is to suggest what these might be, and to propose possible solutions that could allow for the creation of something like a social intelligence network (SIN).
Pitfall 1: The Tyranny of the Lowest Common Denominator
Pictures of cute animals spread easily on Facebook but challenging ideas do not and cannot. Similarly, among the most commonly propagated materials on Google+ you will find a great many pictures from the Hubble space telescope. Because these systems are based on Like or +1 (i.e. “I agree!” or “Interesting...”) tallies and motivated reshares, social networks thrive on the concurrence of interests at the lowest common denominator. This is precisely what you want for entertainment and petty diversions – not to mention advertising. But it is not a way to leverage intelligence in problem solving, nor to aggregate knowledge.
Solution 1: Aggregate and Iterate
If problems are identifiable nodes linked to conceptual labels, solutions can aggregate around these nodes for discussion and refinement. Let's suppose that one of the databases at the root of a social intelligence network is a wiki-esque keyword tree. Each keyword has linked to it clusters of related thematic discussions, perhaps grouped as questions. Some machine curation is possible with such a system, but it might also rely on users being willing to connect their posts into the existing knowledge infrastructure.
However, they would be motivated to do so by the requirements of discoverability – if you don't link well, nobody could find what you wrote!
Related to this, the system should be set up to iterate on the material, so content can become refined through successive stages. The Wikipedia does a reasonable job with this, but it may be better for a SIN to actively encourage iteration by asking users if they would like to issue a revision – or put out a community call to revise and synthesize discussion on a given topic or problem.
Such an aggregation and iteration system needs more that a one dimensional response mechanic if it is to make the best value from user contributions. Users need to be able to tag content as (say) “Agree” or “Disagree”, as well as marking some content as “Interesting”, to show desire for further engagement. The robot curation system would need to process agreement and disagreement in respect of claims in order to tackle the next pitfall.
Pitfall 2: Singular Truth
The Wikipedia runs aground on Platonic metaphysics: there is one truth, we must discover what it is and enforce it. This, as I observed in The Mythology of Evolution, is precisely the mistake medieval Christianity made – and today it can just as often be found among those who have a non-religious commitment to Science. Similar fault lines occur around moral and political positions, as I outline in Chaos Ethics – such problems are widespread within contemporary culture precisely because it is not a unified culture at all, but rather a ‘multiverse’ of competing worldviews. We cannot eliminate this issue and we must therefore plan a way around it if we are going to avoid fostering the empty argumentation that occurs between those with opposing worldviews. How could we prevent time and information from being wasted by irresolvable disputants conversing with each other?
Solution 2: Situated Perspective
Rather than fight over singular claims, an alternative is to collect competing or related claims in their own network nodes, which would be crosslinked. Thus wherever there is substantial disagreement, the competing claims would form their own cluster under the relevant concept. Undisputed information can then filter through into a bridging node, with the alternative perspectives clustered around it. In effect, ‘rival’ clusters around a node represent shared worldviews (or at least, elements of shared worldviews), and thanks to the “agree/disagree” tagging can offer insights into these worldviews. This is more promising than self-identifying a ‘faction’ since only religious individuals have any skill at doing so.
Consider the age of the Earth as one example of a topic that is contentious. The main positions within scientific orthodoxy will likely fall out as a dispute within the same worldview (one likely to agree with 'measurement is a reliable route to truth' or some such claim or cluster of claims). Conversely, Young Earth Creationist claims will be united in agreement with ‘the Bible is true’, while more moderate positions can be subsumed into the orthodox clusters on this topic, while deviating on specific claims concerning evolutionary theory. It will be possible to see from such a system that disputes within orthodox scientific positions (e.g. over group selection) are more varied than contesting views from other sources. However, all knowledge claims would be collated in such a system – as indeed they should be. Even if you are dedicated to scientific process and are absolutely convinced all forms of Creationism are wrong, Creationist knowledge claims are still a part of human information as a whole: it would be perfectly plausible, for instance, for an anthropologist to study Creationism.
Suppression of disagreement is suppression of knowledge, and the ideological assumptions that are deployed to do so always assume the worth of a claim is to be judged against some conception of knowledge as a single coherent whole. Yet you could not, for instance, study Palestine without examining the framework behind the competing positions on this national claim. The same is correct of all disputed knowledge. Only by avoiding premature conclusions about what might be potentially of interest, irrespective of context, can you be sure to record everything of possible value to any hypothetical individual. “Disagree” yet “Interesting” is a whole category of discourse suppressed in conventional internet media.
Technical vs. Popular
The above discussion makes it sound as if the primary role of situated perspective is to separate out religious and non-religious background contexts. But actually, it would be just as useful for dividing up content via its technical content since not everyone is equipped to read every piece of written material. To put this another way, the idea that everyone speaking (say) English is using the same language is woefully misleading since each specialisation has its own special language.
Nothing could be easier for a robot than distinguishing popular and technical discussions, and this could be fantastically useful in SIN. It would allow users to be differentiated into subcultures just on the basis of the words used. As before, everything is cross-linked and accessible – but content tagged Technical can be ignored by those without the requisite lexicon, and content tagged Popular ignored by those more interested in the more complex discussions. This also suggests a role for those ambassadors capable of bridging between the two.
