Cointelligence R&D Lab Proposal
- I'm looking to run a small team doing groundbreaking research into small-scale-but-scaleable collective intelligence & collective wisdom, based on mindset shifts, synergetics, & we-space practices.
- This has been a field of inquiry for me for a decade, and something I've been investigating seriously for 6 years, primarily in the context of groups of people living together.
- This proposal is written in the context of the prospect of setting up such an R&D project as something more like a small institution—a purpose-driven team, not a family or co-living experiment.
- I'm essentially intending to work on this in some form in any case, as it seems to be clearly the most meaningful and important thing I could be doing with my life given my experience and appetite and the world's need for more collective wisdom & coordination capacity.
- The aims of the project
- To create and grow a group...
- of 10-20 people who are profoundly in sync and able to actually think as well as...
- it's hard to put it but...
- as well as a single human could if it had 10-20x as many neurons
- a five year old is to an adult as an adult is to this collective brain
- We're able to solve problems better than almost any individual could (except given specific expertise).
- There's a sense of the whole system being a perceiving-sensemaking-acting agent.
- We experience not just collective intelligence but collective consciousness.
- a group of people that has profound non-naive trust of their collective ability to relate to each other in ways that support the other individuals and the whole
- —trust which comes:
- not from a sense of collective identification with the particular group (as opposed to the rest of humanity)
- but out of the capacity to respect the coherence of each thought or behavior or control system that arises on any scale of all sensemaking systems (parts, people, groups)
- and of the capacity to recognize that the rest of the group is engaged in such respect
- (although some of the group flow would come from people knowing each other personally/uniquely)
- I’m talking about a collaborative superorganism in which the following elements are experienced not just some of the time or most of the time, but consistently:
- it’s safe, workable, and usually enjoyable to look at anything together
- people’s concerns are able to be resolved in good time, or they’re able to reach a shared understanding of why-not that feels acceptable to everyone in the meantime
- everyone is able to give & receive feedback effectively, without using blame-based explanations (ie essentially zero defensiveness, victimhood, or shame)
- nobody is attempting to control their own or each others’ behavior
- everything is seen as part of a learning experience
- nobody feels rushed or pushed through their own learning process
- broadly the group’s wisdom is the union of the members’ wisdom
- ie "there is nothing that I am wise about that WE aren't wise about". it's not quite the "sum of the parts" metaphor.
- (the effect of the wisdom compounds superlinearly)
- authority can flow naturally without guru dynamics
- we have a very high __c__ factor (a measurement of cointelligence independent of individual intelligence)
- …and many other facets of collective sanity
- (note that the above properties are closely entangled; it would be hard to have all but one of them, or hard to have one of them deeply without having a fair bit of most of the others)
- for many people, this will go without saying, but for others not, and in any case sometimes things need to be said even though they feel like they really really should go without saying, so
- when I'm talking about collective consciousness here, I'm not talking about some sort of merging in which people become all the same, or lose their individual identities
- increased personal differentiation and selfhood is a vital part of being able to come into states of high group confluence
- There are two closely-related puzzles here:
- how to create such a group at literally any scale, including 2 people
- (many elements even apply to 1 person)
- (therefore, to say the obvious, I'm talking about a leverage point that isn't about scaling typical sub-Dunbar coordination above Dunbar, but about creating a new attractor for sub-Dunbar coordination, that would also scale differently-and-better)
- how to grow such a group—by default, someone joining a group that has these properties will result in a group that does not have these properties
- this isn't just on the scale of like, being a team, but literally joining a conversation
- I have embodied experience & detailed models of how to do both.
- What day-to-day work would look like
- Each member of the team would have some concrete projects & responsibilities.
- Examples (this could be almost any job description)
- working on a solo software project
- working on a team software project (as coder, designer, whatever)
- interviewing users/customers or doing market research
- responding to customer support mail
- writing for the organization's website or social media
- leadership or project management of a team
- coherence coaching for other members of the larger organization
- learning some new skill
- The team itself would have some shared projects & responsibilities
- Running events within and external to the org
- Examples:
- "Living Room Weekend" style events
- trainings in specific we-space practices, or in listening skills
- learnathons, hackathons
- dynamic facilitation for teams or communities to make decisions
- Documenting the team's learnings internally to the org and externally where that makes sense
- Doing some coaching or meditation for other members of the larger org, since the members of this team would need to have those skills anyway.
- Each day (in some sense, each moment) we co-what-now
- We behold everything together, sit with the unknown of "what now?" and surface awareness of what needs our individual or collective attention.
