I'm not so sure we really have examples of collective intelligence, type II.
In the free market, you have individuals making decisions in the context of the information that the market provides - which is the aggregrate of everyone else's decisions. But the individual element is there, it is just not published, cited, etc.
In the scientific world, you have an individual putting together the pieces of the scientific puzzle in the context of the information provided by all the other scientists. Simply because the individual element is in this case published, recorded, talked about does not mean that it is different from the free market.
We could have had a "anonymous" culture in the sciences and arts too.
I think Rutherford said that he could not achieved what he did without the expert glass-blowers that were available to his lab. CERN would not work without the electronics, which is at its advanced state because of the efforts of hundreds of thousands of people, many anonymous. Likewise, the people at Canon designing a DSLR or Steve Jobs and Apple designing a phone - they are doing this with the input from the market. I don't see much difference between the free market and the scientific world here.
Regarding your example of estimating the beans in a bowl, true, it encodes information that no one person has. **BUT** it is in part because there is an objective answer to how many beans are there in the bowl. The similar "wisdom of crowds" also results in a "GOD", of whom no one person is sure of all the properties. Prices in the free market are more akin to "GOD".
Further, the prize for guessing the beans in the bowl goes to whomever guesses the closest, and not to the collective; the same is true of science.
Collectives regress toward the mean as outliers are shunned for being devisive. Harvard computing attacked Bill Gates creating terse BASIC. Collectives' outputs are beige rainbows.
Fundamental observables hate emergence. Symmetry-breaking ruins elegant theory. The Standard model chokes on mass; GR chokes on spin. Matter predominates over antimatter while physical theory plays Lysenko.
Tolerances average in knowledge but add in illusions of knowlege. Religion reaches no conclusion. Einstein was not hoisted upon jubilant shoulders. He waited until the old guard died. Eddington crushed Chandrasekhar. Yang and Lee were pariahs until experiment.
"the academic system" American public education had McGuffey's 1836 First Eclectic Primer, and it worked (even for the Irish). President Johnson collectivized American education: 11 April 1965, "Elementary and Secondary Education Act." It was improved means to deteriorated ends, extruding intellectual cripples at astounding costs. A BS in STEM now costs $(US)200,000 at a public university. Outsource!
If you only look at the local surrounding you'll never figure out additional information in the topology of the whole thing. You'll have to do something more for that, like looking for closed loops.
http://www.mazepath.com/uncleal/erotor1.jpg The worst it can do is succeed.
Progress requires a sharp nail and a sound pounding - the antithesis of collectivism. Development is collective, squeezing out intellect and craftmanship by statistical process improvement. Journals publish product. Never confuse product with process. Process like birth is an unwelcome painful mess after the main player gets... screwed.
Michael Nielsen writes a lot about this sort of topic (indeed, his last two posts in months were about a book titled "Collective Intelligence"), and I recall at least two really interesting examples, one of which I think might satisfy your criteria for type II; specifically, this post describing an annual programming competition held by MathWorks: "The competitors don’t just submit programs at the end of the week, they can (and do) submit programs all through the week. The reason they do this is because when they submit their program it’s immediately and automatically scored. [...] What makes this a collaboration is that programs submitted to the competition are open. Once you submit your program anyone else can come along and simply download the code you’ve just submitted, tweak a single line, and resubmit it as their own. The result is a spectacular free-for-all. Contestants are constantly “stealing” one another’s code, making small tweaks that let them leapfrog to the top of the leaderboard. [...] The result is that the winning entry is often fantastically good. After the first contest, in 1999, the contest co-ordinator, Ned Gulley, said: 'no single person on the planet could have written such an optimized algorithm. Yet it appeared at the end of the contest, sculpted out of thin air by people from around the world, most of whom had never met before.'"
A similar example he described was an internet-based chess game that appears to have made the group much stronger than any individual participant, as they gave Kasparov a run for his money.
Two more related examples: Nielsen discussing Intel's design process, "[An Intel engineer] stated that there was no one person who came even close to understanding the chip in its entirety. Instead, Intel has fashioned a very clever social process where that is not necessary, and the engineers working on the chip only understand it collectively."
And Matt Ridley's TED talk, "When Ideas have Sex," which is just full of great ideas which (I think) relate to this.
But if you don't think any of these qualify as type II, maybe I should re-read your post.
It seems that type I is potentially instantaneous in the sense that it merely realizes an output as a result of the connection of information (be it developed inside each one and/or as the product of collaborative work per se), whereas type II would require evolutionary (process-related) conditions, whose integration is not "instantaneously" present but requires some "historical" and/or continuously changing factors (be them local or global in nature) in order to provide an output (at least, according to cody's previous examples).
I believe in Collective UN-Intelligence. Unlike the CIA, no, come to think of it, that's an excellent example.
Think Mob Violence (for example "the mob" attacking Frankenstein's castle. If they had brains they would have cheered Dr. Frankenstein for discovering immortality).
"One boy is a boy, two boys are half a boy, and three or more boys are no boys at all" is heard on many school US Elementary school playgrounds by the teachers.
What is the problem with The Singularity? How fast is AI progressing? Computers are idiot savants, like Dustin Hoffman's character in "Rain Man." IQ=17.
Humanity's IQ is 100. Less so in countries that worship revenge.
The point I was trying to make with the economy is that you have an additional structure that allows the whole system to be "more intelligent" than just putting together many people. The additional structure in case of the economy is introducing money and prices for goods. Letting this system operate gives you a better result than, say, asking some king how to best price goods. In that sense the "whole" is more than the sum of the parts. But you have to do something for it. The knowledge that you get in the end, the "true value" of a good, is not something that was decided upon by any of the people participating. It is a result of this particular aggregation mechanism.
As I wrote, I agree that in academia when it comes to solving scientific puzzles there's no collective intelligence of type II (except for a few trivial examples). The example that I mentioned as an example for CI of type II in academia was not solving scientific puzzles, but deciding which theories are the most useful to keep. As I mentioned, this "system" is not well understood and it doesn't built on written down laws and we have no good model for it, but I think it does exist anyway, in some vague sense because scientists do use a shared ethics indeed (or most of them do), so there's your additional structure.
