15. Reputations and Credentials
THE CYPHERNOMICON: Cypherpunks FAQ and More, Version 0.666, 1994-09-10, Copyright Timothy C. May. All rights reserved. See the detailed disclaimer. Use short sections under "fair use" provisions, with appropriate credit, but don't put your name on my words.
15.2. SUMMARY: Reputations and Credentials
15.2.1. Main Points
- "a man's word is his bond"
- reputations matter
- the expectation of future interaction/business is crucial
15.2.2. Connections to Other Sections
- see section on Crypto Anarchy for why reputations matter
15.2.3. Where to Find Additional Information
- very little published on this
- Bruce Benson's "The Enterprise of Law"
15.2.4. Miscellaneous Comments
- this is another "transition" chapter, laying the groundwork for Crypto Anarchy
15.3. The Nature of Reputations
15.3.1. The claim by many of us that "reputations" will take care of many problems in crypto anarchic markets is disputed by some (notably Eric Hughes). To be sure, it will not be a trivial issue. Institutions take years or decades to evolve.
15.3.2. However, think of how often we use reputations: friends, books, movies, restaurants, etc
15.3.3. Reputations and other institutions will take time to evolve. Saying "the market will talke care of things" may be true, but this may take time. The "invisible hand" doesn't necessarily move swiftly.
15.3.4. "What are 'reputations' and why are they so important?"
- a vague concept related to degree of believability, of trust, etc.
- "we know it when we see it"
- (sorry for the cop out, but I don't have a good definition handy...James Donald says studying reputatons
is "nominalist hot air" [1994-09-02], but I think it's quite important)
- obvious, in ordinary life, but in the cyberspatial context
- reputation-based systems
- escrow, expectations
- "reputation capital"
- like book or music recommendations
- obvious, in ordinary life, but in the cyberspatial context
- web of trust (is different than just "trust"---tensor, rather than scalar)
- Actually very common: how most of us deal with our friends, our enemies, the books we read, the restaurants we frequent, etc.
- we mentally downcheck and upcheck on the basis of experience; we learn
- Are there examples?
- Eric's objections
15.3.5. "How are reputations acquired, ruined, transferred, etc.?"
- First, reputations are not "owned" by the person to whom they are attached by others
- the algebra is tricky...maybe Eric Hughes or one of the other pure math types can help straighten out the "calculus of reputations"
- reputations are not symmetric: just because Alice esteems Bob does mean the reverse is so
- reputations are not transitive, though they are partially transitive: if Alice esteems Bob and Bob esteems Charles, this may cause Alice to be somewhat more esteemful of Charles.
- a tensor matrix?
- a graph?
- Any holder of a reputation can "spend" some of his
- in praise or criticism of another agent
- Any holder of a reputation can "spend" some of his reputation capital
- in reviews (think of Siskel and Ebert "spending" some of their reputation capital in the praise of a movie, and how their own reptutations will go up and down as a function of many things, including especially how much the viewing audience agrees with them)
15.3.6. "Are they foolproof? Are all the questions answered?"
- Of course not.
- And Eric Hughes has in the past said that too much importance is being invested in this idea of reputations, though many or even most of us (who comment on the matter) clearly think otherwise.
- In any case, much more study is needed. Hal Finney and I have debated this a couple of times (first on the Extropians list, then a couple or more times on the Cypherpunks list), and we are mostly in agreement that this area is very promising and is deserving of much more thought--and even experimentation. (One of my interests in crypto simulations, in "protocol ecologies," is to simulate agents which play games involving reputations, spoofing, transfers of reputations, etc.)
15.3.7. Reputations have many aspects
- the trading firm which runs others people's money is probably less "reputable" in an important sense than the trading firm in which partners have their own personal fortunes riding...or at least I know which one I'd trust!
- (But how to guarantee one isn't being fooled, by a spoof, a sham? Hard to say. Perhaps the "encrypted open books" protocol Eric Hughes is working on will be of use here.)