- Some days this might involve an extended dialogue about something, whether that's a strategic question for the team or larger organization, a communication tension in the group, some feedback that needs to flow, someone's personal block, some concrete technical problem, or anything else.
- At other times, immediate individual or subgroup priorities will be apparent for all or part of the group, and those people will break off to go take care of those priorities. And then maybe some or all of the group will recenter later in the day to reorient.
- For more on what this is, read this article:
- In this context, functional co-what-nowing is a fluid bridge between strategy/tactics, and a simple stand-up meeting.
- Obviously it can break down
- either by
- tending towards too much strategy & meta
& losing touch with the object level- which is problematic, both
- because it means nothing gets done
- because it means becoming untethered from tangible feedback loops (as opposed to social ones)
- becoming just an ordinary stand-up meeting
& losing touch with the deep unknown- which is problematic, both
- because it means that double-loop learning doesn't really happen (changing strategy)
- because it means that triple-loop learning doesn't really happen (changing one's relationship to strategy)
- To some extent, this is how any sane organization would want all of its teams to operate. So what makes this a special weird R&D group?
- That's a good question!
- The simple answer is:
- Show me a sane organization—a fully meta-rational workplace
- That's the thing we're trying to build, and nobody really knows how to do it yet, and that is this project.
- Somewhat more verbosely:
- In most organizational contexts, there isn't enough slack, in terms of expectations, for adequate attention to be directed towards learning, not just doing.
- One of the only organizations I've seen try to do anything different on this front is Roam Research, where the theoretical norm is "if you spend an hour finding a new way to document your workflow and half an hour working, that's a win". I'm not sure how well it worked in practice.
- Many potential employees don't have immediate interest in having deep strategy or meta conversations.
- And there's a whole learning process for how to actually do so effectively—a learning process that hasn't been understood well enough yet to be able to be streamlined for onboarding new employees.
- Most people have some degree of motivational tangles that makes it hard for them to fluidly orient to priorities.
- Examples:
- Trying to do something may produce resistance that someone doesn't know how to voice or integrate. Pushing past that can work in the short-term, but repeatedly doing so undermines self-trust and thus undermines wisdom.
- The need to let go of a project can feel disorienting and frustrating.
- whether for internal reasons (it turns out to be 10× as hard as originally expected)
- or external reasons (the organization as a whole realizes something else is more important)
- Learning to organize one's motivation differently also involves a whole process that's still being understood.
- We also don't yet know how this functions at scale.
- Obviously in a 1000-person company, you're not going to get all 1000 people in the same room very often for a dialogue, if ever. You need smaller teams, then some way for those teams to interface fractally.
- For organizations of 20-30 people, you would sometimes want everyone in a big dialogue, but most of the time smaller groups would be better—probably up to about 8.
- We're still at the early stages of learning best practices for how to interface high-throughput face-to-face out-loud conversation with technological systems for scaling conversation.
- We can't yet solve this purely with technological solutions, because our concept of what it even is that we're trying to facilitate is insufficiently advanced.
- (As the group matures, however, the value of certain kinds of software (or perhaps even hardware!) will become apparent, which the group or another team on the org may build)
- For waaayyy more on how we get there, see How we get there
- (not publicly shared; send me some context if requesting access)
- What the organization would look like
- Cointelligence is fractal, so one can (in some sense) practice it on an individual scale, but in order to be making serious headway on this I'd need to be working with others.
- Some of the work with others could be with people who don't also have this as their focus
- Some of it would want to be with people who are consciously investigating the same questions
- Hence over the coming year I'd be aiming to hire 1-2 additional people who share a strong appetite & aptitude for this exploration, and another dozen or two over the following decade or so if all goes well.
- I have a few people in mind who I think might be good fits for the early phase of this.
- I'd expect that within about 2 years the group would be "paying for itself" via some combination of:
- small concrete projects that make money
- (eg building software tools, creating books/courses)
- paid consulting for other people/orgs
- (eg running trust-dancing workshops for startups and/or group houses)
- I'll talk about measuring progress below, but my strong sense is that time pressure is usually a huge hindrance to doing this work (note the "nobody feels rushed or pushed through their own learning process" as one of the properties listed above) which gives me a sense of wanting something like tenure for me and the other people who are primarily working in this group.
- This is complex, and of course would make hiring decisions higher-stakes.
- I'm not certain how to best design this and I'm open to input!
- It's obviously very costly to have people stuck in the situation who are a sufficiently bad match for it that it impedes bootstrapping.
- One element would be to have an extended trial/onboarding period, where we collaborate with them on a smaller project and practice co-what-now'ing and see if we have fairly consistent upward spiral dynamics.
