public + private reputations

eBay is an example of a public reputation system. What motivates users to provide feedback? What motivates them to be honest? Are they equally motivated to provide positive and negative feedback? Compare the relationship of the feedback provider with the subject of the feedback and with the audience for the feedback. What influence does this has on the reliability of the system?

as an occasional user of ebay nowadays, in the initial stages (my membership infancy) i tended to give feedback after a transaction, either reminded by the other as a reciprocal courtesy ("please provide feedback for me, i'll do for you") or if something went disastrously wrong (which was very rare). now i hardly do so, because it takes effort to manuever through ebay's site to rate, as well as decide if my experience was truly 'positive.' i would suspect that users would be motivated to provide feedback if (1) they were reminded through email or messaging to do so, preserving ebay as a robust and self-maintaining community (i.e. purely member courtesy), (2) there was some sort of reward to offset the "cost" of providing feedback (e.g. rating a seller would give you a chance to win a daily giveaway from ebay), (3) something in the transaction was noticeably extraordinary (either uberwonderful, like many freebies and samples along with your order, or utterly terrible, like a no-show or an item that was extremely ill-described), (4) if it gave them incentive in the long-run (the more ratings they give, the higher benefits or prestige they accumulate in the system) or the short-run (giving a rating would warrant a small microdiscount on their next transaction), (5) the buyer and seller had an established relationship and would want to continue cordial business relationships, or (6) if there was a public display of showing how many transactions the user participated in comparison to how many of those the user rated (i.e. "cyberbuyer01 has given feedback to 20 of his 56 transactions", or to mask quantity and reflect quality, "cyberbuyer01 has given feedback to 38% of his transactions). this way, the pressure of peer comparison and public display of trustworthiness and followup-ship would encourage users to consummate their auction dealings fully with a rating.

for users to provide honest feedback, there are different cases. i believe there should be a more robust rating system other than +, -, or neutral. there are different components to a transaction, like timeliness, quality, communication, etc. if one could rate on different facets (similar to epinions, or zdnet, where things from 'value' to 'ease of use' to 'would recommend to others' are addressed) . that way, a user could honest report 'this person had great communication, although the item was a little shoddier than expected' with different ratings on different criteria, without completely generalizing the entire experience with a single + or - rating. in this situation, the user would most likely put a + because the average experience was positive although there were a few blips the seller could have improved. the comments area can also be very valuable for honest feedback; the text is provided for others to read if they wish, with clearer details on what exactly happened. in this way, people can defend their ratings extensively.

the 'dishonesty' of putting generous positives instead of malicious negatives seems to be the trend. unwarranted positives occur to soften an online critical blow, or easy forgiveness. perhaps different methods of rating would help. 'was your experience positive' might be less helpful that 'would you do business with this seller again' or 'would you recommend it to a friend'... something that reflects a more personal connection with the transaction might infuse more honesty into it. otherwise, it's like a poll, survey, or test, where users may feel disconnected with the process and click 'this whole thing was fine' without much thought.

as for malicious negatives, the wronged party could petition the moderator of the site, and after further investigation, the rating would be erased or modified. i could see a scenario in which the wronged party would put an initial investment (he/she pays a fee to have the transaction undergo inquiry), and if the party was indeed wronged, the malicious rater would pay a penalty to the wronged party and feedback would update accordingly. this way, a wronged party would only prosecute if they really thought their reputation was endangered unnecessarily, and malicious raters would think twice before reporting something abnormally dishonest.

as for the current system, i believe users are more inclined to provide positive feedback if they're reminded somehow of their duty, and everything went okay. i do think that users provide negative feedback (without external reminder) if the transaction was abhorent. so the motivation is different for different situations. neutral to positive to 'positive average' to 'slightly negative average' experiences receive a +, whereas 'extremely positive negative' experiences receive a -. i would say the motivation to rate a slightly negative to neutral experience as - would be low, given the common courtesy we all grant on community members, a cordial business forgiving. who wants to brand someone as NEGATIVE when (1) you're complete strangers and (2) you're only going to have one or few transactions with this person anyway? branded someone as NEGATIVE is a clear public admonishment, warning others to 'stay away', since negative ratings are quite prominent (even one can bruise a business reputation) when viewing someone's profile.

therefore, because users are tentative to brandishing someone with whom that they've only marginally interacted with an online scarlet letter of sorts, and with high probability of not doing business dealings any longer, the reliability of the system is scant. if i do business with someone who has a very good record established online, yet i felt our particular dealing to be a bit shady, i'd feel guilty if i permanently and visibly tainted the seller's reputation (with my name signed behind it), a reputation that would directly affect all of this seller's future transactions. that's a huge responsibility on the rater's part, to provide feedback on this one particular instance without thinking too much about how it will reflect the seller on all transactions in aggregate. if every buyer who rates one seller with a thus-flawless record, for instance, is afraid of being the first to admit that the seller isn't positive because they'd be the 'ruiner', then no one will cast the first stone, and the seller would continue to give slightly sub-par service while still maintaining a good profile.

