Will we comprehend supra-human emergence?
Thinking about complexity, emergence and ants, I went to a lecture by Deborah Gordon, and remain fascinated by the different time scales of learning at each layer of abstraction. For example, the hive will learn lessons (e.g., don’t attack the termites) over long periods of time – longer than the life span of the ants themselves. The hive itself is a locus of learning, not just individual ants.
Can an analogy be drawn to societal memes? Human communication sets the clock rate for the human hive (and the Interet expands the fanout and clock rate). Norms, beliefs, philosophy and various societal behaviors seem to change at a glacial pace, so that we don’t notice them day-to-day (slow clock rate). But when we look back, we think and act very differently as a society than we did in the 50’s.
As I look at the progression of:
Groups : Humans
Flocks : Birds
Hive : Ants
Brain : Neurons
I notice that as the number of nodes grows (as you go down the list), the “intelligence” and hierarchical complexity of the nodes drops, and the “emergent gap” between the node and the collective grows. There’s more value to the network with more nodes (grows ~ as n^2), so it makes sense that the gap is greater. At one end, humans have some understanding of emergent group phenomena and organizational value, and on the other end, a neuron has no model for brain activity.
One question I am wrestling with: does the minimally-sufficient critical mass of nodes needed to generate emergent behavior necessitate a certain incomprehensibility of the emergent properties by the nodal members? Does it follow that the more powerful the emergent properties, the more incomprehensible they must be to their members? So, I guess I am wondering about the "emergent gap" between layers of abstraction, and whether the incomprehensibility across layers is based on complexity (numbers of nodes and connections) AND/OR time scales of operation?
Can an analogy be drawn to societal memes? Human communication sets the clock rate for the human hive (and the Interet expands the fanout and clock rate). Norms, beliefs, philosophy and various societal behaviors seem to change at a glacial pace, so that we don’t notice them day-to-day (slow clock rate). But when we look back, we think and act very differently as a society than we did in the 50’s.
As I look at the progression of:
Groups : Humans
Flocks : Birds
Hive : Ants
Brain : Neurons
I notice that as the number of nodes grows (as you go down the list), the “intelligence” and hierarchical complexity of the nodes drops, and the “emergent gap” between the node and the collective grows. There’s more value to the network with more nodes (grows ~ as n^2), so it makes sense that the gap is greater. At one end, humans have some understanding of emergent group phenomena and organizational value, and on the other end, a neuron has no model for brain activity.
One question I am wrestling with: does the minimally-sufficient critical mass of nodes needed to generate emergent behavior necessitate a certain incomprehensibility of the emergent properties by the nodal members? Does it follow that the more powerful the emergent properties, the more incomprehensible they must be to their members? So, I guess I am wondering about the "emergent gap" between layers of abstraction, and whether the incomprehensibility across layers is based on complexity (numbers of nodes and connections) AND/OR time scales of operation?
posted by Steve Jurvetson at
3:55 PM
9 Comments:
Very interesting pointers. Thanks!
I was working off the simple topology baiss for Metcalfe's law, where the number of connections betweeen n nodes is (n - 1)! = n*(n-1)/2 = O(n^2)
So Reed looks at group forming networks like Orkut and finds O(2^n) potential sub-groups of interest. This contribution to network value dominates al others for large values of n.
So it begs the question, does group formation drive the properties of emergence in a system? For the brain, does this presume neural plasticity? Will groups of humans exhibit emergence of a fundamentally different nature - with groups and their memeplexes as the vectors?
By
Steve Jurvetson, at 5:15 PM
a nice paper to read about that is this one -->
http://web.media.mit.edu/~minsky/papers/sciam.inherit.html
Minsky is the advisor of Foresight, he is big
By
Anonymous, at 4:06 AM
Back to the big picture question: for the past 1000 years the pace of change has been glacial. The "clock rate" of the emergent intelligence would be the pace of inter-human communication. That was scattered, intermittent, and topologically clustered. We look back to the 50s and see plenty of change in norms, beliefs and culture, but it's harder to see the changes on the time scale of the nodes' activity (our mental activity).
With modern telecommunication networks, the fanout broadened and connectivity started to span geography. With the Internet, the fanout widened considerably (with multipoint communications like this post) and the clock rate increased with IM and email. With agent-based systems, the clock rate would increase further still.
Fanout is important. The adult brain has an average synaptic fanout of 1000. 2 year olds (aka "learning machines") have a fanout of 10,000. Neural network simulations with low fanouts achieve very little.
So I wonder if the glacial collective mind is transforming into a more powerful intelligence in the modern era, and if the pace of its operations is moving from decades to months?