You need an objective answer if you want to quantify really how intelligent a collective is or isn't. That a decision is "beneficial" is not easily quantifiable. The number you get for the marbles is quantifiable in its precision. The marbles-in-a-jar example is a very trivial one, but there too you have made use of an additional structure on your group that allows it to produce new output.
Well, yeah, Smith's reference to the "invisible hand" definitely smells like God's guidance, but it's a metaphor for there being a mechanism that the individual doesn't determine. These are features that are very hard to understand because humans have little if no intuition for what effects can emerge in a crowd. Here is a recent very sad example of that. Best,
Yes, collectives can also be more stupid than individuals. As I wrote, it depends on how you let the individuals interact and how you arrive at a new knowledge or a decision or an action and also what the problem was to begin with. Of course if you do it wrong, the result will be bullshit. But the point is, if you do it right, everyone in the group can gain from it. The MIT has a Center for Collective Intelligence whose studies are dedicated to exactly these questions: under which circumstances can a group be more intelligent than a person. And when does that fail? I think these are extremely important questions. Not so much for companies who want to do crowdsourcing but because it's questions that we need to answer to understand the functionality of our economic and political (and academic) systems.
What's worrisome though is if you look around, the properties that you see in how people connect and how information is distributed are exactly those that are known to promote dumb rather than intelligent decision: oversharing (of information and opinions) reducing heterogeneity (leading not necessary to streamlining, but more commonly to polarization), and information being distributed by authorities which vastly amplifies the impact of mistakes and inaccuracies. Think think, where are we headed?
I can really recommend Surowiecki's book. It is, in contrast to what many people seem to believe, *not* a praise of the wisdom of the crowds, but a summary of the circumstances under which crowds are wise or aren't. It is also well written. Best,
A very interesting piece referencing to a book I will now have to read as I find time. You make a good distinction between what you find as the two types of collective intelligence and yet for me they stand as Type 1 being cooperating intellects and type II as a fully shared or unified intellect.
However the key difference between the two for me relates to what is called consciousness or self. In Type 1 there always remains relevant many selves or consciousnesses, while in the other at first glance it appears there could be room for only one. However then I am reminded of Penrose’s arguments where when it comes to living things all are conscious right on down to the simplest one celled form,, with self awareness not being something attained at a certain level of complexity yet rather only increases by degree; that is as the light a lense can gather increase with the size of its aperture.
So In this respect (as you pointed out) one might consider the human mind itself already an example of a Type II intelligence, which would also require a advanced form of consciousness to have it work and this is the aspect we tend to overlook. So in order for a collective of human minds to form a super intelligence it would seem to me it would mandate the creation of a super consciousness. Perhaps then as a result being that the new knowledge it comes to discover can only be made fully aware to this new consciousness and then never fully understood by any individual mind. That is just as one brain cell is not aware of although does benefit by what is known by the collective we call a mind. For me the consequence of managing such a feat would not just create a superior intelligence yet in the process a new and distinct being. I would then wonder once created if it would ever want or allow for it to be separated again or would we be content to surrender ourselves forever to have it created?
"So in order for a collective of human minds to form a super intelligence it would seem to me it would mandate the creation of a super consciousness."
That would be the final consequence I think, yes. But I'm not sure we'll ever get there, and certainly we're not even close to. We're more on a level where we occasionally find together to collectives for one or the other purpose and try to find intelligent ways to make decisions that serve best our needs. It would take significant changes to human nature to ever create something like a "super consciousness," I'm not sure these changes can or will ever take place, and the internet certainly isn't even remotely sufficient for that. It might be sufficient however to do exploit some advantages of collective intelligence. Best,
BTW I don't believe in collective intelligence neither - for example dense aether model is incredible simple, but no one from myriads of aether proponents or so-called "natural thinkers" has considered it as anyway.
Technologies such as the internet at best facilitate cooperative collectives, as although it allows the interaction of multiple minds it has no power to form a super consciousness to unify them in awareness and purpose. However I do think we will in time discover as to invent technological means to create such an interface, with perhaps quantum computing being the key as it holds the promise of exploiting the non local aspect of nature.
I’m reminded of a blog piece I wrote awhile ago that showed that although leaf cutter ants individually are not conscious of such a capability and purpose, that as a collective they do support the group through what we would find as the selection, care and nurture of another species they all then consume ,which we would might find as having developed the technology of farming.
In similar fashion I can envision that interfaced human minds could combine to form a consciousness that although not known to the individual would form to exist in terms of the whole that could act in unified singular purpose and capable of things well beyond that of the individuals. Then for that matter, in some respect, we might cite society as a wholeness of purpose and utility exceeding that of any individual and yet benefitting all. It has one to wonder if there is even now a difference between recognizing a distinction between the consciousness of a human and that of humanity?
“It is absurd to suppose that purpose is not present because we do not observe the agent deliberating. Art does not deliberate. If the ship-building art were in the wood, it would produce the same results by nature. If, therefore, purpose is present in art, it is present also in nature. The best illustration is a doctor doctoring himself: nature is like that. It is plain then that nature is a cause, a cause that operates for a purpose.”
in my opinion, some extreme sort of your type I intelligence is the work by R. Brout and F. Englert: arXiv:hep-th/9802142, who were complementary working together.
"In similar fashion I can envision that interfaced human minds could combine to form a consciousness that although not known to the individual would form to exist in terms of the whole that could act in unified singular purpose and capable of things well beyond that of the individuals. "
Yes, I can imagine that as well. But that I can imagine it doesn't mean it's likely to happen. I'm not at all sure we'll ever be able to create such an interface and, if we were, would be able to use it in a beneficial sort of way.
"It has one to wonder if there is even now a difference between recognizing a distinction between the consciousness of a human and that of humanity?"
