Mr. Stephen Downes recently commented on a previous post critiquing Collective Intelligence (another Google Alerts fan?). I have reproduced it below, with my responses. He and I may be talking apples and oranges, but I will give my best shot.
Gross mass-based phenomena such as yes-no votes are not emergent phenomena and are not what is meant by ‘collective intelligence’.
That would be like attempting to analyze the meaning of a set of pixels by counting how many are ‘off’ versus ‘on’, instead of looking at the organization and recognizing in that a picture of Richard Nixon.
The fruit of collective intelligence, which I (and others) have described as an emergent phenomenon, results from the linkages and connections between individuals, and not a counting of properties (such as survey results) of those individuals.
On a fundamental basis of communication theory, as I understand it from Wilden’s essays, “information” is the consequence of quantifying, or “digitalizing”, relationships through codes, such as numbers, alphabets, phonetics, etc. Humans digitalize information to express “Exchange Value”, which is simply a qualification of a relationship independent of its “Use Value”, or matter/energy value. Thus “male” is a Use Value; the expression of “brother” “father” “son” are Exchange Values.
On this basis, I interpret gross mass-based phenomenon as a form of digitalizing a topic. Where there is quantification of agreement/disagreement, or like/dislike, there are relationships being codified. My argument is that the patterns that are created by these tactics across a broad communication ecology, such as Yahoo News, may indeed be construed as relationships between groups of individuals, and in the aggregate, creates an emerging pattern if one chooses to categorized topics and generate statistics (which I believe we do in informal ways as we read, such as, are there more people who agree/disagree that the Huteree bust was a “good bust”).
I understand that it is not the yes-no event itself that constitutes Collective Intelligence, but it is the kernel by which it is created. This begins to push the debate into metaphysical terms (which I will avoid). But it points to whether we may choose to acknowledge the point at which emerging information phenomenon occurs as the moment a yes-no vote is cast, at the moment when a statistically significant sample has been achieved, or when people express how they feel to each other about the patterns created by what other people feel.
My point is that I believe that the manner by which relationships are codified determines the resolution of the emergent pattern within the Collective Intelligence ecology. I agree that the more meaningful channels of interpersonal connections lay between individuals, but the “results” being referred to in Mr. Downes description is not a phenomenon that can be defined as we choose. The relationships are saturated with innumerable Exchange Values, and human pattern-seeking behavior will find information in the patterns in any way it needs to to help reduce uncertainty about that linkage. We cannot control by what measure people choose to create distinction between themselves and others.
I consider Collective Intelligence a legitimate description of this phenomenon, though I am apprehensive in describing it as the bearer of sweet fruit. Again, Mr. Downes:
This emergent knowledge is not intended to compete with, or replace, qualitative or quantitative knowledge. The assessment of whether Obama is a Muslim is not the subject of collective intelligence, no more than the assessment of how many children he has would be based on what colour jacket he is wearing. Just as we should not confuse qualitative and quantitative data, we should not confuse wither of those with data describing connections and relations.
As to whether observation of emergent phenomena based on linkages or relations is based on “inherent validity”, or “objective measure, evidence of intellectual virtue, rational thinking, or consideration of viable alternatives”, depends on “reliability and validity of information”, and demonstrates “smart, correct, educated, having wisdom, having valid experience in an area of knowledge or skill”, such data – just like assessments of quality or quantity – are and ought to be subject to assessments of reliability, and not accepted as fact uncritically.
I agree with this description in principle. However, I do not believe that is what occurs in practice. On the contrary, I believe some people think Obama is a Muslim for precisely because of inaccurate interpretations of unrelated information, such as misunderstanding that a turban is a garment reflecting tribal or cultural practice rather than an exclusive representation of one’s religious faith – yet an image of Obama wearing a turban is what is used as evidence to sustain the argument, and if enough people “agree” with it, it becomes codified into “truth”.
Mr. Downes is entirely correct that emergent patterns should be subject to quality assessment. However, the instruments by which information is generated will largely determine whether or not this a feasible process. It is arguable that if the debate over Obama’s religious faith were conducted in the format of the Lincoln-Douglas debate, we may arrive at a point of finality about the subject with few points left uncovered, and proponents of each position relegated to a position of public accountability for their beliefs.
In the Web milieu, however, we lose both the attachment and personal accountability to one’s opinions and identity (or at least certain parts of it) as well as the context for immediate exchange and counterpoint as a continuum. (We also gain some things in the Web context, but I see those gains as amplification of dissemination rather than increases in information resolution).
In this realm, it is not people who suffer, it is information who suffers. (I deliberately ascribed information as a “who” to emphasize my sensitivity to information integrity). If we felt that information had feelings and needs, perhaps we would care for it more so than it being a commodity to be swished around because we have the means to do so with light-speed efficiency and indefinite boundaries of time and space.
Just as nobody would accept a claim like “Obama is purple” or “Obama is really two people” uncritically, and without corroboration or verification, nor either should we uncritically accept statement like “Obama is a Muslim” or even “this arrangement of pixels depicts Richard Nixon” uncritically, without corroboration or verification.
The idea of emergent properties, or collective intelligence, or (as I would call it) connective knowledge, is not inherently opposed even to the strong realism assumed in the assessment above. It is not inconsistent to assert that “there are facts of the matter” and “these facts are expressed as connective knowledge”.
The point of an assertion that there is connective knowledge is to assert that “this domain of facts is not exhausted by observing qualities and counting entities or their properties; there is a distinct set of facts represented by the *connections* between these entities.” This is a proposition, even when granting the naive sort of realism assumed above, that is difficult to refute, and is not refuted by assertions such as “a large quantity of people express the belief that Obama is Muslim.”