Describe in clear and concise detail the costs and benefits to the participants in a model public reputation system of the different possible actions they may make (i.e. deceptive or honest acting; punishing or not punishing transgressors). Do the same for a model private (gossip based) system. Don't forget issues such as the possible mismatch between the rater and the audience, etc.

[online!]

public : costs to honest ::
- if the rating is negative and traceable, the ratee may seek retribution by reciprocating a negative rating or spreading dishonest or malicious information about the rater
- if a buyer on ebay is known to give honest, if negative, reviews, sellers may manuever or manipulate the system in order to avoid transactions with this buyer, therefore removing the buyer's full choice or opportunity of goods
- if the seller gives a buyer a negative rating, it might pose a serious risk for further business with this buyer or other potential buyers who will fear a negative rating themselves
- a person who rates the other as 'negative' suffers guilt that they are publicly marring the other's reputation for countless others to view and judge from, a person they may never encounter again

public : benefits to honest ::
- if a rating is negative, the ratee will be able to more visibly and effectively recognize their transgressions and proactively improve or renew their reputation to improve the community for the general good
- if a transaction is honestly rated positive, both the rater and ratee are in favorable terms, and their public reputation looks good for both their future transactions
- if more ratings are honest (positive or negative), it may remove the stigma attached to the stray negative rating (i.e. the fear of casting the first stone dissipates)

public : costs to dishonest ::
- if a dishonest rater is discovered, others will be wary of dealing with this person, fearing an undeserved poor rating on their public reputation
- if a subpar seller is rated positive overall because of reasons dealing with courtesy or persuasion, for example, future buyers who may feel comfortable dealing with this 'good' seller may end up suffering from worse service than they expected
- conversely, if a reasonably good seller is rated poor when the transaction was just fine, the seller's reputation is marred unnecessarily, which will cost them in future business as other buyers might seek other sellers with more positive ratings

public : benefits to dishonest ::
- a subpar seller that receives undeserved positive ratings will continue business with others with a positive appearance, and may abuse his admirable reputation to cheat unsuspecting buyers
- positive ratings, even when inaccurate, can prolong the 'everything is fine' feeling, giving all parties an ego stroke, and blossom feelings of satisfaction
- a buyer may be able to negotiate a reduced price of sale if a positive rating is promised, or vice versa; bribery via marks of reputation

public : costs to punishing transgressors ::
- users will be more wary of giving a negative rating, in fear of the ratee prosecuting them as dishonest; positive ratings will be inflated
- perhaps the users of the community need to pay a membership or maintenance fee in order to support the moderation of the system

public : benefits to punishing transgressors ::
- the ratings system will be more reliable if dishonest ratings are punished and minimized; users will be more likely to believe in the system and be motivated to uphold their good reputations in order to continue beneficial relations with others
- raters will be more confident that their comments /ratings will be read as useful and accurate

public : costs to not punishing transgressors ::
- unreliable raters can infiltrate the system, thereby diluting the reliability of ratings and basically rendering the system useless and inaccurate
- users will be very cautious as to avoid any risk of reputation soiling; this cautiousness may cause stagnation of the system and unease during encounters

public : benefits to not punishing transgressors ::
- users would not feel as obliged to post positive ratings; they wouldn't fear retribution or public ostrasization
- retaining freedom of expressing one's personal belief of the situation; how can an outsider judge whether someone's rating is viably true or false?

[gossip!]

private : costs to honest ::
- if someone makes an honest, yet negative, remark about somebody else, they risk revenge or retribution on their own reputation; rumors can fly
- a negative report on someone, even if honest, may harm feelings, cause distrust, sever relationships, etc
- outward behaviors become falsified and superficial, as others maintain their 'best behavior' to avoid a slip-up to be passed among the others via gossip; less risks or social 'mistakes' are ventured; authenticity gets smothered by niceties and perhaps-fake cordialities

private : benefits to honest ::
- if one's remarks are eventually known to be reliable and accurate through eventual actions and words of others (proof of concept), then other's will trust this person's testimonies in the future (benefit to the truthful teller)
- if there are established reliable tellers, then the ones who benefit are those who act in good spirit and have others see them as such; their good reputation is cross-checked by many
- a social misfit who hears what others think of them may be motivated to change or check their behavior in order to minimize offending others

private : costs to dishonest ::
- if someone spreads false rumors about one or more persons and is discovered, their credibility is greatly disminished
- offensive people who don't possess a negative reputation (for any given reason, like other people are afraid of being discourteous or impolite by saying so) may obliviously continue to be obnoxious or nasty because they are mislabelled as agreeable people