I don't think that presently there's anything that could plausibly be called a "consciousness of humanity." What we have is a collection of consciousnesses for which the whole is pretty much the sum of its parts and not more. With some exceptions (see post) that however do cover only very few areas of human actions, mostly locally so, and for what I am concerned don't even work particularly well even in these cases. Best,
Although not about collective intelligence, Lanier has inspired a certain appreciation that I develop into the following point: A purely computational intelligence (as some like Dan Dennett suppose even we are) cannot formulate the thought of special real existence apart from logical structures (ie, such a mind cannot even represent disbelief or an alternative to modal realism/MUH.) That's becasue computations are just math, they can't represent "this is just math" versus "this is my thinking here in a real material world." Well, I don't think CI/AI is true - we are not gadgets! - but that would be the implication. (Something breathes "afierie" into the equations, my Captcha which is often uncannily relevant.)
I would say that as still the forming of consciousness is much of a mystery that being sceptical is resultant more of what is not known as opposed to what is. What is known is that as life continues to evolve in complexity that with this increase comes a corresponding increased ability for it to communicate, Therefore I would contend by virtue of what is known there is no reason to understand things have reached any kind of limit in respect either life continuing to evolve in complexity or communication. As for the advantage to such communication I would agree that much like the leaf cutter ant’s inability to fathom the implications of the collective’s farming a singular human consciousness may be similarly fated. None the less a failure to understand has never been evidence to have something not to exist or able to. This is also essentially the contention of Aristotle’s in what I had previously quoted.
It was always an interpretation in terms of a "historical views of the thirties," that a savior of a collapse would distinguish the rescue of one group over another?
So we are defining this zone of extreme possibilities, in changes thought about in context of cultural extremes of democracies?
Shock Therapy? This is a purposeful initiated phase to new changes within those cultures?
In no way should any alignment of thought about these two extremes be thought of as "existing in society" only that it might be interpreted as such(groups), distilled from the collective conscious?:)
The concept of "governance" is not new. It is as old as human civilization. Simply put "governance" means: the process of decision-making and the process by which decisions are implemented (or not implemented). Governance can be used in several contexts such as corporate governance, international governance, national governance and local governance.Good Governance
As a set, derived from the collective conscious, oversight(governance) distinguishes a level of thinking above as(a "super" conscious?), and helps to distinguish a "topological movement(dynamcial exchange between polarities?)" within the context of that collective realization?
Sorry for capitalizing here on comments...last one on this topic.
In this Chautauqua I would like not to cut any new channels of consciousness but simply dig deeper into old ones that have become silted in with the debris of thoughts grown stale and platitudes too often repeated.
I feel most people confuse organization -systems- with type 2. Intel chips parts, or my work are black boxes, but people decided the black boxes organization.
Type 2, as said, is not "better" than individual consciousness, it's different. Bird flying in pack is a good example. Ants and bee are even better ones.
In that sense, even ecosystems can be considered of type 2 intelligence. Sociology covers this field too.
What people are interested in is the happening of a type 2 "conscience" -avatar like. What is the difference between a gas giant and a star? can a sparkle ignite consciousness in the internet? Would incoming aliens speak to it instead of us?
In that context, conscience is the ability to answer questions. Descartes'"I think, therefore I am" follows a big pondering if he is or not. It follows a question of his.
Google can't answer "do I exist?" (... yet ^^').
A type 2 intelligence can't come out of thin air. It always answer a need, a pressure. And it survives because it's better than what individuals can come up with.
Yet, as humans, we are only barely conscious. I have ideas before I can be conscious of them. I'm not conscious of my hearth beating all the time, or how my eyes interpret what is written on this screen. Consciousness is a little boat floating on the sea of our cognitional activities.
Conscience out of an human brain is a concept, nothing else. A idea to play with, art. (speaking of art, I like Masamune Shirow approach to it a lot). Yet type 2 intelligence surround us all the time, we are just not wired to see it properly, like we are not wired to see N dimensions.
If you had never understood this flocking behavior, or how a school of fish could move "as a unit" then what use for any intelligence to write the program that allows us to see the behavior in nature?
Not just in the natural settings, but in societal conditions that allow us to interpret behavior of economic attitudes complacent to the fate of the programs written to choreograph the whole process?
What is it that we can identify as the "substance of the whole process" that runs through the very economic viability free from human constraint and we do not need the programmers to write the programs?:)
I regard now the duo R. Brout and F. Englert as of type II intelligence. As being complementary they extend each other, what makes them connected in summary.
Best, Kay
8:22 AM, August 08, 2010
On a recent flight I was reading Jaron Lanier's book You Are Not a Gadget: A Manifesto. I got stuck somewhere in the middle and then dozed off watching Avatar. (Brilliant. Best movie I've seen for a long time, even on a tiny in-seat monitor.) This combination got me thinking about a common theme in both: collective intelligence. Lanier is a skeptic. He writes “The intentions of the cybernetic totalist tribe are good. They are simply following a path that was blazed in earlier times by well-meaning Freudian and Marxists [...] A self-proclaimed materialist movement that attempts to base itself on science starts to look like a religion rather quickly [...] The Singularity and the noospehre, the idea that a collective consciousness emerges from all the users on the web, echo Marxist social determinism and Freud's calculus of perversion.”And later, in a section “Why It Matters,” he writes “Emphasizing the crowd means deemphasizing individual humans in the design of the society.”
I am very sympathetic to many points Lanier is making, but I dislike the “Manifesto”-style in which he's trying to lead his arguments. In any case, should I make it to the end of his book, I'll write a review. For now however, I want to focus on the topic of collective intelligence, for despite all the words Lanier is quite fuzzy on the use of terminology. I started wondering: What do people actually mean when they talk about collective intelligence? Are they really all talking about the same thing? I came to the conclusion that there's two different notions of collective intelligence, and I thought I'd amuse you with writing about a topic that I know pretty much nothing about.
Setting the Stage
First, let us be clear on what we're talking about: A collective or a group or a crowd will in the following simply be a set with elements. The elements of that set operate on input and create output. I will refer to the input and output likewise as “knowledge.”
