Automated Trust Mechanisms and the One World Market

1. Introduction:

In his famous essays, "The Nature of the Firm", from 1937 [RHC 1988], and "The Problem of Social Cost", written in 1960, [RHC 1988]the economist and Nobel laureate Ronald Coase investigated the balance between the firm (command mechanism) and the market (price mechanisms). Assuming that price mechanisms are the most efficient method of conducting business, Coase postulated that in a zero-transaction-cost environment, the firm would be outperformed by traders operating in open markets.

Though Coase, in questioning the rationale for doing business by decree rather than negotiation, does not speculate on the scope and extent of markets themselves, it follows that such limits must also be affected by transaction costs ¹. The thought arises, that if increasing the size and scope of a market were to bring about no marginal increase in costs then a market could expand to eventually encompass all economic activity.

2. Iron curtains, walled gardens and irrational behavior

The argument against Coase is that "everything can not be measured in costs" and that his (and consequently my) hypothesis assumes a rational human/market behavior that simply does not exist. I agree inasmuch as that which is immeasurable or unpredictable could influence the size and scope of anything. If it can be shown that these immeasurable and unpredictable forces are moving consistently in any particular direction, then they must of course be taken into account ².

Be it through rational behavior or not, economic activity can be walled-off and constrained by arbitrary decisions supplanting price mechanisms. Market limits may be determined by political decree, monopolization, oligopolization, oligarchization, trade protectionism, or panopolies promoting their self-interest. These iron curtains and walled gardensare by nature neither contractive or expansive, but wherever they occur, the effect of transaction costs on the size and scope of markets will be distorted.

3. What determines the size and scope of markets?

Historically, markets have walls and limited space inside these walls. Walls might actually be of brick and mortar or, as intended here, symbols for the perimeters of homogeneous economic activity. Each market participant strives for a maximization of return on investment inside these walls and normally, actors will attempt to identify one or more profit-centerswhere the ratio of income to cost is at its zenith. As transactions move outwardly from this center, it is expected that the income/cost ratio will diminish and that at some point along the radius turn negative, thereby establishing a perimeter beyond which, it might be foolhardy to do business. Unfortunately for the entrepreneur, this is seldom a linear calculation, as simple as, say measuring sales income against transport costs, when sending out delivery trucks further and further from a factory. The ratio may vary bumpily along the path of the radius.

3.1. Proximities (demand-side)

Distances between the profit center and the perimeter can be expressed as proximities, of which temporal-spatial proximities have historically had the most decisive effect on the size and scope of markets. We bunch together the terms of time and space, since they are often co-dependent. In many cases it is not the geometric distance between two trading partners that matters but rather the time it takes to move things back and forth.

Other proximities influencing market limits are those of culture, language, politics, and ethnicity. And ever increasingly, as the ability to pinpoint them grows, the qualitative distinctions of age, sex, income bracket, social status, etc., often sorted under the category of demographics.

3.2. Specificities (supply side)

Each asset specificity defined by Williamson [WO 85]corresponds to possible gains through the concentration of resources and therefore influences the limits of markets, as well as firms, where such concentrations are thought to lower cost or increase profit ³. There is a temporal aspect to specificities as well — maintaining continuity in concentrations over time.

Asset specificities

• Physical asset specificity

  Similar products and similar tools of production

• Human asset specificity

  Similar human experience and expertise - similar solutions and goals

• Dedicated asset specificity

  Similar investments and investment strategies

Specificities imply the benefit of concentration and focus. They are contractive forces.

To put the proximities and specificities into context, imagine the Swedish labor market, made up of all possible jobs and all possible workers in Sweden. It is easy to see how the proximities of time, distance, culture and language shape this market, (within the walled garden of Sweden of course). It is also easy to see how asset specificities divide this market into sectors such as the power and utilities market or the shoe market.

Proximities lead to further segmentation at a subnational level; for example to one particular region of the country, such as Norrland, or into sectors of permanent and temporary employment. Specificities give rise to vertical market segments for the employers and employees of the forestry sector or the academic sector. Or to horizontal markets for secretaries, computer technicians and executives. These constellations are obvious to any observer— we all know the benefits of having the unessential and irrelevant sorted out before getting down to the specific task at hand. But these markets are migrating to the Internet, and this raises questions as to how one best defines and exploits proximities and specificities using new technology.

4. The causes of expansion

"Modern natural science regulates the direction of economic development by establishing a constantly changing horizon of production possibilities." "Specialized tasks which were unprofitable when a factory was selling to a couple of local villages suddenly become worthwhile when one sells to an entire nation or to an even broader international market." Francis Fukuyama The End Of History And The Last Man [FF92]

What makes markets expand? I would suspect that markets themselves are inherently organisms of expansion. I assume that the very reasons that bring providers and consumers together - the search for more and better product at a better price - are sufficient to warrant market expansion up to the point where the costs of expansion become prohibitive. This does not automatically mean bigger is better. If smaller markets translated into greater profits, then markets would stay small, but the reasons for keeping them small, if we disregard than the distortions of walled-gardens and iron-curtains, appears to be the costs of making them larger.

The proximal relationships between buyer and seller are constantly being augmented by technology. Transportation technology in particular has served to radically extend market range. Today even the most unlikely and cumbersome of goods can seemingly be flown, shipped and driven between any two points on the globe and still sold at their destination for attractive prices. This development goes hand in hand with financial technologies that expedite clearance and settlement in all viable currencies, and trading technologies that can time-shift transactions to enable effective risk management.

