A Theory of Scalability

Posted by Rafe Furst in Agency, Alternative Institutions, Autocatalysis, Evolution, Interconnectedness, Interventions, Memes, Models, Philanthropy, Social Entrepreneurship, Socio-technical systems on October 7th, 2009 | View Comments

One of the hidden themes of The Feast this past week has been how to scale successful social ventures.  This has been on my mind a lot recently as I have been working informally with both Self Enhancement, Inc. (SEI) and Decision Education Foundation (DEF) on this puzzle.  SEI is extremely successful in the Portland locale where they began 25+ years ago, achieving 98% high school graduation rate (working against hard socioeconomic realities).  Like with many models that are very successful “in the small”, the biggest challenge is to translate that same success to larger scales (e.g. all across America, or all around the world).  DEF is attempting to build scalability into its model from the start, and has found that this is extremely challenging.

In thinking about this I am reminded about a duet of innovators who spoke at the Pop!Tech conference last year about scaling.  Both Bunker Roy and Paul Polack have some profound lessons to teach us about scalability.  You will learn these lessons by watching Roy talk about Barefoot College and by watching Polack talk about his Out of Poverty approach (also see BarefootCollege.org and PaulPolack.com).  But despite all of the incredible wisdom to be gleaned from observing how Roy and Polack achieve scale, I’ve been wondering about how their success can be translated to other realms.

Replicators

In creating a general theory of scalability, I think there is a key conceptual anchor from Susan Backmore’s TED talk on the third replicator.  Now, I have to pause here because as simple and great as the universal Darwinism principle is, I know from conversations that many people have a really hard viewing evolution in non-biological systems as anything more than a good metaphor.  It’s hard for most people to see that “true evolution” — the kind that Darwin was talking about — is actually what is happening in these non-biological systems.  I will address this in detail in a later post, but ask that you indulge me for the time being so that we can talk about replicators.

When we talk about scaling sociotechnical systems, really we’re talking about one of two things: either growing the original system to handle “more”, or replicating the original system (or enabling it to replicate itself) with appropriate variation for the new context.  Growth models are the more familiar and comforting to governments and policy makers for reasons that should be obvious to anyone who has noticed how scared these types get when faced with systems that scale via replicators.  Formal organizations (corporations, non-profits, governments, anything with a legal structure or formal set of rules) are growers; networks of cooperating agents (open source software, social change movements, revolutionaries, anything that is formed in a grass-roots / bottom-up manner) are replicators.

I am not here to argue that either type of system is dispensable, indeed they are both essential.  I will leave it as an unproven conjecture that we are at a point in history wherein the ecology of sociotechnical systems is dominated by growers that are straining and stretching to the edges of their dynamic range.  Societal edifices are crumbling under their own weight, and are thus vulnerable to subversion by an algal bloom of replicators in their midst.  For those that want the argument and evidence, go read The Chaos Point by the grandfather of complex systems theory, Ervin László.

And I will leave alone in this theory of scalability the entire grower side of the equation.  It’s been systematized and refined since at least the days of Machiavelli;  we know it today as management science.  Instead I want to suggest that there is lacking an entire half of the formalization project for a unified theory of scale, and that’s a formal model for scaling via replication.  The reason this formalism has eluded us for so long is the same reason Darwinian evolution is so hotly contested: it requires a fundamentally different way of thinking than the Western analytic tradition is based on.  That’s not to say that the complex systems paradigm is not scientific, just that the scientific method as it exists today has not yet incorporated the bottom-up, emergent calculus required to be complete.

The first question we must ask is what exactly is being replicated, and only then we can ask how that replication is achieved.  Blackmore names three classes of replicators which I would like to refine by pointing out (as she does) that these are really self-replicators.  In her TED talk she observes that biological self-replicators exist (i.e. what we normally refer to as “life”), that mental self-replicators do indeed exist (though most people don’t take this notion seriously enough yet), and that technological self-replicators are in the process of being born.  If we think about it though, it is easy to see that certain forms of this third replicator already exist: computer viruses, bot nets (e.g. as are used in DDoS attacks), digital agents in artificial life simulations and genetic algorithm systems, and others.  What Blackmore was hinting at with the her more restrictive definition of technological self-replicator is one in which the artifact being replicated has a physical form (as opposed to digital information form).

I must digress here for a moment to point out that it is a red herring to try to neatly circumscribe the system being replicated (the “artifact” or agent) from its environment.  In reality there is no such thing as a true self-replicator; there are always some resources or information that is outside the self-replicator that is required for replication to occur.  Neither the chicken nor the egg can recreate itself.  And if you (rightly) view the chicken/egg system as the thing self-replicating, you only need observe that food is also essential (as are many other things) for replication to occur.  Given this truth in the realm of biology, is it really so far fetched to view digital cameras self-replicating technological agents, that is replicators of the third kind?  Sure they require humans, manufacturing processes and other technology from their environment to replicate, but I’ll reiterate that there are no biological life forms either that are entirely self-replicating.  (This blog post puts an even finer point on it all, if you are still not convinced).