I further do not actually want to talk about the “intelligence” of a collective in the common sense for two reasons. One is that a collective can be intelligent one day and stupid the next day. I'd rather focus on one particular process the collective makes instead of assigning a qualifier to it as a whole. A process could be a decision as well as a direct action. Second reason is that using the term intelligent isn't useful without defining it. Thus, I would instead like to talk about collective processes that are beneficial for the collective. That's basically because I don't think measures like the IQ are particularly meaningful to assess intelligence since they also test for knowledge. Sure, the both are in humans correlated with each other because both draw upon the functionality of the same brain, but in principle it's different things: The one, knowledge, is the input a system has to work with, the other one, intelligence, is the procedure by which the input is processed. Clearly, to make good decisions both is needed. That's true for individuals as much as for groups. Consequently, to make beneficial decisions the collective needs to be well-informed as well as have a good way to process that information. Googeling doesn't make you more intelligent. It gives you more information.
Thus, what I will mean in the following with “collective intelligence” is the ability of a set of elements operating on some input to perform processes that are beneficial for the collective. Note that you'll have to define what your set is before you can make any statements. I know, it sounds pretty abstract, but it will be useful in its generality for the following.
Collective Intelligence Type I
[Image]The first sort of collective intelligence is simply the gain in knowledge that you can get when you bring many elements of your set together. If every element brings some knowledge, you've now trivially more knowledge together. But that isn't the interesting aspect. The interesting aspect is that you can give the elements of your set the possibility to operate on each other's output, which means that you can create more knowledge than you could have done had they be disconnected.
There's many examples for this sort of collective knowledge. It's what you count on if you bring smart people together and let them talk to each other. They'll exchange ideas, they will build on each others' conclusions, and that way they can produce something genuinely new, say, a groundbreaking paper. I'll count that as a process that's beneficial for the collective. (A recent meeting of this sort is this one, report unfortunately in German.) It is this sort of collective intelligence that has become vastly easier to make use of with the internet, web2.0 and other advances of information technology. It is much easier today than it was two decades ago to give people with shared interests a platform to exchange their ideas.
Note that I carefully wrote that you can get a gain of knowledge by better connectivity. It is however not a given. Just providing a possibility to share knowledge is not necessarily a way to arrive at a good decision, conclusion, or even useful creation of knowledge. Under certain circumstances, too much sharing of knowledge actually dumbs down a group because it reduces heterogeneity. Besides, human cognitive processes are messy and affected by all sorts of biases. As a consequence, the benefit of sharing knowledge depends on how the knowledge is shared, for example because judgement about the truth-value of a piece of knowledge is often dependent on the source it came from. These are some pitfalls that Surowiecki pointed out in his book The Wisdom of Crowds (read my review here). To exploit the additional knowledge gain you get from connecting groups of people you thus have to do some research about the effects and side-effects of social interactions.
You, and you, and you and I, we are some sort of collective and we exchange knowledge. If you give me a piece of information that's useful for my work or if you learn something from me, it's beneficial and I would argue we're thus part of an collective intelligence in that sense. Clearly, the internet has opened a vast potential for this. There's all sorts of unused possibilities in connecting billions of people online which we have only begin to explore. However promising, this sort of collective intelligence is at the same time of a very trivial sort. We're not actually doing anything new here. People have talked to each other and exchanged ideas as long as humans have wondered how to best skin a bear. The difference is just that now we're connecting more people faster and easier.
A mathematical example for this sort of collective intelligence is a group with some basis elements and operators acting on them to create the full group. The basis elements are in this case the knowledge you start with. The operators are the collective. If you only allow each operator to repeatedly act on the same element (his own “knowledge”) you'll generate only a small part of what you'll generate if you allow all operators to act on all elements (use all knowledge).
Collective Intelligence Type II
[Image]The first type of collective intelligence is common and readily found in human groups, but in my opinion actually not the interesting type. The interesting type of collective intelligence is one in which knowledge is created by the collective as a whole and not by any of its elements. An example for that is your brain. Your brain is some sort of collective. It consists of neurons that process input and create output. Yet the thought processes that allow you to make conclusions are circuits in your brain that are not assigned to any specific neuron. The steps of your decision can not be broken down to processes on the elements simply because they don't exist on that level. Your intelligence is a truly emergent feature. It doesn't make sense on the level of a neutron.
Note how very different this is to the first example of bringing together elements that create knowledge individually. The first type of collective intelligence is very common among human groups, the latter isn't. Wikipedia is an example of the first sort of (trivial) collective intelligence: It thrives from adding up the knowledge of many individuals. It doesn't actually collectively create anything truly new. The same is the case for crowdsourcing: Posing a problem to a large group of people allows you to use all their knowledge as well as their intelligence. Yet you're not creating anything that wasn't previously there already.
A simple example of actual collective intelligence of type II is the starting point of Surowiecki's book. Have a group of people estimate something like the number of marbles in a jar or the weigh of an ox or something like that. Then take the average value. It will generally give a pretty good result, basically because individual errors average out. (This is an example that does not work well if you allow people to exchange their guesses in advance, it will skew the result. Recall above cautionary note about how to exploit collective wisdom.) What's new here is that one has added a process that was not previously there, a procedure to aggregate individual knowledge other than just adding it up. Note here that it's relevant to first define what you mean with the collective you're talking about before you decide what sort of intelligence you're dealing with. You could easily enough add Bob to your ox-estimating group and what Bob does is that he asks everybody for their guess and takes the average value. Now who is intelligent: Bob or the group? It would depend on whether you did count Bob with his input processing as being part of the group originally, so one has to be precise. Anyway, in this simple case one is just taking an average value, not a very sophisticated aggregation of knowledge, but there are less trivial cases.
For example our economic system, according to the standard general equilibrium theory, the well-known interplay of supply and demand that, ideally, results in pricing products in such a way that the economy runs maximally efficient. The equilibrium that one works towards is not something known by anybody participating in that system. The aggregation mechanism is a free market economy. Now one can debate how well that actually works and under which circumstances the model doesn't apply, but that's not really the point here. The point is that this “invisible hand” of optimization is a features that adds something truly new, some intelligence that is not present on the individual level.