Mass media and broadcasting technologies have significantly extended marketing reach, and consumer profiling technology has extended the proximal reach of demographics. In the last decade, the rapid advancement of communication technology, in networking and above all the Internet has given us the means of radically reducing transaction costs. And as we know, the Internet offers new solutions to the concentration of asset specificities, through narrowcasting and distributed collaboration. Needless to say these potentials have not yet been fully exploited.

Transaction costs are no more deadthan distance [GF99] but automated transactions taking place completely within the realm of one computer or one computer network, have marginal costs, so low in some cases, that one could justifiably call them quasi-zero or approaching-zero. Of course all aspects of transactions are not wholly situated in the computer realm, and some never will be. The extent to which the individual transactional components can be computerized and networked is significant both for the total transaction costs of an interaction and inversely the future survival of those particular non-computable, non-networkable components.

5. Semantic Web - Semantic Trust

This paper is being given at a conference concerned with the interoperability of information. Our method is to enable a flow path between the universal and the atomic by the mechanism of referential extensibility. Our vehicle is the Internet. At this time there is only one Internet, even if that just might be an accident. There are of course other long distance networks, hermetically isolated in search of security or proprietary expedience and the Internet itself harbors tunnels of virtually isolated nodes and networks via schemes of encryption, but it is the oneness of the Internet above all else that has brought us here. The network is the computer. The Internet is the network. One network - one computer.

It is the integral singularness of this technology that gives it supremacy over preceding technologies.

Machine readable information is a natural evolution of the codification process that precedes the interchange of utilitarian ideas and values in science, engineering, business and law. Codification, or representation in domain specific controlled languages, classes and types, considerably predates computers. These are trust structures, since they imply an (out of channel) object of reference necessary to a trustworthy exchange of material or immaterial assets. In some cases these trusted references are physical objects (ex. A bar of platinum-iridium), sometimes they are based on empirical observation of natural phenomena (ex. The movements of heavenly bodies), and some times they are the representations of humanly derived and dynamic values (ex. The price of pork bellies six months from today).

I won't dwell on the truth aspects of these referents, but it is clear that the modeling of referential systems can not be uniform for all forms of knowledge - that some forms lend themselves happily to interoperability in mathematics, calculus, and statistical comparison — others not. ⁴.

Poetry, for example does not lend itself to interoperability in mathematics, calculus, and statistical comparison; The referent for poetry is the entirety of language and language is ambiguous ⁵, but science does lend itself, and so to a lesser extent, does commerce, and to a lesser extent still - law. Law as we know, is a conglomeration of science, business and poetry.

Referents may be in private hands, and referential frameworks may be closed, but reliable, efficient and accurate access to referents within any framework, be it the family, the firm or the world, is the foundation of trust. To portray the Semantic Web as a web of meaning is misleading. But as a web of trust, as the universal channel for World–3 ⁶ reference, the Semantic Web represents an historical paradigm shift.

"Without a technology tending to process experience homogeneously, a society cannot go far toward the control of natural forces nor even the organization of human effort" Marshall McCluhan— The Gutenberg Galaxy

6. Metcalfe's law

I assume that most of us here are familiar with Metcalfe's Law, named after Robert Metcalfe, the inventor of Ethernet. Roughly stated— as each new node is added, the value of the network increases exponentially. According to this formula, one network of 100 nodes is not just equally as valuable as five networks of 20 nodes, but five times more valuable!

As stated, this law in itself would provide a powerful argument for extending markets - the larger the market connected through networks, the greater the value. But the law seemingly takes for granted that each node in the network actually has something to give or take from all other nodes. A network that was maintained but never used, would only increase in operating costs as it grew.

I often think of Metcalfe's law when waiting for my bags at airports. If the law were to be implemented in arrival halls, we would not have the separate luggage ramps of today. It would be better to put the baggage of all incoming flights on one belt, inviting incoming passengers to participate in a considerably more valuable network. This would not be a bad idea if every passenger had a potential interest in every item on the ramp, but such is not the case. Each passenger (hopefully) is interested only in his or her own luggage. From the passenger's point of view it might be more convenient to give each of us our own private ramp, but since this would put undue strain on airports, we have arrived at the present compromise ⁷.

This sort of compromise is going on wherever we turn; the accommodation of physical properties to abstract efficient systems or, the accommodation of abstract efficient systems to physical properties.

We are aware, as many writers have pointed out, that there is a great difference between that which can be digitalized and that which cannot. You don't have to be digital to take advantage of Metcalfe's Law, as anyone attending a good party can ascertain, but it does have its advantages in communication networks. In communications networks, no matter what we originally use them for, there is always the possibility of using them for something else — as long as that something else is carried out in a nondisruptive manner to the network as a whole ⁸. The cost of doing something else is directly related to that something else's ability to exploit what we are already doing.

An example: Exploiting commonalities with SkiCal

In Sweden, we are gradually incorporating the SkiCal http:/www.skical.orgontology for Super-Directories ⁹. We are exploiting the commonalties of resource directory publishing, the fact that so much information, such as names and addresses, scheduled happenings, hours of business, payment methods, transportation and parking facilities, and what have you, are common to all events, services and businesses. If successful, we will (blush) have the most effective marketplace in the world for helping people decide on how to spend their time. Please observe that we are talking about machine-readable information objects here, not particular applications or portals.

SkiCal information is made available on the Internet, which is of course a global network. In a Metcalfian sense, each additional Swedish dentist office publishing her telephone consultation times; the location of each new public toilet in Stockholm; or the occurrence of a happy hour in a bar in Kiruna, will exponentially increase the value of the Internet to a potential TimeSpender ¹⁰In Rarotonga, on the other side of the planet, which strikes us as absurd.