- Ideally we'd be able to get on the same page about something being not a fit, and both sides would agree to part ways.
- But of course in practice, usually the most unworkable situations are ones where you can't really see eye-to-eye on such things!
- The tenure could perhaps include sabbaticals, so if things are downward spiralling within the group, an individual could take a multi-month break from working with the group (and either work elsewhere in the larger org or do some personal development work on their own).
- And during these sabbaticals there might still be some coordination with me or some of the other group members.
- Measuring progress
- Part of the whole point of the project is to cultivating a mode of being that is consistently able to orient towards what actually matters, rather than goodharting.
- Focusing on any quantitative metric is necessarily not the way to learn how to cultivate an ungoodhartable state of mind.
- Meanwhile, social things can goodhart exceptionally easily in a different manner than highly quantified goodharting: one ends up chasing the feeling of progress or insight, which may have little to do with actual progress or insight.
- So there's a puzzle here!
- As long as there isn't shortsighted pressure or cynicism, being accountable to leadership within the larger organization seems like it could be advantageous.
- It means the group can't go in circles for years while feeling like it's making amazing progress and is almost there, because someone outside will have a sense of "yeah but you said that before", and then we'll need to deal with that!
- There's a weird sense in which this is what being on an exponential or superexponential curve would feel like: you'd always look back and feel like you'd made no progress, and looking forward would always feel like you were about to take off.
- But this is only in a relative sense—insofar an agent is actually working on tangible problems, its capacity to tackle those problems should overall be increasing.
- There is however a question of the extent to which progress here would be expected to be more monotonic (always increasing) or whether it would involve more {2 steps forward, 5 steps back, 10 steps forward}
- My rough sense is that some things would be more monotonic and some would be more non-monotonic. Figuring out which things those are is part of the research project.
- Older thoughts here:
- Let's differentiate lead & lag measures:
- The lag measures would include:
- high trust & talkaboutability
- all of the items described above
- clearly effective teamwork on a wide range of projects
- which would mean ability to decisively & profitably execute quickly on new ideas
- effective teamwork with other people who aren't part of the R&D group, including other members of the larger org
- which would mean being able to augment the capacity of other team members by pairing with them or cothinking with them about decisions or designs
- Lead measures would be evolving, but might include:
- developing theoretical and pragmatic answers to relevant questions, like
- (we'd figure out which questions we're trying to answer)
- running a certain number of events and measuring their effectiveness
- ---
- Setting aside how it would be turned into a number, what does progress look like? How would even know, experientially, we were making progress?
- One of the things that's confusing about it is that it's partially a gradual skill that can build up over time, and it's partially a sudden shift that makes an immediate qualitative difference and that most people can get a taste of within a few days, given the right context.
- So measuring progress clearly can't be about the peaks, since those can be really high while the rest of the time things are messy.
- We could experiment with measures like Woolley's collective intelligence tests, although they seem like the sort of thing that would have some amount of practice effects.
- Why now?
- The simplest and truest answer is
- Why now? Because I want to. In particular, I've stared at everything and this seems like the thing to do in the context of everything.
- This is a classic question asked by investors to startup founders all the time.
- It has two subcomponents:
- Why not already?
- If this is possible, why didn't someone do it last year? Or a century ago?
- Why not too early?
- Why makes it tractable at this point, given that it wasn't before?
- I feel like I have a sense of this but it's hard to put it into words, and I've definitely often been optimistic about how close it is.
- Ah—what I know is:
- we can make progress on this right now
- progress on this is one of the most meaningful things for humanity right now
- I have less of a clear sense of "over the next N years, we can reach target X"
- Although broadly I have the sense that a 10-20-person group of people experiencing the levels of coflow described above is achievable within 10-15 years of sustained focus, and it'll start to pay dividends much sooner.
- It's hard to predict because the growth & development of such a superorganism seems likely to be superexponential, which means it never feels like you're making any progress at all, but a tipping point always looks imminent. I've experienced this from the inside and it's very weird.
- Another comment I can make regarding the why now? and why not before? question is:
- when I first started doing this work 10 years ago
- it was weird as fuck
- things like
- a culture operating without coercion
- these sorts of things were well outside any public overton window at that time, and now they're still not mainstream but there are dozens, hundreds, or thousands of people all jamming on these ideas on twitter
- and I know a lot more people who are very ripe for getting more seriously involved, and who are developing related ideas & practices on their own
- so there are more early adopters and more of a scene for this
- and a much larger and more developed pool of people to recruit from
- and a lot of pieces coming together memetically
- 50 years ago, when Synergetics was written (a book about this sort of project), the concept of a "meme" didn't even exist yet, nor did "meta" in the way we use it now. The word "fractal" had just been coined. These cultural technologies are vital for doing this kind of work and they're very new.
