notes on "Coases Penguin, or, Linux and the Nature of the Firm" by Yochai Benkler - two existing known modes of economic production: individuals following price signals in a market, or employees following managers commands in a firm. - third mode emerging (and has emerged in the past 15 years already) : commons-basd peer production. - Benkler notes that in free software, production is not organized by firms, but is it significant that there may still be hierarchies (as in linux)? this is very firm-like. however it is true that usually there is nothing preventing someone under another person in the chain of command from "forking" the product if they become frustrated. also I imagine that it is important that price signals (may) be completely absent (in firms, the managers are following price signals-- someone always has to) - preconditions: object of production is information or culture. physical capital necessary from production is widely distributed. - peer produciton better than firms or markets because: lower "information opportunity cost": loses less information about who is best for a particular job. lower transaction costs; easier to find new projects and opportunities - i think one of the questions this paper answers is "why is it that open source software is the path of least resistance to software development so often these days?" that is, why is it not even really something that developers have to think about too much? ("software development" can be replaced with culture creation too) - commons-based production beating corporations, traditionally organized economic production structures, in their own markets - Coase: * costs associated with defining and enforcing property (transaction costs) * firms emerge when these costs exceed benefits * use markets when the opposite is true - you *CAN* make a case that open source software development is completely done because of the expectation of plain economic gain in the future. in ESRian terms, a homestead on the noosphere can be worth considerably money later on in the meatsphere (even ESR ended up with 40$mil in stock at one point) (look at footnote 7 on page 4) although this has some relevance and truth to it, I think the paper is right in that participation in commons-based development cannot be explained by the existing theories of production. - this paper's conclusions are generalized away from hackers or free software (good) - this work may actually imply that community-authored sites tend to be stable and self-sustaining, rather than unstable and in constant need of very careful supervision and planning. that is, it may be that community-authored sites that fail must have really done something to screw up the situation... - i see implications for the battle between publishing firms and community- based publishing (including provisions for editing, review, etc). maybe it is not even possible for firms to compete digitally, without really locking everything down in terms of IP and DRM, etc - i think he cites slashdot and kuro5hin =) - specification problem: imperfection, transaction losses - proximity problem: propery and contract make agents and resources "sticky": people more likely to work with their own firm's resources, etc - human creativity an especially difficult resource to manage, hence putting the assignment of creative responsibility with the individual is optimal! this is an excellent, pivotal point. - weeding out: commons-based production systems (if successful) must have a process for filtering out mistaken judgments producers make about themselves (ratings, rankings, etc) - bottom line: improved allocation of human creativity with commons-based peer production - for information and culture, production consists of : existing information and cultural inputs, creativity, and physical capital (media for storage, infrastructure for transmission) - for 150 years, cost of physical capital was organizing impetus! caused capital-intensive, industrial model to dominate. - falling price of computation has inverted capital structure - primary remaining scarce resource is human creativity! - arguments againts commons-based production (tragedy of commons) : no one will invest if they are not guaranteed benefits, and lack of organization will cause venture to fail. - past few years: rise of non-property based schemes for structuring coop. - thesis one: THERE EXIST RANGES OF HUMAN EXPERIENCE IN WHICH THE PRESENCE OF MONETARY AWARDS IS INVERSELY RELATED TO THE PRESENCE OF OTHER (SOCIO- PSYCHOLOGICAL) AWARDS!!! i think this is pretty important. many computer hobbyists have noticed that the joy extracted from programming seems inversely related to the amount of pay being furnished for it! (That is, it is most fun pursuing one's own interests, rather than something someone is offering money for) another example: academics either selling consulting, or writing research respected by peers - thesis two: when a project is broken up into simple chunks, motivation to participate need only be small. - project needs to be modulary, fine-grained, to be amenable to peer production. (i.e. novels vs planetmath) - successful commons-based peer production projects will accomodate contribs of varying size. - need low-cost integration of separate contributions * this includes "defense mechanism" against malicious or incompetant contribs * many projects are using automated, free integration mechanisms! - GNU GPL a formal "Set of rules" - check out Eben Molgen's "Anarchism Triumphant" - emergence of commons-based production is at odds with rapid expanding of scope of exclusive rights in information - scientific/academic community the original "commons-based" production community (produce a body of knowledge) * value appropriated by SERVICE rather than PRODUCT ... (teaching, contracting, etc) * support from grant sources * motivation from prizes, titles, etc * select own projects - nonprofessional information production: casual conversations, entertainment, everyday life - 3 stages in information production: * utterance * relevance/accreditation * promulgation - net is allowing more decentralization/dissaggragation of the above - NASA clickworkers: amazing results. this might be a new role for NASA; simply an organizer of volunteer effort. * "What the NASA scientists running this experiment had tapped into was a vast pool of increments of human judgement applied with motivation to a task which is unrelated to keeping together the bodies and souls of the agents" - web itself largely takes the role of encyclopedia - interesting comment about how people from wikipedia may also be contributors to everything2 (actually, some are contributors to planetmath as well). i have noticed that people are very adept at "context-switching" between online communities, that is, sensing