15.4. Reputations, Institutions
15.5. Reputation-Based Systems and Agoric Open Systems
15.5.1. Evolutionary systems and markets
- markets, emergent order, Hayek, connectionism
15.5.2. shell games...who knows what?
15.5.3. key is that would-be "burners" must never know when they are actually being tested
- with devastating effects if they burn the tester
- example: how to guarantee (to some degree of certainty) that an anonymous bank is not renegging (or whatever)?
- e.g., a Swiss bank that denies knowledge of an account
- key is that bank never know when a withdrawal is just a test (and these tests may be done frequently)
- the importance of repeat business
15.5.4. another key: repeat business...when the gains from burning
someone are greater than the expected future business...
15.5.5. reputations are what keep CA systems from degenerating into flamefests
- digital pseudonyms mean a trail is left, kill files can be used, and people will take care about what they say
- and the systems will not be truly anonymous: some people will see the same other people, allowing the development of histories and continued interactions (recall that in cases where no future interaction is exected, rudeness and flaming creeps in)
- "Rumormonger" at Apple (and elsewhere) always degenerates into flames and crudities, says Johann Strandberg
- but this is what reputations will partly offset
15.5.6. "brilliant pennies" scam
15.5.7. "reputation float" is how money can be pulled out of the future value of a reputation
15.5.8. Reputation-based systems and repeat business
- reputations matter...this is the main basis of our economic system
- repeat business...people stop doing business with those they don't trust, or who mistreat them, or those who just don't seem to be reputable
- and even in centrally-controlled systems, reputations matter (can't force people to undertake some relations)
- credit ratings (even for pseudonyms) matter
- escrow agents, bonding, etc.
- criminal systems still rely on reputations and even on honor
- ironically, it is often in cases where there are restrictions on choice that the advantages of reputations are lost, as when the government bans discrimination, limits choice, or insists on determining who can do business with who
- Repeat business is the most important aspect
- granularity of transactions, cash flow, game-theoretic analysis of advantages of "defecting"
- anytime a transaction has a value that is very large (compared to expected future profits from transactions, or on absolute basis), watch out
- ideally, a series of smaller transactions are more conducive to fair trading...for example, if one gets a bad meal at a restaurant, one avoids that restaurant in the future, rather than suing (even though one can claim to have been "damaged")
- issues of contract as well
15.6. Reputations and Evolutionary Game Theory
15.6.1. game of "chicken," where gaining a rep as tough guy, or king of the hill, can head off many future challenges (and hence aid in survival, differential reproduction)
15.7. Positive Reputations
15.7.1. better than negative reputations, because neg reps can be discarded by pseudonym holdes (neg reps are like allowing a credit card to be used then abandoned with a debt on it)
15.7.2. "reputation capital"
15.8. Practical Examples
15.8.1. "Are there any actual examples of software-mediated reputation systems?"
- credit databases...positive and negative reputations
15.8.2. Absent laws which ban strong crypto (and such laws are themselves nearly unenforceable), it will be essentially impossible to stop anonymous transactions and purely reputation-based systems.
- For example, Pr0duct Cypher and Sue D. Nym will be able to use private channels of their own choosing (possibly using anonymous pools, etc.) to communicate and arrange deals. If some form of digital cash exists, they will even be able to transfer this cash. (If not, barter of informations, whatever.)
- So, the issues raised by Hal Finney and others, expressing doubts about the adequacy of reputation capital as a building block (and good concerns they are, by the way), become moot. Society cannot stop willing participants from using reputation and anonymity. This is a major theme of crypto anarchy: the bypassing of convention by willing participants.
- If Alice and Bob don't care that their physical identies are unknown to each other, why should we care? That is, why should society step in and try to ban this arrangement?