Other examples for human collectives are arguably political decision making processes. Again, one can debate how well these work and how beneficial the outcome really is, more research is clearly needed, but it suffices to say here that at least they work better than nothing at all. The thing is however that for human collectives you need some mechanism to bring back the insight gained from the aggregate to the individual level, either by communicating the result of a decision or directly by converting it into an action or recommendation. Just leaving it standing inaccessibly at the aggregate level isn't useful because actions are still made by the individuals.
I want to add another example of human collectives here which is the academic system. In the academic system we do not have written down rules to aggregate knowledge, neither do we have a model for how it works. However, if you look at the history of science, we nevertheless de facto have some aggregation mechanism. It's not like there was ever a vote whether or not Coulomb's (wrong) magnetic field law should be kept in the textbooks, yet here we are without it. I think Smolin in his book The Trouble with Physics (read my review here) summarized it well: “Science has succeeded because scientists comprise a community that is defined and maintained by adherence to a shared ethics.” The shared ethics Smolin is talking about is basically some sort of scientific method. Again one can debate whether Smolin's suggestion is the best way to set up the system, but that isn't the point here. He's right in that, written law or not, scientists have used some sort of ethics and that's what made the scientific community more than just a bunch of smart people. It has created a solid and growing body of knowledge that spreads though educational systems and is the driver of innovation. Note however that this sort of collective intelligence does not create a theory, it (ideally) merely singles out the ones to be kept.
Another example for this sort of collective intelligence outside the human realm is the DNA of the bacterium Escherichia Coli containing information in its topology and not only in its sequence. That's not some “knowledge” that any of the elements of the DNA string encodes - it can only be read off from the whole string. An example of the mathematical sort might be a manifold. Consider every point of the manifold the knowledge and parallel transport by some infinitesimal step the elements of your collective. If you only look at the local surrounding you'll never figure out additional information in the topology of the whole thing. You'll have to do something more for that, like looking for closed loops.
Limitations
Clearly for me as a scientist the interesting question is what can collective intelligence do for scientific progress. I think we're doing well with the first type of collective intelligence. It is frequently used. The second type of intelligence is however one that does not exist for the creation of scientific knowledge and I am not sure it ever will.
What would it mean, this collective intelligence of type 2 in science? Consider a large group of scientists, maybe thousands, working on a notoriously difficult problem, possibly for decades. They'll collect and publish many pieces of knowledge. Collective intelligence of type 2 would be an aggregation process not working on the individual level that finds a solution to this difficult problem from the pieces scientists have found. However, the aggregation process that commonly worked for this is an individual finding the right pieces and being able to draw the right conclusions, thus type 1 of collective intelligence. Remember what I said earlier, as long as our society is run by the actions of individuals you need some way to bring back collective knowledge to the individual level, otherwise it's useless. But what mechanism do we have for that in science other than a scientist?
There are some few cases in which indeed scientific conclusions have been drawn by aggregation of accumulated knowledge. For example finding relations between seemingly distinct (medical) topics from citation analysis. But these examples are rare and I'm not sure how far one can ever get with this. This then opens the following question: Can it happen that we will not be able to arrive at some insights simply due to the limitations of the human brain to recognize knowledge that is not present at the individual level? And what is the next step of evolution? Can we ever go beyond that?
I've argued there's two types of collective intelligence: One in which bringing together elements of a group and connecting them creates output that was not possible to create without the connections. In this case however, the output is still created by the single elements, it's just that the connections allow more of it and thus making them can open unused potential. This is the trivial, type 1, sort of collective intelligence. The second, more interesting, sort of collective intelligence one has when the knowledge is not present at the individual level at all. It is contained on the collective level, and its production cannot be assigned to any element of the group in particular.
For what groups of humans are concerned, the first sort of collective intelligence is already well in use and with a better understanding of social dynamics and psychology we will be able to exploit more unused potential. For what the second type is concerned, I think one should rightfully be skeptic. It is questionable how much can be achieved by it, and unclear how it can be created or used. We're a long way from global collective intelligence like that connecting the lifeforms on Pandora.
In his book, Lanier offers the reader a list with suggestions for “what each of us can do” featuring the item “Write a blog post that took weeks of reflection before you heard the inner voice that needed to come out.” Well, I report that I've been reflecting on this for some weeks, though not continuously. After all, I have a job to do. My inner voices are typically busy with some other things than wanting to come out, like reminding me I should put the laundry into the dryer now. In any case, I hope it won't take you several weeks to read it ;-)
posted by Sabine Hossenfelder at 11:32 AM on Aug 2, 2010
"Collective Intelligence"
32 Comments -
Dear Bee,
I'm not so sure we really have examples of collective intelligence, type II.
In the free market, you have individuals making decisions in the context of the information that the market provides - which is the aggregrate of everyone else's decisions. But the individual element is there, it is just not published, cited, etc.
In the scientific world, you have an individual putting together the pieces of the scientific puzzle in the context of the information provided by all the other scientists. Simply because the individual element is in this case published, recorded, talked about does not mean that it is different from the free market.
We could have had a "anonymous" culture in the sciences and arts too.
I think Rutherford said that he could not achieved what he did without the expert glass-blowers that were available to his lab. CERN would not work without the electronics, which is at its advanced state because of the efforts of hundreds of thousands of people, many anonymous. Likewise, the people at Canon designing a DSLR or Steve Jobs and Apple designing a phone - they are doing this with the input from the market. I don't see much difference between the free market and the scientific world here.
Regarding your example of estimating the beans in a bowl, true, it encodes information that no one person has. **BUT** it is in part because there is an objective answer to how many beans are there in the bowl. The similar "wisdom of crowds" also results in a "GOD", of whom no one person is sure of all the properties. Prices in the free market are more akin to "GOD".
Further, the prize for guessing the beans in the bowl goes to whomever guesses the closest, and not to the collective; the same is true of science.