By
Steve Jurvetson, at 8:26 PM
EY: Very interesting observations. I should clarify my question; I am not trying to focus on an out-of-system analysis of the high-level emergent systems (the social sciences do just fine there). I am interested in the in-system “emergent gap.” I would agree that the flocking patterns of birds and schools of fish are understandable from the lower-level simple rules. But what about the more complex systems?
I realize that there can be confusion whether we are approaching this as an in-system “node” or as an out-of-system human observer. As an observer, we can try to analyze a system of comparable complexity to the hive mind as you point out (I don’t think we can understand the emergent gaps yet, but I’ll get to that later).
But if we want to understand our individual role in the system (our contribution to culture, norms, etc.) and the architecture of mind, will we be baffled, and notice shimmering patterns of beauty at best?
And we might want to understand how to bridge this emergent gap more urgently than when analyzing other systems. We are the nodes. We care about the nodes. And we care about the societal constructs since they can affect all of us. For other systems, the nodes are not the focus of agency.Philosopher poet, Hunter S. Thomspon wrote: “Kill the body and the head will die.” The converse is also true.
I am puzzled about the claim that there are no selection pressures for the global mind. 1) It is possible that we have competing memeplexes or emergent minds at work today. Must there be one mind? 2) Perhaps the glacial timescale comments from before suggest that our global selection test has just not occurred yet. Will our society survive the selection test presented by modern terrorism and annual decreases in the “barriers to entry” for WMD? Will we develop a societal immune system? Do we start to recognize our interdependence and “commonality of genetic interest.”
And now, some quibbles about the relative complexity arguments of the brain. The ~6B population of human nodes should be compared to the ~60B neurons, not the ~100T synaptic connections. The synapses are analogous to the myriad communication networks between humans. (As an aside, a 2 year old has ~1 quadrillion synapses, but that does not enable the infant to deconstruct and understand the emergent gap of the relatively simpler “adult’ with 1/10th the synapses. I think you would agree that the brain, as a system of complexity N does not yet understand all of the phenomena and the emergent gaps in systems of complexity N (e.g., our own brains) or even .1N (simpler animals)).
What is the fanout of human social networks? We have broadcast mechanisms and multi-modal communications. We have the power of dynamic group-forming networks (discussed above with the Reed links).
And compared to the brain, “humans++” are powerful nodes. We have large local and near-line memory stores via our technologies and databases. (How much of the brain’s neurons are functioning as local memory vs. a computational capability?) We also have local agency - perhaps entering the domain of recursion and reentrant mimetic code.
What is the sensory I/O of a hive mind? The neurons of the brain are not the I/O; they do not interface with the world directly. As nodes in a hive mind, we may be contributing to activities entirely beyond our senses. As Ember, emWare and others hook billions of embedded sensors and machines to the network, the potential array of I/O interfaces starts to compound and extend the metaphor beyond human nodes to include other symbionts – much like the body is a federation of cells and cells are a historical endosymbiosis of simpler organisms.
By
Steve Jurvetson, at 2:09 PM
Thanks for you comments on selective pressure gradients and the speed limit of evolution. Fascinating stuff! If the global brain is analogous to competing memplexes rather than one Gaia-inspired sentience, then what is the reproduction rate? The competition is among memes. It is no longer coupled to our biological clocks. Selection pressure applies to the memes, as it does with genes, and the emergent intelligences are indirect effects. Perhaps I am misreading your comment, but it seems tightly coupled to physical biology. And I realize that the lack of rigor around meme definitions leave this a bit fuzzy. And I have been mixing scenarios, I now realize. The “maybe the selection pressure hasn’t happened yet” comment was a random thought relating to the Gaia/Selfish Biocosm scenario, and the rest of my comments relate more to the memeplex scenario. All of the parameters differ between these two scenarios. Sorry for that muddle!
Regarding your earlier comment about the ease of understanding societal emergence, I think the disconnects between our comments might trace from two root sources:
1) Emergence. From your other post, I think we have a different perspective on the emergent complexity and computational irreducibility across different layers of abstraction (by irreducible, I just mean that there are no major computational shortcuts to reduce the complexity of a system simulation). I don’t think we have figured out how to analytically bridge or “reduce” that gap for large systems of interest. (or perhaps we do agree. I’m not sure. I don’t think any of this will “save us,” but I do think it’s an interesting area of speculation, and the unresolved core of complexity theory)
2) In system vs. Out of system perspective. I realize that maybe my assumptions are influenced by my wife’s career. She is a psychiatrist. In her work experience, something may be perfectly clear to an outside observer that is opaque to an equally smart in-system participant (e.g., understanding oneself or relationship dynamics with a spouse). This relates to the notion of humans as the nodes and their ability to understand and take in-system actions.