On the Internet the profusion of localized information is seen as a cost to all those who couldn't give a hoot about things they will never interact with. A Cook Islander looking for a dentist on the Internet, not being able to distinguish between local and Swedish dentists would only loose time and run up her ISP bill ¹¹.

There are two solutions; cut off the wires or get structured. If we opt for the former, we will loose all the collective benefits of being part of a network as related to things which are of value to Cook Islanders. The later choice entails creating structures that reduce the costs for a Cook Islander to find that information which she values. Since this involves a considerable amount of work, someone thinking in the short-term, might just say - cut the wires.

7. The cost of a transaction is the cost of trust

I spoke above of "one network - one computer". I will now extend that to "one network - one computer - one market". In a world of perfect trust there would be only one market.

This might seem incongruous, since the arguments for firms contra markets cited above, which I have related to the size and scope of markets themselves, have to do with transaction costs. So what has trust got to do with it? Coarsely stated; trust is the transaction cost.

In protest, you might ask, 'If I place a telephone call to Kuala Lumpur to buy some tiger tooth elixir, is the price of that telephone call trust? Are the shipping costs to Berlin the costs of trust? Is the commission on my credit card payment the cost of trust?'

The answer to all three questions is yes. Transaction costs are trust costs. The cost of a transaction is the cost of trust. For what else would you be paying for, if not the assurance that your transactions will transact in the manner you expect? If you did not expect reliable performance, then hopefully you would pay accordingly. You might not expect any performance at all, in which case you would pay nothing.

Edgardo Gerck, in his paper "Toward Real-World Models of Trust" 1998-2001 [GER], presents a formal and abstract definition of trust: "Trust is that which is essential to a communication channel" and qualifies that with " but cannot be transferred from a source to a destination using that channel ."

Gerck's paper is not an easy read. He is attempting to create a unified theory of communication by the coupling of Shannon's mathematical theory of information [CES48] and his own theory of trust: "Trust and information are to be understood thus as two cardinal properties of communication systems..." The combination, loosely speaking, comes out like this: "Information is what you do not expect (Shannon) - and - trust is what you know. (Gerck)"

This seems a far cry from the capricious adventures of Alice, Bob and Carol common in the literature of trust ¹², though it really isn't. Though in the public eye, the issue of trust is often focused on financial transactions and privacy, in this paper we are more concerned with trust as that which is essential to a communication channel and to what extent transaction costs determine trust and vice versa.

I believe Gerck makes a significant contribution towards understanding the role of trust in the system itself. And it is in the makeup of the system; the market, society, the computer, the Internet, that the role of trust should be understood by the would-be architects of such systems.

8. In the grand scheme of things

Von Mises in his study; Human Action, 1949 [VM 49]writes:

"The market economy is the social system of the division of labor under private ownership of the means of production. Everybody acts on his own behalf; but everybody's actions aim at the satisfaction of other people's needs as well as the satisfaction of his own. Everybody in acting serves his fellow citizens. Everybody is both a means and an end in himself, an ultimate end for himself and a means for other people in their endeavors to attain their own ends."

In the world of Von Mises, and if you will, his opponent Karl Marx ¹³, it is the system which fosters trust through its ability to promote the optimal behavior of man, be it laissez-faire or proletarian dictatorship ¹⁴.

Though that which constitutes an ideal system belongs to the realm of essentially contested concepts, it is apparent that in matters of trust, harvesting the optimal behavior of man (and machines) is a worthwhile goal. For it certainly must be cheaper to promote trustworthy actions within a trustworthy system than to do so in an untrustworthy one. Therefore it is essential in the design of a system to ask to what extent does trust profit within this system? And this is true both for the systems of computers and men and anything there in between ¹⁵.

9. Mechanisms of Trust

Arguably, eye-to-eye, or handshake trust is the original trust between people. Handshake trust has short range and when not formalized in writing, it is prone to arbitrary interpretation and dispute. Mechanized trust, in the form of tools such as locks and balance scales addressed some of these problems. But at some historical point, in order to accommodate the greater range (extension of proximities) man was striving for, the need for signed trust manifested itself resulting in the abstraction, codification and signing of laws, constitutions, contracts, treaties, testaments, and their like.

As society developed in complexity and as communication technology and mass marketing prevailed, branded trust became necessary in order to compensate for our limited capacity to keep track of and reckon with the many disparate fragments of trust. Branded trust lets us conceptualize past disparate fragments to a concerted, manageable whole. ¹⁶Branding fills in the gaps, acting as a halfway station between superstition and first hand knowledge. And this applies to our transactions as well as our membership in polities, religions and football supporter clubs.

Have we stumbled upon another transaction cost problem? Can technology, by reducing the cost of access to empirical and certified "facts" through the Internet, threaten the branded trust model? Can a technology replace psychological phenomena so congenial to humans. To some extent I believe so, but to what extent remains to be seen.

We are in the midst of the age of branded trust, as the merchants of network-mechanized trust begin to set up shop. Tim Berners-Lee in addressing XML2000 described the "Semantic test ..., which is passed if, when you give data to a machine, it will do the right thing with it". ¹⁷

So, I don't mean to suggest that networked-mechanized trust or Semantic trust will totally subsume or replace branded trust. Branded trust did not replace handshake trust or signed trust. Only power or singularity can replace trust. Trust will continue to coexist in many forms. Furthermore, it is obvious that we humans like "brands" and "branding". We like living in branded environments and shopping in branded marketplaces and doing branded things. Brands reinforce — or even supply our own identities. Brands provide us with entertainment, security, status, and a sense of community belonging ¹⁸. At the same time branding also serves as a tool for simplifying choice and trust ¹⁹. We neither can, nor will be bothered, to flush out every detail that can be known about a trading partner, information source, commodity or transaction.