- A related question: why have I not made more progress than I have?
- As I said, I've been working on related stuff for 6-10 years at this point. What do I have to show for it?
- Well, what do I have to show for it?
- My embodied presence of it!
- And some amount of capacity to invite others into tasting it.
- Less meaningfully but more legibly...
- I have a ton of new models & writing on the subject
- I've been able to have conversations with a wide range of people that feel, particularly in the last few years, qualitatively different.
- ...
- And what have been the main things holding me back?
- Some of these are related to blindspots or confusions that were blocking things at Liminal Space Agency from 2012-2020, that my non-naive trust dance framework solves:
- trying to do it counteractively, for instance
- not slowing down & integrating oscillations
- / rushing peoples' learning processes
- etc etc etc
- Others might directly translate into the benefit of doing this as a workplace:
- lack of focus
- the groups that I've been doing this with have been groups of people living together, which in some sense has meant a lot of focus as we were basically aiming to inhabit collaborative culture 24/7, but that also led to people getting burnt out and pulling back
- also "inhabit collaborative culture 24/7" is a somewhat different frame than "learn how to have a meta-rational workplace", and I think the former produces a lot more resistance and is harder to tackle, even though they both kind of reach the same place in the limit
- lack of team
- also related to doing it with people I'm living with
- but I want to point at something more around not having a clear shared purpose
- and relatedly, the teams having various personal life conflicts due to kind of being too all-up-in-each-others'-business
- there are probably also blindspots or schemas of mine, or conceptual confusions, that are keeping things more stuck than they'd otherwise need to be
- necessarily, I don't know what these are!
- but I'll find them more readily working on a team with people who have different blindspots & confusions
- I do already have a rough sense of the perspective I'm coming from and other perspectives that are also valuable and necessary that I don't really understand or feel comfortable with yet, or know how to integrate with my main perspective
- but there are surely unknown unknowns as well
- Michael Smith's thoughts:
- 2 pieces
- embodiment takes time
- we needed the internet
- awareness of goodhart dynamics
- ~rapid enough iteration on rational memes
- the need for something else wasn't as present
- (rewind 1000 years and the church was adequate for the task of organizing culture)
- interconnectedness
- collision of belief structures
- speeds up goodhart etc
- (phase 3)
- choiceless
- right/wrong us/other
- MO > antirivalrous tech
- copyright law
- the need & lack for something else stifles creativity in the name of maintaining the incentive structures for people to be creative
- strikes me as the human version of Eliezer's case for working on AI safety
- Best way to convey the why now? is give people a taste for the experience of it.
- Potential Advisors
- (I haven't asked most of them, but they're all people I know who I think would be interested in some level of visibility into this project and would have useful stuff to say to us)
- Jordan Hall, meta-protocol & collaborative culture researcher
- Jean Robertson, wisdom anchor of Liminal Space Agency
- Conor White-Sullivan, entrepreneur & collective intelligence visionary
- Ali Tataryn, we-space facilitator & theorizer
- Mark Lippmann, meditation writer & teacher
- Catherine Olsson, AI alignment researcher
- David Chapman, meta-rationality writer
- Faith Curtin-Koch, coherence therapist and trainer at Coherence Psychology Institute
- Abhay Prasanna, management consultant
- (people I don't know directly)
- Related writing
- elaborates a bit more on the capacities I expect a 10-20 person collective consciousness to have, and how I see that interplaying with a larger "meta-team" and the global economy
- goes into extensive detail on my theory of change model for full-stack trust-building, to get from having a group of people who are kinda-sometimes doing the thing to a group who is consistently doing the thing
- very-unfinished draft that I started writing summer 2021 towards Conor White-Sullivan, that was intended to be a very methodical model articulation of why collaborative culture is an attractor that one can move towards
- meta thoughts
- it occurs to me that one interesting way to write this would be to write it as if someone had just read Malcolm's 100× vision & How we get there and understood them completely, and then all I'm trying to fill in is the specific form proposal
- ..although if they completely understood those docs
- ...another way to say maybe exactly the same thing though would be that instead of this being a proposal to someone else, it should just be me writing out the obvious (and the unknowns) of how I would imagine running such a thing
- and I don't think that's actually quite what I want to do here
- part of the whole point of this is to make my thinking more legible
4 linked references
June 27th, 2022
- progress towards Cointelligence R&D Lab Proposal via increased sense of what I'm working on and why it matters?
June 14th, 2022
- R > improv becomes good when people get a collective mind going