By
Steve Jurvetson, at 6:49 PM
Certainly, I am brave to write some words here, where more than 50% of the words I read are complete not of my use. However, the form of treating this subject might be unreachable to me, but not its substance. And actually, posting with these obstacles and distances in our comprehension is a reaffirmation itself of the points I want to make with my comment. =)
I believe that the glacial pace of evolution within the last 1000 years is inversely proportional to the accerelating increment of formal education, amount of information and its storage available for more and more people. I mean: population has not become more intelligent because they have access to education and to all these technology changes. Quite the opposite. Seems that since we are not obliged to fight for survival knowledge, it is always there, even free, at a click... why should we make an effort of "thinking"? of building a personal new core of understanding about the world or about how to make it a better place, for example?
Information and its logistics and instrument has become a prosthesis. You know what real prosthesis make to people, right? A knee replacement might let you walk but you will never run again, and probably you will start to suffer other related problems caused by the use of the prosthesis (at the hip bones i.e.). With glasses you might be able to see but your next pair will need more magnification... endless examples. To sum up: everything tends to get truck in a comfort zone when it is not necessary to go forward. When survival is assured since you are born, fat changes are that you will never try to make any efforts, that may end in an optimization or enhancement of your nature.
Take this thread as an example. I think your questions on emergence are unanswerable from these big theories and formulas. Funny thing that those things that seem impossible to understand to an ordinary mortal may be a comfort zone to you. Here you need free thinking, creative activities, some body excercise too may help to open all the channels of understanding that are not the usual intellectual ones. This urges you to leave behind your preconceptions about the world. Because you are trying to visualize change in human race, the turning into something different than it is now, but you are using tools to think that belongs to this current now-and-here. I believe this requires more of playing, the "lego" thinking, rather than pre-thought theories. Also, you are trying to predict evolution, by reducing its nature to a system of complex equations. But since the variables: "emotions", "hope", "faith", "belief" are not included in the scope of causes that may alter the course of human evolution, I wonder how will you ever come to a accurate answer.
Everything about us might be reduced to our genome, even our conception of the soul and God. I am sure of this. All starts and ends at our body-mind system. But the keys to understand our behaviors, and thus predict them and their evolution, remain yet undiscovered.
My last comment: I think the in or out of the system view has nothing to do with the human ability of self awareness or posibilities to examine himself. What makes that self conciousness and recognition opaque most of times is a survival defense mechanism of the mind that put out of reach (in the dark) perceptions about ourselves that it (the mind) judges that threathen the whole system. It even put the system at mortal risk in order to keep on defending it from those perceptions. The most clear exmaple: someone that take drugs without understanding why (which means that the addiction is truly effective). The addict is keeping in the dark things from its past (...childhood let´s say) that his mind perceive as a big danger or unbareble suffering to live with conciously. Even more terrible than intoxicating the whole system with susbtances. Maybe because our experience and absorving capacity (learning) when we are children make a greater impact in our mind, so the memory of some traumatic thing at that age is recalled in the present of the adult still as a big drama, when it is no longer so. Seems crazy? Ilogical? It is not. It sounds very natural to me. I like the idea of our models and theories about what we are and how we work being wrong and ilogical rather than saying those addicts are mentally ill or insane.
But this requires an open mind. That´s why I recommend boundless thinking... to let your deepest side of yourself, your purest mind speak up... trying to feel (decode) the answers it gives. Like some kind of meditation. At least it can be a good experience. Sounds odd that I am suggesting a way of thinking to one of the greatest minds ever, uh? =)
I know I have not answered if we will comprehend suprahuman emergence. I am focusing in if it is this possible from this foundations.... My deepest respect, mon ami. I am truly blessed to have met you. Aur revoir.
By
Gisela Giardino, at 1:16 AM
Well, I finally got a picture of a Hive Mind in formation, so I think this settles the issue. ;-)
By
Steve Jurvetson, at 6:27 PM
In a recent retort to Metcalfe’s Law and Reed’s Law, variation in link strength leads Odlyzko and Tilly to conclude that the value of a network grows with n*log(n) instead of n^2 or 2^n.
“That means, for example, that the total value of two networks with 1,048,576 members each is only 5 percent more valuable together compared to separate. Metcalfe's Law predicts a 100 percent increase in value by merging the networks.
It's not a merely academic issue. "Historically there have been many cases of networks that resisted interconnection for a long time," the researchers say, pointing to incompatible telephone, e-mail and text messaging standards. Their network effect law, in contrast to Metcalfe's, shows that incumbent powers have a reason to shut out smaller new arrivals.
When two networks merge, "the smaller network gains considerably more than the larger one. This produces an incentive for larger networks to refuse to interconnect without payment, a very common phenomenon in the real economy," the researchers conclude.”
By
Steve Jurvetson, at 11:29 AM
Cool new research in SEED on this topic
By
Steve Jurvetson, at 7:29 PM
Post a Comment
<< Home