This dual usage of branding as a lifestyle enhancer and a decision maker makes it difficult to predict our future willingness to accept Semantic trust. I find the problem rather overwhelming. Branded trust is not founded on thin air. Branded trust normally has some relation to personal experiences, empirical findings and believed-to-be-true facts, but we willingly combine all this with a considerable amount of obvious fiction. Does Semantic trust seem all too impersonal and drab? I can hear someone saying; "You can spout facts and figures until you are blue in the face, and you can program your little bots to do whatever they please, but I am going for the Gucci", or when branding is the rallying point for loyalty; "No matter what, I am and always will be a Norwegian".

9.1. Shopping for shoes

Lets say you wanted to buy some shoes by post order and the choices were products from Nike, Reebok and Adidas. You have no previous "handshake" experience with any of these companies, but you have been exposed to the advertising campaigns of all three.

You have some notions about the shoe business, leading you to believe that the shoes of most manufacturers are probably made in similar factories somewhere in China, most likely by children, who perhaps should be in school. If in your catalogue you find an equally attractive pair from each company, at the same price for all, how do you choose?

Lets say that the year is 2006 and there are 36 trust mechanisms in place to help you in your shoe purchasing decision. These mechanisms are not flashy web pages or portals; you will not be expected to browse about for the information provided. Your buying agent will do that for you while you are asleep in bed.

These trust mechanisms range from indicators as to the financial health of each of the three firms, test results as to the strength and durability of materials used, the ethical track records of each company, their compliance with environmental— and labor, guidelines and laws (national, international and NGO's), their post-transaction service performance, and perhaps most importantly for your feet; personalized profile matches between your size, which by now has been expanded from one simple ordinal number to a fifteen fold measurement system encompassing all aspects of podiatry; from the height of your arches to the ballistics of your gait. Will these mechanisms as accessed and weighted by your shopping agent decide which shoe you buy — or will you consciously or subconsciously pick your footwear on the strength of its brand and disregard all or most of the above.

9.2. You choose

What if you are the keeper of a family, and you want the best nutrition, the most reliable and safest means of housing, transportation and medical care; the most durable, yet comfortable, yet stylish clothes you can afford? And suppose that Semantic Trust is a functioning reality. Do you go for the Cheerios or do you go for the certified nutrition, certified crunch, certified hygienic packaging, and certified Robert Parker seal-of-great-taste score of 97.5, all for a certified price 30% cheaper than the Cheerios?

If you are the keeper of a firm, and you want the best workers, the most reliable machines, the best transport and communication, the most accurate tools and the finest raw materials? Do you go for AT&T, Wintel, FedEx, and Manpower or do go for the certified yet perhaps anonymous products, services and human resources chosen for you by 369 certified and perhaps anonymous agents of semantic trust? ²⁰

I don't know the answer to these questions. We don't have such automated trust mechanisms at our disposal. Today we could spend a considerable amount of time and find out at least some of the information exemplified above, but this will not be a concerted automated process, we will lack a system of common references, and our findings will not be conclusive. In most cases we will not consider it worth the time and effort.

Today price comparison can be automated, but not much else. We are not all happy about this, since low prices can be obtained at the expense of service and quality. This focus on price is considered by many a detriment to innovation. Price is an easy metric — performance, reliability and QOS measurements are harder to come by.

One method of obtaining metrics is through binary certification. A company might undergo an extensive set of quality measurements in order to receive an ISO9001 certificate. Either you have one - or your don't. Sort of like; either you are a doctor (you have a degree from an institution recognized to issue degrees) or you are not. Such a binary certificate is the sum of a slew of various test scores and performance records. But just as individual plays in a sporting game loose significance in relation to who won the game, these intermittent measurements are subsumed by the top level binary certificate. Networked-mechanized-trust offers us access to these granular fragments as well as the consummate whole.

How does it strike you that in some factory where female workers outnumber male workers by 4 to 1, there are less toilettes for woman then for men? Are you deeply concerned, sympathetic, or nonplused? Perhaps your reaction is proximity related: "Will I or anyone I know ever work there? How far away is this factory? Did you say women were being discriminated against? Are they Europeans?" When it comes to making purchase decisions this information might rate fairly low — an aleatory circumstance. But not necessarily.

Reebok thinks they can sell more gear if they alleviate this situation at their supplier's factories and they want to inject this "we care" position into their brand image. They are also looking for ways to verify this stance through certification from third parties. One current practice to signify compliance with standards requirements or codes of conduct is through logo or icontrust, where a trademark such as TrustUK is placed on a web site for human observation. Even if the advocates of icon trust have not grasped that the visual space for accommodating the multitude of icons, each representing some measure of trust could take dozens of web pages, this development can be seen as preliminary to making icons machine-readable and machine-accountable. One day you might just contribute to better working conditions in places you have no idea existed, merely by turning up the ethical concern knob on your shopping bot control panel.