private : benefits to dishonest ::
- if a negative spell is given a positive spin, human mistakes are forgiven + forgotton... transgressions are elided over; conflicts are smoothed or put into perspective
- a person may give a dishonest impression of someone if they are accepted into a desirable social group; e.g. a teen might fawn over a certain person that they secretly dislike, in order to feel more accepted by a clique at school

private : costs to punishing transgressors ::
- exclusion from a desirable social group, or fear of looking too judgmental; e.g. i won't rat out my boss who's spreading nasty rumors because i don't want to lose my job
- quelling personal freedom of opinion of others; bad-mouthing someone might be an effective form of catharsis

private : benefits to punishing transgressors ::
- someone who spreads false gossip or rumors can be ostrasized, and others can live with less fear that their reputation will be tainted or questioned
- good people will be more motivated to maintain their good image

private : costs to not punishing transgressors ::
- rumors can run rampant, wrongly marring otherwise reputable people; e.g. people might not want to talk to this girl because they have heard stories about her that render her untrustworthy or dispicable, even though the stories are completely untrue
- tongues can run either way; reliability of the word-of-mouth system is more diluted; more costly methods to confirm suspicions on people may be necessary

private : benefits to punishing transgressors ::
- good people can maintain their good reputation more easily
- the motivation to be honest runs high if dishonesty warrants a high cost of social punishment

Describe a real life situation in which reputation information is exchanged, either publicly or privately. Ebay has been extensively written about, so pick something else. You can use Amazon.com reviews, teacher recommendations, gossip exchange in a social group, etc. Describe the situation in detail. Is the information reliable? What keeps it so? What is the relationship among subjects, raters and audience? What are the costs and benefits the participants receive? How well does this fit with the model you described above?

let's envision a scenario that a prospective graduate student is visiting a department of choice and seeking the best fit for an academic advisor. this student (named josi, say) isn't terribly familiar with the professors, save for some papers she's read online and perusal of the academic websites. josi arrives at the university and has some interviews for the day, but beforehand she has some time to scope out the different styles of the professors. she has arranged to speak with some of the current graduate students; in these informal talks she will be able to ask questions regarding the professor. the current students probably will be very reliable and truthful in their testimony, since they have a personal connection (and empathy) for the prospective, and they work very closely with their respective advisors. the costs for being honest are relatively low: students who exchange more dicey, truthful gossip have an assumption that such details ("prof smith is a wonderful person, but he's so inaccessible!") won't be aired publicly or extensively. the benefits for being honest are high: telling about one's experience helps everyone involved, and it's an act of good will to ensure that josi gets a good match (and saves some headache in the long-run). dishonest testimony from students would be rare, probably with fear that the professor might be in earshot, being paranoid that certain words would eventually be traced back to them, or having an established decorum to not say anything negative about anyone else. the relationship between the current students and josi remains very close; the students were once in josi's shoes, and can understand the confusion and determination of such a quest. graduate students are a reliable source because they have no incentive to lie to a peer. josi realizes this and is grateful; she's convinced that the students she's talked to are honest. however, she knows also to take everything with a grain of salt, because sometimes clashes are due to differences in personality, habits, experience, etc that's very individual-based rather than general.

josi then has time to go to the library and google some of the professors she's interested in. each professor's website contains lots of information, from their CV (very much reliable, since others in the field can easily view to cross-check or confirm facts) to classes they're teaching (reliable: can talk to students in said class) to (perhaps) their personal page with pictures of family and friends. josi can also see professional references to their papers. having lots of references to oneself is a signal of prestige within an academic community, and is usually reliable if the field is broad and established with many contributing parties. after viewing lots of papers and seeing current projects, josi has a better sense of what their styles are like, but online profiles can definitely skew her perceptions. if a professor has an ugly, ill-designed website, is that a direct reflection on them, or did they just have a non-artistic UROP code it up for them? if a professor has a staid, solemn picture of themselves online, is that because they're traditional and boring, or just a professional digital facade to a joking and vibrant real-live personality? since online sources are very much public, and in this case professionally-oriented, indicators of the person are filtered down to pose a professional/creative/prolific profile. helpful to josi, but not terribly indicative on what kind of advisor they'd be.

she then goes off to her interviews, talking one-on-one with each professor. seeing them in the flesh carries a lot more meaning now, after processing the words from the other fellow students and having an idea of their public online profile. here, she can bounce ideas around and balance others' experience with that of her own. now, individual personality and dynamics can take over, as josi has her own direct dealings. although the first impressions she gathers at the interviews won't necessarily point her in the direction of the best advisor (half an hour in an office is hardly indicative of behavior throughout months, years) she can get a feel for interaction style. the professor themselves are a mostly reliable indicator of who they are (since they are them), but it's up to josi to integrate her experiences with the words of others to make a final decision.