Principles

The scaling brilliance of Bunker Roy and Paul Polack was hard-won, after many years of solving specific problems at the bottom.  It was only after gaining a deep understanding all of the interacting subsystems was it possible for each of them to engineer an overall system that was scalable via replication.  Looking at various attempts to scale sociotechnical systems, both successful and unsuccessful, a pattern starts to emerge of the key principles and dynamics.  Here are a few:

  • Counterintuitive: Brilliant solutions are only obvious in retrospect.  Crazy. Crazy. Crazy. Obvious.
  • Self-Replicators: It is important to identify the parts of the system that are — or that can be made to be — self-replicating.
  • Fecundity: Digital information replicators are more easily replicable than mental constructs (i.e. memes), which are in turn more easily replicable than organizations of humans.
  • Mutation: The more fecund the replicator, the easier it is to co-opt for ulterior motives, and the more likely it is that random variation will throw the overall system off course.
  • Environment: It is easy to mistakenly believe that a prospective environment is suitable for replication when it’s not.
  • Side-effects: With any complex dynamic process there are always side-effects. If ignored, this usually leads to collateral damage, but on the flip side there is usually an opportunity to accomplish other goals and turn side-effects into new benefits.

In thinking about how to engineer a system to bring solar electric installations to rural villages around the world, it is counterintuitive to think that poor, illiterate grandmothers (with no formal education and very little social standing in their village) could learn to be solar engineers.  To further think that they could be taught by illiterate trainers (who don’t speak the same language) is crazy.  Until Bunker Roy proved it was possible.

Microcredit was crazy too, until Muhammad Yunus proved that it wasn’t, and then it was obvious.  So obvious in fact that it became a viral meme and has spread all over the world.  The concept of microcredit is a very fecund self-replicator.  Unfortunately, the practice of microcredit in many places has ignored the nuances of different environmental contexts and unintended side-effects.  Add to that a high mutation rate: the model being tweaked to confer greater benefit to lenders (at the expense of borrowers); the introduction of middlemen who screw up the incentive structure and unwritten social contracts; etc.  The net effect has been that in some areas microcredit has been a net negative to the economy, and especially negative to the borrowers, whom the model was originally designed to help most.

Polack’s franchise model (an indeed all franchise models) are inherently replicators.  They are also good self-replicators because customers and other locals get exposure to the idea of becoming an entrepreneur themselves. And some of them end up as franchisees.  That is replication.  But to move from solving one problem (e.g. clean drinking water) to solving a very different one (e.g. locally available energy), new technologies that are also “radically affordable” have to be created on a regular basis.  And this type of innovation does not self-replicate.  So Polack created an entirely separate institution, the non-profit R&D lab, specifically to tackle the problem of replicating franchises (i.e. going from an electrochlorinator franchise to a solar concentrator franchise).

Applications

With this nascent framework in mind, I’d like to invite you to evaluate some of the social ventures that I encountered at The Feast (and a few of my favorites from Pop!Tech last year) and see if you can predict how scalable their model will be based on the replicator principles above.  And in cases where they have achieved some amount of scale (like charity: water and frontlineSMS), can you explain their success using the theory?

I would love to hear your thoughts, both on the specifics of these ventures, and on the theory of scaling through replication.


Big shout out to the newly formed Brains of Change group whose speakeasy jam session helped crystallize many of these thoughts: Daniela Papi of PEPY - Taryn Miller-Stevens of StartingBloc -Daniel Epstein of Unreasonable Institute.  Be sure to follow their sailing trip around Madagascar as part of the #spintheglobe initiative!

Related posts:

  1. Alfred Hubler on Stabilizing CAS
  2. Autocatalytic Systems
  3. Daniel Nocera’s Gift
  4. Game Theory and Military Planning
  5. The Future of Evolutionary Theory?

  • Scaling up "should" require a separation of behavior types. High frequency flows ("tightly bound" behaviors) might benefit from ritual (entertainment) rather than policy (microcredit).

    Bind the ritual into the hand-off mechanism to the enlarging compartments - add a team to the league as the league grows - add a song to the choirs as the congregation grows.

    just some thoughts
    dwc
  • plektix
    I think an important aspect of replicability is modularity. It's much harder to replicate something whose functionality is "all or nothing"--ideas that have self-contained "sub-ideas" which are useful on their own are more likely to spread.

    Also, you seem to suggest that mutability is determinental to a replicator's success because of the possibility of co-option and degradation. But there is an important tradeoff between fidelity to the original good ideas and flexibility to adapt to new situations. Ideas that are too rigid will always be limited in scope.
  • Rafe Furst
    Jerri, I think you are onto something, but could you please explain what you mean by "platform"? I imagine it has to do with technology, but also other resources, processes, etc. Can you give examples on the extreme ends (i.e. the simplest possible platform and one of the more complex)?
  • Have a bit of a theory that platforms are the wave of the future and are the only real way to leverage the power of replicators. Platforms, frameworks, franchises. Especially in the social good world. I think the problem is that replicators (because they don't inherently have set structure) are "risky". It's not "risky" to get a job in a growth type organization (big NGO, or corp) but it is risky to be an entrepreneur. It also takes a lot of creativity to start your own.

    Platforms allow people to latch on and "call" something their own, activating their own network and potential in the process. We kind of noticed this with Cause for Drinks. It was a super simple idea, everyone got it, and because we had already created the materials, concept, process, people latched on and started contacting us across the country to host them. They wanted to do good but the barrier of starting their own was too great. They just needed someone to tell them what to do, allowing them to feel part of a movement but while also giving them enough creative leeway to feel ownership.

    Where many platform models seem to fall apart are balance of top down management and bottom up outliers, laziness, etc. as well as money (re: Clay Shirky's talk on open source being the act of collaboratively creating things out of love.. money and ownership has trouble fitting into an equation that's built on that principle). But it seems those with the hugest potential to succeed are those that facilitate, connect, create frameworks for personal action and responsibility while making it easy for people to become involved.
  • Thanks for the mention Rafe.
blog comments powered by Disqus