The process of quantifying large performance samples into binary states of true and false or numerical scales always presents problems. Some values and nuances of value will be lost in the process. Neither can any such process of reduction be considered neutral. By definition, quantification and the compression of information promotes salience and represses redundancy, but this is carried out aspectualy, based on someone's — or something's value system. ²¹

Before you can make value judgements you must be in agreement about the metrics of value. And these metrics must be accessible to all parties. Though it comes as a rude shock to many who have initiated themselves in standard's work via the IETF, the W3C or open-source communities, most standards are not freely available and for that matter not free. The vast collection of existing standards considerably predates the Internet, and are subscription (or publication) financed. The success of TCP/IP — and the example set by the IETF standards process show that open standards are the necessary infrastructure of interoperability, but the migration of subscription financed standards into this realm is not only contended by some, but costly. Someone has to pay the bill.

In Sweden the opponents of the end-to-end labor market argue that unified schemas, such as those proposed by the HR-XMLhttp://www.hr-xml.org/consortium, and automated searches based on such schemas, would result in a significant loss of human expertise. They argue that the human resource industry is a face-to-face business. The face-to-face character of HR we can all agree on; but if you had to choose between picking 10 names out of a random bin of 100,000 possible job candidates in order to conduct four hours of interviews with each — or having that random bin first narrowed down to say, 50 job-seekers through schema based profile matching, which method would you choose?

10. Roles

If we do have 36 agents working on a shoe purchase as in the example above, how do we know that each is trustworthy? Through third party Certification Authorities? Okay, but why would we choose one CA over another. Would we rely on the strength of brands like Identrus, Consumer Reports, Human Rights Watch, Greenpeace, Underwriters Laboratory, SCS, SEMKO and so forth, or would we put our trust in the brands of CA's that keep these organizations honest, or the brands of CA's that keep those CA's honest? At some point this hierarchy ends up in government. Is there anybody out there who would like to work out the implications of government through semantic-trust as opposed to government through branded trust? Do you have a publisher?

Bertrand Meyer [BM97] describes Design by Contract as; "viewing the relationship between a class and its clients as a formal agreement, expressing each party's rights and obligations. Only through such a precise definition of every module's claims and responsibilities can we hope to attain a significant degree of trust in large software systems."

Meyer does not discuss a party's motivesfor entering into such an agreement. And why should he? Objects are not supposed to have motives, but rather functions. Functions might have roles and perform routines but in a confined world, objects should function deterministically. They can not exercise free will or possess motives.

Of course their creator had a motive. She might have wished to express a point of view through an object of information or she might have wished to carry out a task through an object of action.

In a world with only one creator — a hermetically sealed system containing objects lacking free will, collisions of intent must be the results of (faulty?) programming. But in a larger world comprised of the efforts of more than one creator, objects will be instilled with intents and purposes derived from the will and motives of their creator and collisions will occur as the consequences of these opposing wills and motives.

An informational object might say "France won the World Cup" and in the same world another informational object might say, "Spain won the world Cup". Either both objects are wrong or just one of them is right. They might be right or wrong by accident and they might be right or wrong by intent.

An object of instruction might say "Everyone who believes France won the world Cup — turn left at the next exit" and another instructional object might say "Everyone who believes France won the world Cup — turn right at the next exit". Here neither object is "wrong", they just want those who believe that France won the world cup to do different things. These objects have inherited from their creators a motive for saying what they are saying or telling you to do certain things. In order to trust that they are telling the truth or instructing you to do the right thing, you would be well advised to have an (out of channel) understanding of the roles these objects are playing. Can the portrayal of roles be incorporated into Semantic Trust?

10.1. Roles People play

• In a home appliance store, a salesperson conscientiously aids you in your purchase, but turns in to Dracula at the cash register, trying to sell you a ridiculous insurance policy you have never heard of. Unknown to you, she has switched roles. She is no longer just working for the store; she is working for the Caribbean Cruise prize in an insurance sales competition.

• You are watching a track and field event with a friend. In the 5,000 meter race there is a runner who holds the record for this distance, knowing this, you propose a bet with your friend that this runner is going to win. Taken, says your friend, for she knows more than you. Your favorite, having only entered the race in the role of a pacer, to help the others to a better time, will never cross the finish line.

• A seemingly trustworthy company, with financial stability, organizational quality, excellence of product and so forth, enters into a transaction with only the intention of jacking up the price in order to benefit an ally.

• The concierge of a hotel recommends restaurants in the area on the strength of a kickback on your dinner check.

• A supplier offers below cost software, planning to recoup this loss on expensive service and consulting.

• A cable company offers you free broadband Internet service, expecting to regain this cost by locking you in with no-choice product offerings.

• Your on-line food market is a time saving convenience and delivers on schedule, but the worker who fills your grocery list has the ancillary roleof maintaining a first-in-first-out inventory flow for her employer. Any one shopping for themselves would normally opt for last-in-first-out in order to get the freshest produce available.

Behavior in the above examples ranges from criminal to conventional. In all cases, information is hidden from one or more transacting parties. In some, such as the foot race, a discerning punter could easily discover the true state of affairs; in others opaqueness is strategic and rigidly maintained.

The ability to determine roles in human interaction is part of our survival kit. Life is easier to deal with if we think we know why people are acting in a certain fashion. We can understand why love, hate, jealousy, envy, greed, loyalty, conceit and any other human emotion would motivate people to do things differently than when they felt no such emotion, and we can anticipate actions and outcomes if we know that people are in particular emotional states or acting on particular motives.

In business we can expect that all participants are in the state of being "greedy" and this tremendously simplifies our dealings. "Greedy" doesn't sound very nice but it's a lot shorter than Von Mises's quote to the same effect in this paper. And it constitutes acceptable behavior in business, preferable to companies doing things because they are in "love" or because they "hate". In markets greed is taken for granted.