the boundaries between them and instantly shifting the set of operating norms they act by. this is possible even in real-time chat, this participation in multiple commons-based production loci is another manifestation. has anyone done a sociological/psychological study on this? - doesn't seem everything2 has automatic linking either =] - the self-description of K5 is a little bit off ... these days there is plenty of "noise," though not intolerable. * emphasis on quality enforced by pre-publication peer review, and post publication commentary could planetmath benefit from this pre-/post-publication dichotomy? - interesting point about online gaming: the role of the commercial provider is more to provide an infrastructure and set of tolls than to provide a start to end finished product. - important comment: "... they are playing a game-- but they are spending real economic goods-- their attention and subscription fees-- on a form of entertainment that displaces what used to be passive reception of a finished, commercially and professionaly manufactured good with a platform for active co-production of a storyline." - amazon as example of relevance/accreditation through peer production * i agree that this is really the "killer app" of amazon. this is the single most innovative thing the corporation has done, and possibly stands as the most innovative use of the internet by a retail outlet to date. * purchase circles: group people geographically or by organization, take into account trends in the circle itself. very interesting! - google really has a peer-produced ranking, through the PageRank algorithm (links to a site are a "vote" for that site") - google not self-consciously a product of peer production - slashdot moderation system: give many users small bits of power * troll filters - prevent sabateurs - napster, et al, are peer-based distribution networks - distributed proofreading: very interesting-- this satisfies the "Granularity" criteria for success - Coases' "natural" limit to size and number of firms runs counter to the folk wisdom that monopolies are bound to arise, and firms to grown in size. - information is * non-rival (consuming it does not diminish its availability) * it is produced by inputting more information - physical capital costs of producing information dramatically tending towards zero - human creativity much more variable than labor - CD's mentioned as 20th century manufacturing paradigm; its noteworthy that they too were included in the decentralization via technological progress that the computer revolution entailed (now everyone can make their own CDs, there is no longer anything special about old-paradigm CDs except that they may be MORE restricted) actually, it should be obvious to everyone (record companies, hello?) that consumers will seek the versions of a desired good (CD) that will give them the most ability, so they are just going to make their own CDs or seek black-market custom-burned versions that are unrestricted. the mess we see now with all the copy protection is what arises when a firm-model of production becomes obsolete and doesn't want to die. - treat all approaches to organizing production as methods by which individual agents reduce uncertainty as to the value of various courses of productive action - CIOs : mitigating information lossiness within firms - Xerox Eureka : knowledge content of machine (about machine) was understood to be INCOMPLETE when it left the drawing board this is a motivator towards formalizing the production of knowledge based on what happens in the field, which is best done in a peer-production manner (use implies many instances, decentralized from the producer) - Eureka demonstrates that peer production can be combined with market and hierarchies, creating a 3-dimensional "productive organization" space - primary arena of information gain: information about human productive capital - thought: in the current firm paradigm, even into largely information- productive arenas, there are "HR" departments which manage the plugging-in of people (the human resources) into predefined slots. This seems increasingly odd in light of the fact that more and more human effort is creating information rather than applying simple labor to produce physical artifacts. It seems like probably most firms could benefit from adopting a peer-production model, to some extent. It just seems to me that your typical HR department is hopelessly incapable of managing things like human creativity, which as mentioned earlier, is much more variable than labor. - commons-based production must overcome losses of management, pricing, and lack of property and contract. - one of the key reasons there is a benefit to be had from commons-based peer production is that it allows individuals to self-identify their participation and their roles. - there is a comment here that strengthens what i said above: "... as human intellectual effort increases in importance as input into a given production process...", showing that this is a recognized trend - distributed information production identifies best individual to produce a component, with abilities and availability (given the time frame) auto- matically taken into account. - i would say there is also a social/situational cost that appears in the firm, human resource management situation. in a sense, it is the individual pitted against the firm here, in a battle of wits. the individual's goal is to fool the firm into thinking they are 1) hireable and 2) doing a good job. there is ample leeway for expenditure along these lines, which isn't helping the primary goal of production, yet still results in a less-than-optimal person doing a particular job. i think this goes a little further than just being imperfect information, because we actually have the employee acting towards a distinctly different goal (simply that of staying hired) than the goal of production. there is a threshold, of course, where the employee becomes obviously detrimental or does not meet some fixed requirements. but i think what happens in firm-based information production is that individuals are consciously not working optimally simply because they are not motivated internally towards the goal, yet they do the minimum (possibly with some nonconfrontational cushioning within the corporate structure) and remain hired and paid. this situation is *impossible* for commons-based peer production. in fact, i surmise that this is one of the big reasons commons based production is feasible, because there is no massive influx of opportunists whose goals differ from the goals of the project (of course this is possible, but nowhere near the extent we are acquainted with in firms) - peer production can used its own decentralization and technological basis to provide fine-grained