Pitfall 3: Forum Cock Fights
When nerds accumulate in a sealed ‘room’, they get into arguments that devolve into flame wars and noise. This was the blight of the old Usenet forums and it persists wherever the forum or mailing list model of internet discussion is dominant (e.g. Google+ Communities). The essential problem is that there is no way out in such a dispute: both parties claim membership in the shared space, and no third parties are as able (or, for that matter, willing) to intercede in such fights as they would be had it occurred 'in the flesh'. The result is unpleasant for everyone involved.
Solution 3: Reduce Gain on Flames
If the automatic curation system sidelines arguments between two parties so they are less prominent, this problem could be lessoned. It could even enforce a 'cooldown' period on willing users to prevent 'posting in anger' – and allow users to mute flames from users with no cooldown, as well as branching content by worldview by default (e.g. few want to read about a theology or atheology they do not hold!). Some experimentation would be required, but a comfortable balance is possible, and all the information would be accessible to anyone by voluntarily defining conditions for exploration.
I am confident this problem is manageable because during the height of the blogosphere, before the social networks drove it to the point of extinction, it was far more possible to avoid flame wars than in the Usenet era. The reason was that your violent disagreements occurred on someone else’s blog, and returning to writing at your own blog was far less likely to escalate conflict than pushing on with the same vehement dispute. Similar techniques could make SIN arguments more civil – or at least, less volatile! – and they could do this without automatically closing down discussions. Two disputants can keep at it - if they want - and the results of their argument could be resurfaced if they managed to reach an accord. In the meantime, they can fight in private since anyone not interested will not be shown the most heated exchanges.
Pitfall 4: TLDR
Some online discussions are too long to be widely read (blogs), and some are too short to have substance (Twitter). Is there a way to get the best of both worlds?
Solution 4: Variable Spaces
The benefits of a word limit on Twitter are that it shapes content for quick and easy consumption – to do it well requires careful construction, but either way it's quick. Setting limits like this control the depth and complexity of discussion – so why not explore this phenomena further by striating discussions into (say) Essays (2,000 words – about 10 minutes reading), Discussion (500 words – 3 minutes reading), Thoughts (100 words – a minute to read), and Blips (25 words – about the length of a tweet).
Now because of machine curation you could choose to launch discussion at any point in this scale, and equally choose to examine any problem or topic node at any scale. And of course, each scale is independently tagged – so you could go to look at “Rainforest preservation” (say) and find the best Discussion-length piece for someone with your approximate worldview based solely on the most valued Blips about it. Indeed, you could monitor Blip-feeds, or Essay-feeds, on any topic you liked, according to whatever filter criteria you liked.
As an additional suggestion to enhance this experience, replies to posts into the SIN can default to one step down. If you wrote an Essay (2,000 words), people's replies would be co-or donated as Discussions (500 words) by default. Nothing stops someone writing their own Essay in reply and linking to you (the equivalent of blog Trackbacks), but the system can make it easier to support shorter replies to help discourse remain fluid.
Pitfall 5: Money Buys Attention
We live in a world where large organisations can literally buy attention and, ironically, already possess a share of our collective attention by being well-known. This is a benefit (of a certain kind) in conventional social media, since wealthy organisations bankroll free usage through paid advertising. But this might not be so desirable in a social intelligence network – you would not want a tobacco conglomerate dominating discourse on smoking, for instance.
Solution 5: Individuals Only
The solution here must be to permit access solely to individuals. Large organisations will still get in by sending representatives, but could not plausibly buy attention in such a set up. However, this does not solve the question of how a SIN would be funded. The brutal truth is that every social network depends upon corporate involvement for its funding, and it might be difficult to pursue a project of this kind without such involvement. Even the Wikipedia doesn't get by without advertising – admittedly, the PBS-style ‘begging’ adverts rather than anything from the commercial marketplace. Perhaps the codebase could come about through open source means, although I doubt it. This problem is one I shall have to leave open.
The above assumes a social intelligence network that is a combination of distributed human intelligence and robotic automation – a cross between social networks and the Wikipedia. But to end this blue sky discussion, I want to look at an entirely different way of leveraging collective intelligence.
What I shall call a virtuous social network (VSN) is predicated on the idea of communities, linked only by common interests. These communities would be capped at a certain number of participants – something between 20 and 50 – so that anyone involved might plausibly get to know the others in their cluster. Each individual would be encouraged to belong to several clusters, thus allowing the collective intelligence present in each community to be leveraged on a grander scale, should it be appropriate. This is a much simpler idea than the SIN fantasy described above – but that makes it radically easier to implement, since at its core is only the idea of linking individuals into potentially productive clusters.
The pitfalls described above – and perhaps some of the solutions, too – would still apply to a VSN, but such an approach would have the added value of being more than just an abstract space within the depths of the internet. It would be a path to creating transnational communities on any topic imaginable. I wistfully like to think that this could be even more valuable than the techno-utopian vision that motivates my description above of a hypothetical network for social intelligence.
Do you think a Social Intelligence Network is plausible and desirable? Do you think Virtuous Networking is something we need? I would love to hear your thoughts in the comments!