A lot of people are obviously participating in the development of the Internet driven by other motives than greed. Be it the desire for thymos— recognition, self-esteem or downright altruism. Actually these roles, though perplexing for many businessmen of more traditional kilt, are often made somewhat clear though open source licences. But involvement in the open source movement is unfortunately not itself a guarantee for trustworthiness.

10.2. The Concierge Problem

In the above example of concierge kickback, an individual's moral character (though perhaps the moral character of society) are not the issue. The system is the issue; Concierges are usually knowledgeable, socially competent and of no lesser moral resolve than any of the rest of us. But hotel management pays them poorly, knowing full well that there are external compensations to be had. Ostensibly these compensations are gratuities from appreciative guests, but in reality concierge are tipped poorly. The remaining course of action for a Concierge is to accept referral "awards"and trust themselves to act conscientiously despite them.

This system could be changed. But someone would have to pay for those changes. Much of the concierge's rewards are in kind, and this gives them the opportunity to try out restaurants and events, giving experience to share with hotel guests. If the hotel management or the guests were to pay for both higher salaries and on-site experience, concierge services would become quite costly.

One solution would be to make all concierge kickbacks transparent to hotel guests, (Politicians in some countries must do something similar) but such a system strikes us instinctively as whacky.

The question that begs an answer here is; can we really create value and trust systems in cyberworld that do not directly parallel those of the physical world or is this just Jolt Cola utopian dreaming?

Perhaps this is another transaction cost issue? Perhaps through the economy of scale of networked-mechanized trust, conditions can be set to solve the Concierge problem. But this might entail eliminating personalization - or the concierge herself.

11.

11.1. Thou shalt only serve one master

We might think of a future, whereas legions of agents are lambently cavorting about on the Internet doing our chores for us. It would seem perspicacious to program these digital servants to obey some sort of code of conduct, such as the robot laws proposed by Asimov ²². Unfortunately Asimov left out one key rule; "a robot shall serve but one master". Nor has this rule been embraced by today's robot builders. Most Internet agents are undoubtedly double agents — serving several masters.

Buying agents, for example, frequently accept kickbacks. You might think they are figuring out the best deals, no doubt they are, but those are the best deals to line two pockets, yours and those of the particular sellers willing to make the kickback. This problem is simple enough; we don't pay our agents a decent salary. We force them to moonlight.

Everyone will tell you there is no free lunch, even when eating one. The free lunch culture proliferating on the Internet is problematic for trust since it fosters opaqueness and muddles roles. Sellers should have no back-door influence on buying agents — and selling agents should not be on the take from buyers, each should be instilled only with the will and strategies of their masters. Once these roles are guaranteed and only then, can these two heteronomous agents make a "fair deal".

The failure to establish viable digital payment systems is a serious concern, a significant impediment to effective trust. Today interruptive advertising in the form of banners fills the absence of mechanisms for petty cash expenditures, but banners have nothing to offer bots and semantic trust can hardly be financed by advertising.

Of course selling bots can offer buying bots deals and special prices and limited offers and buying bots can make bids and offers of their own. Dynamic pricing is something electronic agents can handle quite nicely.

12. End-to-end as an instrument of trust

In papers and talks by members of the Information & Communication Technology (ICT) community, it is quite common to hear terms of system architecture straying outside their original context. The same terms can be used in talking about computers, networks, communication systems in general, and human society. Two such terms in vogue are "layers" and "end-to-end".

The concept of end-to-end (or the buzzier end2end ²³), was formalized in a paper by J.H. Saltzer, D.P. Reed and D.D. Clark, "End-to-end Arguments in System Design" 1981 [SRC81] "a design principle that helps guide placement of functions among the modules of a distributed computer system..."In their conclusion the authors write: "It is fashionable these days to talk about 'layered' communication protocols, but without clearly defined criteria for assigning functions to layers. Such layerings are desirable to enhance modularity. End-to-end arguments may be viewed as part of a set of rational principles for organizing such layered systems."

Both layered and end-to-end system architecture have to do with the division of labor. Both have to do with the strategic allocation of resources within a system.

Threefold this is:

• Maximizing efficiency within each layer

• Optimizing communication between layers

• Maximizing autonomy, plasticity, flexibility and transparency within each layer

I will concentrate on number three. By the division of labor into layers, it is assumed that each layer has more freedom to go about its own business without deliberating with the other layers. As long as each layer meets the requirements and constraints involved in passing on products and services to the layers above and below, it can freely redesign itself as it pleases, and it may also expand horizontally as opportunity arises, leading to innovation and growth.

We should keep in mind that stacked layers are not objects- in the sense that objects could freely choose to interact with whom or whatever they wished. Layers are traditionally confined to their purpose and utility, and the rules of communication between them are so construed. Basically they haven't a clue as how to address anyone else than their immediate neighbors. The original end-to-end paper ²⁴ does not speak of informational objects in the sense that Graham Klyne writes about below. One might say it is application-centric.

In the physical world ²⁵ We also stratify economic activity into horizontal layers such as research, exploration, production, distribution, wholesaling and retailing.

In "A Layered Approach to Information Modeling and Interoperability on the Web" [MD00], Sergey Melnik and Stefan Decker write: "... we suggest a layered approach to interoperability of information models that borrows from layered software structuring techniques used in today's internetworking. We identify the object layer that fills the gap between the syntax and semantic layers and examine it in detail."