- they won't be using "our" court systems, so that's not an issue (and longer term, PPLs will take the place of courts, many of us feel)
- only if Alice and Bob are counting on society, on third parties to the transaction, to do certain things, can society make a claim to be involved
- (A main reason to try to ban anonymity will be to stop "bad" activities, which is a separate issue; banning of "bad" activity is usually pointless, and leads to repressive states. But I digress.)
15.8.3. Part of the "phase change": people opt out of the permissionslip society via strong crypto, making their own decisions on who to trust, who to deal with, who to make financial arrangements with
- example: credit rating agencies that are not traceable, not prosecutable in any court...people deal with them only if they think they are getting value for their money
- no silly rules that credit rating data can "only" go back some arbitrary number of years (7, in U.S.)...no silly rules about how certain bankruptcies "can't" be considered, how one's record is to be "cleared" if conditions are met, etc.
- rather, all data are considered...customer decides how to weight the data...(if a customer is too persnickety about past lapsed bills, or a bad debt many years in the past, he'll find himself never lending any money, so the "invisible hand" of the free market will tend to correct such overzealousnesses)
- data havens, credit havens, etc. (often called "offshore data havens," as the current way to do this would be to locate in Caymans, Isle of Man, etc.)
- but clearly they can be "offshore in cyberspace" (anonymous links, etc.)
15.9. Credentials and Reputations
15.9.1. debate about credentials vs. reputations
- James Donald, Hal Finney, etc.
- (insert details of debate here)
15.9.2. Credentials are not as important as many people seem to think
- "Permisssion slips" for various behaviors: drinking age, admission to movie theaters, business licenses, licenses to drive taxicabs, to read palms (yes, here in Santa Cruz one must have a palm-reading license, separate from the normal "business license")
- Such credentials often are inappropriate extensions of state power into matters which only parents should handle
- underage drinking? Not my problem! Don't force bars to be babysitters.
- underage viewing of movies? Ditto, even more so.
15.9.3. Proving possession of some credential
15.10. Fraud and False Accusations
15.10.1. "What if someone makes a false accusation?"
- one's belief in an assertion is an emergent phenomenon
- assertion does not equal proof
- (even "proof" is variable, too)
- false claims eventually reflect on false claimant
15.10.2. Scams, Ponzi Schemes, and Oceania
- Scams in cyberspace will abound
- anonymous systems will worsen the situaion in some ways, but perhaps help in other ways
- certainly there is the risk of losing one's electronic cash very quickly and irretrievably (it's pretty far gone once it's passed through several remailers)
- conpersons (can't say "con men" anymore!) will be there, too
- Many of you will recall the hype about "Oceania," a proposed independent nation to be built on concrete pontoons, or somesuch. People were encouraged to send in donations. Apparently the scheme/scam collapsed:
- "It turned out to all be a scam, actually. The key people involved, Eric Kline and Chuck Geshlieder, allegedly had a scheme set up where they repeatedly paid themselves out of all of the proceeds." [anonymous post, altp.privacy, (reprint of Scott A. Kjar post on Compuserve), 1994-07-28] - or was it Eric Klein?
15.11. Loose Ends
15.11.1. Selective disclosure of truth
- More euphemestic than "lying."
- Consider how we react when someone asks us about something we consider overly personal, while a friend or loved one may routinely ask such questions.
- Is "personal" the real issue? Or is that we understand truth is a commodity with value, to be given out for something in return?
- At one extreme, the person who casually and consistently lies earns a poor reputation--anyone encountering them is never certain if the truth is being told. At the other extreme, the "always honest" person essentially gives too much away, revealing preferences, plans, and ideas without consideration.
- I'm all for secrets--and lies, when needed. I believe in selective disclosure of the truth, because the truth carries value and need not be "given away" to anyone who asks.
15.11.2. Crytography allows virtual networks to arrange by cryptographic collusion certain goals. Beyond just the standard "cell" system, it allows arrrangements, plans, and execution.
- collecting money to have someone killed is an example, albeit a distasteful one