Best,
-Arun
1:52 PM, August 02, 2010
Collectives regress toward the mean as outliers are shunned for being devisive. Harvard computing attacked Bill Gates creating terse BASIC. Collectives' outputs are beige rainbows.
Fundamental observables hate emergence. Symmetry-breaking ruins elegant theory. The Standard model chokes on mass; GR chokes on spin. Matter predominates over antimatter while physical theory plays Lysenko.
Tolerances average in knowledge but add in illusions of knowlege. Religion reaches no conclusion. Einstein was not hoisted upon jubilant shoulders. He waited until the old guard died. Eddington crushed Chandrasekhar. Yang and Lee were pariahs until experiment.
"the academic system" American public education had McGuffey's 1836 First Eclectic Primer, and it worked (even for the Irish). President Johnson collectivized American education: 11 April 1965, "Elementary and Secondary Education Act." It was improved means to deteriorated ends, extruding intellectual cripples at astounding costs. A BS in STEM now costs $(US)200,000 at a public university. Outsource!
If you only look at the local surrounding you'll never figure out additional information in the topology of the whole thing. You'll have to do something more for that, like looking for closed loops.
http://www.mazepath.com/uncleal/erotor1.jpg
The worst it can do is succeed.
Progress requires a sharp nail and a sound pounding - the antithesis of collectivism. Development is collective, squeezing out intellect and craftmanship by statistical process improvement. Journals publish product. Never confuse product with process. Process like birth is an unwelcome painful mess after the main player gets... screwed.
2:58 PM, August 02, 2010
Michael Nielsen writes a lot about this sort of topic (indeed, his last two posts in months were about a book titled "Collective Intelligence"), and I recall at least two really interesting examples, one of which I think might satisfy your criteria for type II; specifically, this post describing an annual programming competition held by MathWorks: "The competitors don’t just submit programs at the end of the week, they can (and do) submit programs all through the week. The reason they do this is because when they submit their program it’s immediately and automatically scored. [...] What makes this a collaboration is that programs submitted to the competition are open. Once you submit your program anyone else can come along and simply download the code you’ve just submitted, tweak a single line, and resubmit it as their own. The result is a spectacular free-for-all. Contestants are constantly “stealing” one another’s code, making small tweaks that let them leapfrog to the top of the leaderboard. [...] The result is that the winning entry is often fantastically good. After the first contest, in 1999, the contest co-ordinator, Ned Gulley, said: 'no single person on the planet could have written such an optimized algorithm. Yet it appeared at the end of the contest, sculpted out of thin air by people from around the world, most of whom had never met before.'"
A similar example he described was an internet-based chess game that appears to have made the group much stronger than any individual participant, as they gave Kasparov a run for his money.
Nielsen also describes the search for the Higgs in a similar way (as Arun mentioned above), since no one person will be able to fully understand the complete operation & results of the LHC.
Two more related examples: Nielsen discussing Intel's design process, "[An Intel engineer] stated that there was no one person who came even close to understanding the chip in its entirety. Instead, Intel has fashioned a very clever social process where that is not necessary, and the engineers working on the chip only understand it collectively."
And Matt Ridley's TED talk, "When Ideas have Sex," which is just full of great ideas which (I think) relate to this.
But if you don't think any of these qualify as type II, maybe I should re-read your post.
3:08 PM, August 02, 2010
It seems that type I is potentially instantaneous in the sense that it merely realizes an output as a result of the connection of information (be it developed inside each one and/or as the product of collaborative work per se), whereas type II would require evolutionary (process-related) conditions, whose integration is not "instantaneously" present but requires some "historical" and/or continuously changing factors (be them local or global in nature) in order to provide an output (at least, according to cody's previous examples).
4:00 PM, August 02, 2010
This comment has been removed by the author.
5:24 PM, August 02, 2010
I believe in Collective UN-Intelligence. Unlike the CIA, no, come to think of it, that's an excellent example.
Think Mob Violence (for example "the mob" attacking Frankenstein's castle. If they had brains they would have cheered Dr. Frankenstein for discovering immortality).
"One boy is a boy, two boys are half a boy, and three or more boys are no boys at all" is heard on many school US Elementary school playgrounds by the teachers.
What is the problem with The Singularity? How fast is AI progressing? Computers are idiot savants, like Dustin Hoffman's character in "Rain Man." IQ=17.
Humanity's IQ is 100. Less so in countries that worship revenge.
9:52 PM, August 02, 2010
Interesting post.
Something to think about.:)
Best,
2:13 AM, August 03, 2010
Collective Intelligence
There may be some benefit with your perspective along side of the mapping image currently constructed?
How could one see your categorizations being specific with the divisions in Type I and II exhibiting further categories? Does then become to complex?
Best,
2:32 AM, August 03, 2010
Dear Arun,
The point I was trying to make with the economy is that you have an additional structure that allows the whole system to be "more intelligent" than just putting together many people. The additional structure in case of the economy is introducing money and prices for goods. Letting this system operate gives you a better result than, say, asking some king how to best price goods. In that sense the "whole" is more than the sum of the parts. But you have to do something for it. The knowledge that you get in the end, the "true value" of a good, is not something that was decided upon by any of the people participating. It is a result of this particular aggregation mechanism.
As I wrote, I agree that in academia when it comes to solving scientific puzzles there's no collective intelligence of type II (except for a few trivial examples). The example that I mentioned as an example for CI of type II in academia was not solving scientific puzzles, but deciding which theories are the most useful to keep. As I mentioned, this "system" is not well understood and it doesn't built on written down laws and we have no good model for it, but I think it does exist anyway, in some vague sense because scientists do use a shared ethics indeed (or most of them do), so there's your additional structure.
You need an objective answer if you want to quantify really how intelligent a collective is or isn't. That a decision is "beneficial" is not easily quantifiable. The number you get for the marbles is quantifiable in its precision. The marbles-in-a-jar example is a very trivial one, but there too you have made use of an additional structure on your group that allows it to produce new output.