Melnik and Decker are moving layering one step up the food chain; they are suggesting how to reach interoperability between semiotic layers, though they make no reference in their paper to end-to-end.

"It is a view of many in the Internet community, and ours as well, that the extraordinary growth of the Internet rests fundamentally upon its design principles ²⁶. ... but from the very beginning, these principles have been understood to have a social as well as technological significance."

Lawrence Lessig has championed the end-to-end concept in several papers. Here together with Mark A. Lemley: [LL00]

"End-to-end expands the competitive horizon, by enabling a wider variety of applications to connect and use the network. It maximizes the number of entities that can compete for the use and applications of the network. ... An end-to-end network creates a maximally competitive environment for innovation. ... Vis-á-vis transportation generally, end-to-end is also how the principle of common carriage works. The carrier is not to exercise power to discriminate in the carriage. So long as the toll is paid, it must accept the carriage that it is offered."

Lessig and Lemley do not use the word "trust" in this paper, they speak of "reliance" "that innovators can rely upon" protection from strategic behavior.

"The potential of an actor in the distributional network to act strategically is a cost to innovation. The expectation that an actor can act strategically is an expected cost to innovation. Thus to the extent an actor is structurally capable of acting strategically, the rational innovator will reckon that capacity as a cost of innovation. Compromising end-to-end will, then, tend to undermine innovation."

And finally Graham Klyne, writing to the RDF interest group mailing list [GK]:

"The semantic web aims to do for information what the Internet and www have done for data. Today, the Internet, and the www in particular, allow data to be exchanged between an arbitrary pair of systems, without regard for, or discrimination among, different system types, data content or the service to which the data relates.

Today, information tends to be bound up in the data format of the application for which it is designed. Without the right application, the information content of data is inaccessible. The semantic web will make the information content of the exchanged data accessible to any application that understands semantic web protocols. In other words end-to-end architecture for content."

Klyne is expounding, without using my term, unknowns compatible information. Informational objects prepared to meet all comers. Or as Tim Berners-Lee said at XML2000, of the semantic web, "a 'self-navigable space' of self-described, 'machine-understandable' fragments of information in which documents convey meaning through XML markup".

12.1. Evolution of Information and trust

The cited texts above focus on a brief period in the evolution of information and trust, our own age's chapter in the continually changing relationship, between Popper's three worlds. As we follow the evolution of world–3 from stone to papyrus, from the hand copying of scribes to the incunabula of early printing presses, to movable type, to electronic media and digitalization, an evolutionary pattern becomes clear. Information is "liberating" itself from its World-1 captivity.

Digital information began the 20th century as holes in Herman Hollerith's punch cards and ended it as web pages, mp3, and XML. In the fifties information had to be constrained to work reliably within machines. Until quite recently, information had to be constrained to work reliably within databases and applications. Today machines, operating systems, databases and applications have to be constrained to work reliably within information. That, I believe, is a significant evolutionary development. ²⁷Information freed from the restraints of 20th century technology, is capable of achieving astounding levels of interoperability and concurrence, but new structures (infostructures) must be built to take the place of the old.

13. The end-to-end market

An end-to-end market is one in which the endsof the market have unencumbered access to one another. It reflects the design maxims of Saltzer, Reed and Clark, inasmuch as it discourages giving power to intermediaries to encumber or deny this access. The role of intermediaries in an end-to-end market is to provide added-value services in competition with all other would-be intermediaries.

But where exactly are the ends? Do they extend from the depths of iron ore mines to the engine blocks of our automobiles or from the seeds of crops and forests to our hands and mouths? As I see it the ends are simply —ideas. Ideas of creation and production, ideas of desire and necessity.

The middle ground between these ends, the path between A and B, where— and whatever A and B might be, has always been a rocky-road-rivers-to-cross-and-mountains-to-climb sort of thing. This is where the middlemen reside. There are good-guy middlemen, who through knowledge, skill and technology offer a better, safer, speedier, more comfortable journey. And there are bad-guy middlemen, who profit by alleviating hindrance that they themselves have created. These middlemen are not necessarily human, merely that which exists between A and B. The distance between ideas. The distance between capital and manufacturing, conception and realization, buyer and seller. The distance between supply and demand.

In the original end-to-end model, the middle ground was to be populated by obedient dummies, agents of defined and restricted purpose. But that which constitutes a "middle" contrary to an "end" may not always be that easy to define. Once we leave the relative safety of layers (as in the Emerald City everything is green), things get a bit more complicated and our allegory loses some relevance. Can it be, that it is the "business model" rather than the task performed that is the distinguishing factor? Can a business model choose to be an end or a middle. Here are some models:

The Tollbooth: In this model, participants are obliged to pass a tollbooth and pay whatever toll is requested. The tollbooth just might be the only way to gain entrance to a desired category of service, in which case it constitutes a monopoly.

One can hardly argue for the elimination of tollbooths, since they are staples in our economic system, and perhaps our protection against the "tragedy of the commons". But in accordance with end-to-end philosophy, once participants have paid and passed, the tollbooth operators should not be able to exert further influence upon them. This, by the way, is the core argument of Lessig and Lemley in the MediaOne Group- AT&T Corp. case. [LL00]

The Packaged tour: In this model you can choose between different competing tour operators, but you are not able to influence the bouquet of attractions they have chosen for you. One tour might include the Louvre, yet not the Eiffel Tower, while another tour leaving at the same time might offer the opposite.