There is tenuous balance toward rational or irrational belief at the median between where a person is inclined to assert scrutiny (locus of control) about the validity of information, and where such scrutiny is not deemed necessary, or socially risky. For example, Sam Harris has observed that it is commonly accepted that humans operate at a certain level of rationality when it comes to choosing a spouse, whether or not to bear children, or buy a certain house, but they abandon that same faculty for making a rational decision about whether it is an appropriate belief to persecute or even murder other people who do not hold the same beliefs because their holy texts tells them to, and their rational functions of scrutiny are somehow precluded due to some taboo or social cost.
I propose this because I feel it is important to acknowledge that the inclination for applying critical thinking towards deep understanding or comprehension is somewhat arbitrary, pregnant with social consequences, and not a dependable function of human experience in the learning environment, be it formal or informal, towards establishing facts or knowledge. Collective Intelligence can be used to effectively achieve a sample, but I am apprehensive about believing humans are inclined to assert critical scrutiny about the meaning of emergent patterns – at least when it comes to social media!
Towards Mr. Downes’ statement that “there are facts of the matter” and “these facts are expressed as connective knowledge”, I agree. However, those facts aren’t always acted upon as an extension of rational thinking. Harris recently asserted in his TED talk that science – or the scientific method – can be used to determine moral right or wrong. (Whether you agree with him or not is beside the point). His premise is whether a certain act based on a belief produces more or less well-being.
Again, one may or may not agree with this statement, but what Mr. Harris proposes (and what I agree with) is a response to the absence of a foundation by which we scrutinize the validity of belief beyond assertions in a holy text or cultural norm (or in our case, Yahoo News). For example, reason is clouded by dogma when it comes to deciding whether it is a morally wrong murder your daughter who has been raped because it defends the family honor.
I am not suggesting that rational thinking is “absent unless threshed out” in human discourse. Instead, I am only asserting that there are risks involved in the reliance upon Collective Knowledge, and those risks cannot be mitigated within the discipline of the collective ecology except by more Collective Knowledge. Thus, we see the difficulty Democrats had in refuting the false “death panel” claim in the healthcare debate.
If we wanted to learn about Obama’s religion – which is not a simple observable or countable property – then we would not sample what people unconnected to him express as beliefs. That’s like determining the colour of grass by counting pebbles on the beach. Rather, we would amass and collect the set of Obamas *connections* and *interactions* with other people and things, and determine whether this constitutes a set of patterns that more typically resembles a person we typically call a “Muslim”….
Asserting that “Obama is a Muslim” based on a poll would be irresponsible, and no person advocating any form of collective intelligence or connective knowledge would assert otherwise.
But asserting that there is some simple observable property that verifies or confirms that “Obama is not a Muslim” is equally irresponsible. Naive realism does not refute connective knowledge when the reality being described is complex, when there is no simple observable or countable fact of the matter.
Again, this a problem in practice, not principle. The ability for the student to derive or build denotative reference to relationships is encumbered by a system of information management (Internet) that seeks to express and organize information in pre-packaged forms of meaning, such as tags or relevance ratings. Thus, if one wanted to Google “the relationship between Obama and his faith”, you would not find connections and interactions in the results. You would find ranked and categorized information, not resources for collecting connections and interactions per se. Mr. Downes is correct that seeking to answer certain questions via polls irresponsible, yet that is essentially the basis upon which information is organized in electronic network systems. The results of tapping into this body of collective knowledge may or may not hold valid information towards seeking a reliable answer.
The problem is further compounded by the recursive nature of crowdsourcing, which imposes an exponential value of relevance (deemed “true”) once enough people assert that certain information is the definitive resource for answering certain questions – and what’s worse is that there is no recourse for undoing this phenomenon.
Connective knowledge, in other words, does not refute or overturn existing knowledge; rather, it offers us a *new* type of knowledge, that *cannot* be confirmed or refuted by simple observation of data; the employment of connective knowledge *is* to assess and evaluate such assertions *is* a demonstration of being “smart, correct, educated, having wisdom, having valid experience in an area of knowledge or skill”.
In the context of something like Wikipedia, this makes good sense. More eyes and ears presumably mean that diverse forms of scrutiny have gleaned bad information from good. I assert caution, however, that not all platforms for collective scrutiny in the ideal sense of it are equal, and that increasing numbers of micro-channels and narrowcasting of beliefs onto smaller and smaller self-regulated platforms reduces the probability that a diverse audience will uphold the standards for the assessment and evaluation necessary to ascribe validity and reliability to assertions of fact. Wikipedia holds a hallowed position as being an area where we may presume that certain standards and editing behaviors cause a greater probability for validity in certain areas, although it is not necessarily so in certain fringe areas, such as when I observed when it was stated in Wikipedia that Glenn Allison was the first bowler to roll an ABC sanctioned perfect 900 series. (He was the first to do so, but it was not sanctioned. This is important, people!).
Thus we have a continuum upon which we may reasonably believe that information has passed the threshold of connective knowledge scrutiny, though we may never know by some reliable standard where that threshold exists, such as when Senators reliably quote the Congressional Budget Office to assert a bill’s financial viability.
I do not dispute that Collective Knowledge is a powerful tool for performing valuable functions. As a career video content producer and web designer, however, I feel sensitive to the behavior of audiences, especially in the detached, uncontrolled online environment. Now that I am beholden to account for my beliefs about good instructional design practice, I hope to continue to discuss these issues. I would like to thank Mr. Downes for taking a generous amount of his time to respond to my blog post.