Well, yeah, Smith's reference to the "invisible hand" definitely smells like God's guidance, but it's a metaphor for there being a mechanism that the individual doesn't determine. These are features that are very hard to understand because humans have little if no intuition for what effects can emerge in a crowd. Here is a recent very sad example of that. Best,
B.
4:46 AM, August 03, 2010
Hi Steven,
Yes, collectives can also be more stupid than individuals. As I wrote, it depends on how you let the individuals interact and how you arrive at a new knowledge or a decision or an action and also what the problem was to begin with. Of course if you do it wrong, the result will be bullshit. But the point is, if you do it right, everyone in the group can gain from it. The MIT has a Center for Collective Intelligence whose studies are dedicated to exactly these questions: under which circumstances can a group be more intelligent than a person. And when does that fail? I think these are extremely important questions. Not so much for companies who want to do crowdsourcing but because it's questions that we need to answer to understand the functionality of our economic and political (and academic) systems.
What's worrisome though is if you look around, the properties that you see in how people connect and how information is distributed are exactly those that are known to promote dumb rather than intelligent decision: oversharing (of information and opinions) reducing heterogeneity (leading not necessary to streamlining, but more commonly to polarization), and information being distributed by authorities which vastly amplifies the impact of mistakes and inaccuracies. Think think, where are we headed?
I can really recommend Surowiecki's book. It is, in contrast to what many people seem to believe, *not* a praise of the wisdom of the crowds, but a summary of the circumstances under which crowds are wise or aren't. It is also well written. Best,
B.
4:57 AM, August 03, 2010
This comment has been removed by the author.
6:45 AM, August 03, 2010
Hi Bee,
A very interesting piece referencing to a book I will now have to read as I find time. You make a good distinction between what you find as the two types of collective intelligence and yet for me they stand as Type 1 being cooperating intellects and type II as a fully shared or unified intellect.
However the key difference between the two for me relates to what is called consciousness or self. In Type 1 there always remains relevant many selves or consciousnesses, while in the other at first glance it appears there could be room for only one. However then I am reminded of Penrose’s arguments where when it comes to living things all are conscious right on down to the simplest one celled form,, with self awareness not being something attained at a certain level of complexity yet rather only increases by degree; that is as the light a lense can gather increase with the size of its aperture.
So In this respect (as you pointed out) one might consider the human mind itself already an example of a Type II intelligence, which would also require a advanced form of consciousness to have it work and this is the aspect we tend to overlook. So in order for a collective of human minds to form a super intelligence it would seem to me it would mandate the creation of a super consciousness. Perhaps then as a result being that the new knowledge it comes to discover can only be made fully aware to this new consciousness and then never fully understood by any individual mind. That is just as one brain cell is not aware of although does benefit by what is known by the collective we call a mind. For me the consequence of managing such a feat would not just create a superior intelligence yet in the process a new and distinct being. I would then wonder once created if it would ever want or allow for it to be separated again or would we be content to surrender ourselves forever to have it created?
Best,
Phil
6:51 AM, August 03, 2010
Hi Phil,
"So in order for a collective of human minds to form a super intelligence it would seem to me it would mandate the creation of a super consciousness."
That would be the final consequence I think, yes. But I'm not sure we'll ever get there, and certainly we're not even close to. We're more on a level where we occasionally find together to collectives for one or the other purpose and try to find intelligent ways to make decisions that serve best our needs. It would take significant changes to human nature to ever create something like a "super consciousness," I'm not sure these changes can or will ever take place, and the internet certainly isn't even remotely sufficient for that. It might be sufficient however to do exploit some advantages of collective intelligence. Best,
B.
7:42 AM, August 03, 2010
Philip Tetlock: Any individual expert is likely to be wrong.Why experts are usually wrong.
I've basically nothing to add about it.
11:16 AM, August 03, 2010
BTW I don't believe in collective intelligence neither - for example dense aether model is incredible simple, but no one from myriads of aether proponents or so-called "natural thinkers" has considered it as anyway.
11:19 AM, August 03, 2010
This comment has been removed by the author.
4:05 AM, August 04, 2010
Hi Bee,
Technologies such as the internet at best facilitate cooperative collectives, as although it allows the interaction of multiple minds it has no power to form a super consciousness to unify them in awareness and purpose. However I do think we will in time discover as to invent technological means to create such an interface, with perhaps quantum computing being the key as it holds the promise of exploiting the non local aspect of nature.
I’m reminded of a blog piece I wrote awhile ago that showed that although leaf cutter ants individually are not conscious of such a capability and purpose, that as a collective they do support the group through what we would find as the selection, care and nurture of another species they all then consume ,which we would might find as having developed the technology of farming.
In similar fashion I can envision that interfaced human minds could combine to form a consciousness that although not known to the individual would form to exist in terms of the whole that could act in unified singular purpose and capable of things well beyond that of the individuals. Then for that matter, in some respect, we might cite society as a wholeness of purpose and utility exceeding that of any individual and yet benefitting all. It has one to wonder if there is even now a difference between recognizing a distinction between the consciousness of a human and that of humanity?
“It is absurd to suppose that purpose is not present because we do not observe the agent deliberating. Art does not deliberate. If the ship-building art were in the wood, it would produce the same results by nature. If, therefore, purpose is present in art, it is present also in nature. The best illustration is a doctor doctoring himself: nature is like that. It is plain then that nature is a cause, a cause that operates for a purpose.”
-Aristotle- Physics (350 B.C)
Best,
Phil
4:36 AM, August 04, 2010
Hi Bee,
in my opinion, some extreme sort of your type I intelligence is the work by R. Brout and F. Englert: arXiv:hep-th/9802142, who were complementary working together.
Best, Kay
9:23 AM, August 04, 2010
Hi Phil,
"In similar fashion I can envision that interfaced human minds could combine to form a consciousness that although not known to the individual would form to exist in terms of the whole that could act in unified singular purpose and capable of things well beyond that of the individuals. "
Yes, I can imagine that as well. But that I can imagine it doesn't mean it's likely to happen. I'm not at all sure we'll ever be able to create such an interface and, if we were, would be able to use it in a beneficial sort of way.