The package tour trading model also constitutes a lock-in for the end participants. The freedom of choice to choose between tour operators might put competitive pressure on their prices and services, but these competitive pressures are lessened considerably once the tourists are actually on the bus.

The Lemonade stand: Lemonade stand operators can offer thirsty travelers better or cheaper lemonade, they can shout out their barker's cries and marketing slogans, but they cannot forcefully detain anyone - no matter how much they might like to.

The lemonade stand model has a governance problem; exclusivityor more correctly; lack of exclusivity. For how is one to engage in long-term activities or make long-term investments, without ownership in- or control over the object of investment. One solution to the problem of exclusivity is granularity. Instead of granting exclusive rights to the whole of an idea, product or one's person, participants might contract only for lesser parts. Digital technology and infostructures have changed the possible extent of granularity for those parts.

14. Granularity

Granularity in this case means the breaking up ideas, processes, products and services into fragments. The makeup of a set of fragments is dynamic and aspectual to ever-changing utility and circumstance. Each set represents new and unique combinations. The backside of granularity is complexity and the costs involved in extending distributed trust. Unlike Humpty-Dumpty, a granular set must be able to function as effectively as any pre-granular whole.

Imagine a group of children in a playroom. Each child has come to the room with a thousand pieces of their privately owned and highly valued collections of Lego. A teacher says to the children. "Put all your Lego pieces in the middle of the floor and build a great city." The children balk, since once the Lego is removed from their immediate possession, they can no longer identify it as their own. The teacher tells them," I will keep track of each piece and remember whose is whose. If the children trust the teacher's ability to do this they will begin to build.

To match their trust the teacher would need almost supernatural powers since Lego is known for the precision modularity of its product, mostly indistinguishable plastic blocks. The value of playing together and having so much Lego to build with is the value of the network, but the effort to maintain trust regarding individual ownership is the transaction cost.

End-to-end markets are capable of creating a considerable amount of complexity. Through Semantic trust they will admit a swarm of participants (fragments) into one and the same transaction. The shoe shopping experience hypothesized above, could lead to the interest of thousands of agents. Shoes might even be designed and assembled on the fly by any number of subcontractors. Shoes sales might be transacted, shipped, intermittently financed, warranted and serviced by any agent or instantaneously formed consortium of agents with the necessary CA authorization.

The complexity is further exacerbated by the existence of exploratory negotiation. Agents trying to evaluate their participation in a deal would need the same transaction mechanisms as those of a firm deal. The trust mechanisms must be in place regardless.

Nor is the resource expenditure in the process of discovery trivial. Proximity concentrations alleviate the cost of discovery. Slopes in diminishing returns, if not linear, are at least somewhat predictable. If you are digging for some mineral starting from the point of a find, it is expected that concentration of the mineral will either diminish or increase as you dig deeper and wider. If concentration increases you will dig in that direction, if in decreases you will change directions. Like the child's game, you are either getting warmer or colder. But if this mineral is distributed totally at random your task will be much more difficult. You will have to look everywhere.

Despite these considerations, it is my opinion that in the Swedish labor market example sited above the participants are best served by an end-to-end market. That each potential employee, irregardless of their whereabouts, job, or relevant state of affairs, is benefitted in participating as an unhindered end in such a market. That every employer, whatever current needs they might have for human resources, is also benefitted by participation as an end in this market. And that any intermediary wishing to engage in activities, such as recruitment, temporary staffing, economic planning, development of econometric models, trend analysis, forecasting, and what have you, will do so more profitably and effectively in this end-to-end market - not as tollbooths or packaged tours, but in accordance with the lemonade stand model.

15. The One World Market

At first glance end-to-end markets appear, as phrased in popular geomorphic terminology, to be "vertical". This symbolizes that market limits follow the flow of specificities, as in our examples of shoes and jobs. The likeness of ideas and product creates economy of scale at various stages in a product's flow-path from inception to consumption.

Some times the material properties of a particular product are so unique, that one sees no reason to bunch it together with dissimilar products. An example that comes to mind is that of fish; live tropical fish and dead frozen fish. The one can never get cold and the other can never not be. They can't reasonably share warehouses, or packaging, or transport lorries or retail shelf space. And the workers involved, from fishermen to salespeople, can hardly be expected to know about both. It seems the only thing these enterprises have in common is their name — fish.

But on the other hand the transaction process for these two items on their vertical flow path, once product specificity is taken into account, are to a great extent identical. Consider my ad hoc, non-exhaustive list of transactions for both living and dead fish;

The marketing of goods and services; the discovery of goods, services, trading partners and agents of intermediation; representations of intentions and capabilities; exploratory, committal and post committal negotiations; contracts, and the declaration of transaction-specific assets and liabilities; declaration of compliance with laws and treaties, authorizations, certifications, authentication, escrow, bank guarantees; administration of transportation, insurances, asset transfers, settlement and clearance; post-transaction liability and dispute resolution, warranties and service agreements.

For each particular transaction element, there is a ratio between the possible exploitation of commonalities and the exploitation of exclusivities, or as termed in jurisprudence; universalities and particulars.

Digitalization and the standardization of transport protocols has made it so, that any activity, any world-3 representation that can be digitalized and networked, has entered into a space where interaction with any other world-3 representation is a possibility. That space is the one market world. — the potentialinteroperability of all transactions and value systems that can be digitalized. I do not imply that this leads like Ice-9 ²⁸ to one market for all digitally-capable human activity — like jump off the ground and you are on your way to the moon. But what ever happens, the transaction costs of trust will have a lot to do with it.

Design and Development by deepX Ltd.