quality control - sometimes not even expending the effort to actively bring the "best" agent to a role on a project that suits them is more productive, since the next-best person pays for their own lesser performance in the fact that they were "free" to acquire. - the picture of firms merging (as a vehicle to illustrate the peer production system's benefits) is interesting. however, firms merging should really not be equated with commons-based peer production among the set of people the firms employ, as this would kind of defeat the whole point of the paper. when firms merge, there is no magical fluidity of reassignment of resources and tasks, so the situation is still much different than the commons. - good insight about entirely new projects being cooked up, given access to the same resources - see http://www.reed.com/Papers/GFN/reedslaw.html - there is a footnote about reed's law, and it sort of teases that something similar would be applicable to commons-based peer production. it does seem to me that if the set of resources is of size m and the set of agents is of size n, then the limit of value attainable by the commons-based paradigm will be 2^(m+n). the problem is that this is the same upper limit as the case where all agents and resources are bound up in firms, since firms have the capability to shift people and resources, albeit with transaction costs. Hence, to actually quantify the difference, one would need to insert some constants in the equation that, in the case of firms, reduce the number of combinations available. - nice comment about the size of the service-based software industry as compared to the whole software industry (2/3), and how this provides a large pool of potential free software developers. it does, however, neglect the fact that someone can develop free software AND shrink-wrap software at the same time. that is, software developers are not completely "rival" either. - good job on demonstrating that motivations are sundry and can combine - Take a look at : Richard M. Titmuss, The Gift Relationship, From Human Blood to Social Policy - use of s and p in exploring the effect of depressing monitary rewards and correlating monetary rewards and socio-psychological awards is kind of confusing. - jalt factor as applied to free software community is interesting, i think it does factor in (who wasn't jealous of Eric Raymond when he was suddenly a millionaire overnight?) - interesting point about teenagers and young adults, who have plenty of time left to earn money, but high needs for social recognition, being good candidates for peer production participation - corollary to linus's law: given a sufficiently large pool of potential contributors, the direct monetary incentives necessary to realize an information product are trivial - new law: peer production is limited not by the total cost or complexity of a project, but by its modularity, granularity, and the cost of integration. - nice breakdown of true commons versus common property regimes, and regulated versus unregulated commons - good point about how growing free riding does not matter; it is the availability of some minimum number of contributors that matter. i would say that this applies in the linux case, as i'm sure the number of users has grown faster than the number of contributors, though both have grown. - sometimes increased use by "free riders" actually helps the contributors (greater recognition, more opportunities, etc) - need for integration not always present; for example, putting up a web page on a topic (actually, this isn't true: integration in this case is done automatically by search engines...) - appropriation: one party takes a makes their part of a project it too specific to their values or goals (example: behavior in MOOs). * Can be regulated by social norms. * Can be regulated by technical means; keep each person from having too much of a role in shaping the direction of the project (slashdot mod.) - jealousy factor: people don't want their efforts used for commercialization * AOL moderators who left * this doesn't always apply: look at linux development, some got very rich and some did not, yet development continues and is not in jeopardy - on peer review journals: yes, the academic contributes for free (or even pays) and their own institution buys back the finished work (at perhaps high rates), but at least ostensibly, the publisher is assembling, editing, polishing, distributing, etc. So one can at least claim some value is being added. Of course, it is becoming increasingly obvious that brick-and-mortar publishing of academic works is an inferior system to online, peer production of equivalent works (or perhaps even not assembling them into works, as this is perhaps an artifact of constraints of physical distribution) - real problem of appropriation is a scarcity of /motivation/ to contribute, not limited access to product - integration can be automated and peer-produced in the same way the content creation is itself peer-produced i would argue that the lines are increasingly being blurred, so it is making less sense to even consider integration as a separate task. for instance, in planetmath, some of the integration is delegated to the content author, and scoped towards their own contribution's size, much is achieved automatically by the system, and some is done by other users (with a tiny portion done by admins) most of the work, however, is done automatically or by the user themselves at the time of contribution, counted as a part of contribution. on planetmath this is the provision of classification metadata and semantic metadata (synonyms, defines, proper titles, related links to other entries). so here there really is no separation between content authoring and content integration. - interesting idea about property regime or partnership for cooperative using the product of some commons-based product. - it is hard to separate out technological provisions from social norms in peer-production environments. for instance, kuro5hin uses social norms, but these have a technological basis in the publishing system. it seems that a key innovation is using the technology to embody social norms. - good point in the conclusion about firms needing to commit to /not/ appropriating the fruits of peer production, as this dissuades the production - very very good point in warnings to regulators about strengthening intellectual property laws - i agree that the institution of copyrights in peer production projects is mostly a defensive move, to defend FROM intellectual property (public domain is not really a good option), but things like the GPL do go a little bit further into their own philosophies (producing the "viral" nature, which is distinctly unfree and ventures back into the realm of coercion by IP)