"It has one to wonder if there is even now a difference between recognizing a distinction between the consciousness of a human and that of humanity?"
I don't think that presently there's anything that could plausibly be called a "consciousness of humanity." What we have is a collection of consciousnesses for which the whole is pretty much the sum of its parts and not more. With some exceptions (see post) that however do cover only very few areas of human actions, mostly locally so, and for what I am concerned don't even work particularly well even in these cases. Best,
B.
10:58 AM, August 04, 2010
This comment has been removed by the author.
8:30 PM, August 04, 2010
Although not about collective intelligence, Lanier has inspired a certain appreciation that I develop into the following point:
A purely computational intelligence (as some like Dan Dennett suppose even we are) cannot formulate the thought of special real existence apart from logical structures (ie, such a mind cannot even represent disbelief or an alternative to modal realism/MUH.) That's becasue computations are just math, they can't represent "this is just math" versus "this is my thinking here in a real material world." Well, I don't think CI/AI is true - we are not gadgets! - but that would be the implication. (Something breathes "afierie" into the equations, my Captcha which is often uncannily relevant.)
8:33 PM, August 04, 2010
This comment has been removed by the author.
7:03 AM, August 05, 2010
Hi Bee,
I would say that as still the forming of consciousness is much of a mystery that being sceptical is resultant more of what is not known as opposed to what is. What is known is that as life continues to evolve in complexity that with this increase comes a corresponding increased ability for it to communicate, Therefore I would contend by virtue of what is known there is no reason to understand things have reached any kind of limit in respect either life continuing to evolve in complexity or communication. As for the advantage to such communication I would agree that much like the leaf cutter ant’s inability to fathom the implications of the collective’s farming a singular human consciousness may be similarly fated. None the less a failure to understand has never been evidence to have something not to exist or able to. This is also essentially the contention of Aristotle’s in what I had previously quoted.
Best,
Phil
7:07 AM, August 05, 2010
A shift between extremes?
Presented factions within the differences between a social order, and a elite group?
Presented as an idea between Type I and Type II.
Zietgeist and Capitalistic Movements( how are these defined in terms of a elite groups and dollars)?
Best
11:46 AM, August 05, 2010
The Relief of Burdens
It was always an interpretation in terms of a "historical views of the thirties," that a savior of a collapse would distinguish the rescue of one group over another?
Best,
11:56 AM, August 05, 2010
So we are defining this zone of extreme possibilities, in changes thought about in context of cultural extremes of democracies?
Shock Therapy? This is a purposeful initiated phase to new changes within those cultures?
In no way should any alignment of thought about these two extremes be thought of as "existing in society" only that it might be interpreted as such(groups), distilled from the collective conscious?:)
Best,
12:01 PM, August 05, 2010
The concept of "governance" is not new. It is as old as human civilization. Simply put "governance" means: the process of decision-making and the process by which decisions are implemented (or not implemented). Governance can be used in several contexts such as corporate governance, international governance, national governance and local governance.Good Governance
As a set, derived from the collective conscious, oversight(governance) distinguishes a level of thinking above as(a "super" conscious?), and helps to distinguish a "topological movement(dynamcial exchange between polarities?)" within the context of that collective realization?
Best,
12:31 PM, August 05, 2010
"Rising Above the Duality?"
Sorry for capitalizing here on comments...last one on this topic.
In this Chautauqua I would like not to cut any new channels of consciousness but simply dig deeper into old ones that have become silted in with the debris of thoughts grown stale and platitudes too often repeated.
Zen and the Art of Motorcycle Maintenance Part 1 Chapter 1.(Bold added by me for emphasis)
Drawn any pictures of mountains lately as you listen to the lectures? Climbed any mountains lately?
Best,
Best,
12:49 PM, August 05, 2010
I feel most people confuse organization -systems- with type 2. Intel chips parts, or my work are black boxes, but people decided the black boxes organization.
Type 2, as said, is not "better" than individual consciousness, it's different. Bird flying in pack is a good example. Ants and bee are even better ones.
In that sense, even ecosystems can be considered of type 2 intelligence. Sociology covers this field too.
What people are interested in is the happening of a type 2 "conscience" -avatar like. What is the difference between a gas giant and a star? can a sparkle ignite consciousness in the internet? Would incoming aliens speak to it instead of us?
In that context, conscience is the ability to answer questions. Descartes'"I think, therefore I am" follows a big pondering if he is or not. It follows a question of his.
Google can't answer "do I exist?" (... yet ^^').
A type 2 intelligence can't come out of thin air. It always answer a need, a pressure. And it survives because it's better than what individuals can come up with.
Yet, as humans, we are only barely conscious. I have ideas before I can be conscious of them. I'm not conscious of my hearth beating all the time, or how my eyes interpret what is written on this screen. Consciousness is a little boat floating on the sea of our cognitional activities.
Conscience out of an human brain is a concept, nothing else. A idea to play with, art. (speaking of art, I like Masamune Shirow approach to it a lot). Yet type 2 intelligence surround us all the time, we are just not wired to see it properly, like we are not wired to see N dimensions.
3:50 PM, August 06, 2010
This comment has been removed by the author.
5:45 AM, August 07, 2010
Lets look at the Boids?:)
If you had never understood this flocking behavior, or how a school of fish could move "as a unit" then what use for any intelligence to write the program that allows us to see the behavior in nature?
Not just in the natural settings, but in societal conditions that allow us to interpret behavior of economic attitudes complacent to the fate of the programs written to choreograph the whole process?
What is it that we can identify as the "substance of the whole process" that runs through the very economic viability free from human constraint and we do not need the programmers to write the programs?:)
Best,
10:13 AM, August 07, 2010
Hi Bee,
I regard now the duo R. Brout and F. Englert as of type II intelligence. As being complementary they extend each other, what makes them connected in summary.
Best, Kay
8:22 AM, August 08, 2010