Student Comments –
Brett Harrison
This paper analyzes the popular online marketplace, eBay, in the context of reputation systems. eBay employs a reputation system in which both the buyer and seller of a transaction can rate the other after the transaction has been completed. This rating is either +1, 0, or -1, and can be accompanied by a comment. The authors set up an experiment which compared the amount that postcards could be sold for when offered by a reputable seller as opposed to new sellers. As we would expect, the reputable seller was able to sell the postcards at a higher price, since that seller had a well established and positive reputation. The interesting result is that one or two negative ratings did not have profound effects on the new sellers' ability to sell.
My main question is about the universality of these results. Does the behavior seen in the paper's experiment carry over to transactions for items besides postcards? Perhaps if an expensive car is being offered on eBay, a reputable seller can sell the car at a much higher price than a new seller, or perhaps these two prices are roughly equal (i.e. reach some sort of equilibrium). Perhaps the results will not hold with items that are even cheaper than postcards. I am concerned about the breadth of this experiment, although the results seem intuitive enough.
Victor Chan
The main contribution of the paper by Resnick et al, was to investigate the role that user reputation plays in transactions done on ebay. The paper takes an empirical approach to finding out the affects of seller reputation on the price of an item sold. Experiments were done using a strong seller with high reputation and a new seller (no reputation). Then the two would sell the same item and it was found that the STRONG seller was able to get on average 8.1% more for an item than the NEW seller. Another result found was that negative feedback in the reputation did not affect prices. The importance of these results show that online reputation does matter in an ecommerce, however it appears that positive reviews outweigh the effects of negative reviews.
The first result was something I would have intuitively expected, since a higher reputation, means that the seller is more trust worthy so their prices naturally command a higher premium. However, the second result is very strange. If negative feedback does not affect the seller that much, then is the measure of the seller reputation only the number of positive ratings? After thinking about my own ebay experiences, I tend to only purchase from sellers with 10,000+ ratings, and normally with those sellers they would have a 98% positive feedback. Which means then there were 200 negative feedback for that seller, which I would consider quite a few complaints. However the way ebay presents the information, one would only see 98% +ve and 10,000+ ratings, which hides away the fact that 200 people have had bad experiences with this company.
I am also curious to see the results of this experiment, when the items being sold are at a higher value. In this case a negative review might cause the sale to not occur. People might do more research when purchasing a $1000 camera, and if a negative review says a seller took the money and did not send the item, then people will likely not trust them for high value purchases.
Avner May
I thougth this article investigated an important issue -- can product quality be communicated effectively in internet-market places? The issue is that if the quality of the seller/product cannot be communicated well, than "low-quality products and sellers will drive out those of high quality and the market will shrivel." This is equivalent to the problem of "lemons" in used-car markets, which from microeconomic theory predicts that if sellers of good cars are unable to credibly communicate the quality of their product, then good and bad cars look identical to buyers; thus buyers are willing to pay equal amounts for good vs. bad cars (the expected value to them of the car), and if this is less than the amount good car sellers value their cars, the market will shrivel, and only bad cars will be sold on the market, at a low price. Due to this potential problem, it is important that in internet markets, which are becoming increasingly common, and growing in size, there exist effective ways of communicating product quality. Whereas in "physical markets" there are many ways of communicating product quality and seller reliability (advertisements, paying rent for store, buyers can physically inspect the products, etc.), this is much trickier on the internet. This article studies a specific marketplace, eBay, and investigates, via a controlled experiment, whether a seller's reputation allows a seller to credibly communicate the fact that they are reliable, and sell quality products. It was determined that, in fact, buyers are willing to pay on average 8.1% of the selling price more if they are buying from a seller with a high reputation. This shows that high reputation is a (at least partially) effective communication mechanism. For this reason it seems, reliable sellers of quality products choose to enter the market, and thus the "lemon" problem is in this case not causing the market to shrivel.
One thing I am wondering is whether this would hold in all types of markets. Lets imagine the following scenario: Lets say good and bad used cars are being sold on eBay (I'm using the car experiment just to be consistent with the lemon problem), where half the cars being sold are good, and half are bad. Say also that sellers value good cars twice as much as they value bad cars. According to this article, if the sellers of good cars have earned good reputations, whereas the sellers of bad cars have not established good reputations on eBay (they are new), buyers would only be willing to pay 8.1% more for the good cars than for the bad cars. This would NOT be enough to compensate for the true difference in quality of the products. Thus, in this case, the market would unravel, and only bad cars would be sold on eBay. So essentially, it would seem that only markets in which good vs. bad products differ in value by less than ~8% (or in which the cost of being a reliable seller vs. an unreliable seller differs by less than ~8%), would the market not unravel.
A question I think could be interesting to explore is how effective are other ways on eBay of communication product quality? Is reputation the main one? Are pictures and descriptions enough to differentiate high quality products from low quality products? If reputation is the main way of communicating product quality, then the statements I made above would indicate that many markets would unravel. However, if product quailty is communicated mainly in other ways (and good reputation simply indicates that a seller is reliable), than the statement I made above would not hold for as many markets.
I think it could be very interesting to do a similar controlled experiment but with more types of products (to see if some markets unravel, while others don't). I think it could also be very beneficial to test how effective descriptions and pictures are at differentiating high and low quality products. I am more interested in the differentiation of high and low quality products on eBay than I am the differentiation between reliable and unreliable sellers on eBay, and I do not think this article does a satisfactory job at studying this issue.
Hao-Yuh Su
This paper provides empirical results to support some theoretical
arguments about the reputation system on eBay. Under their experiment
setting, first, it is proved that sellers with higher reputation outperform
those with lower reputation; second, negative feedbacks hardly affect
the performance of new sellers. First of all, I have a question about the
object they chose. In the experiment, the object under study is an eBay
dealer whose business is primarily in vintage postcards. Though the
author has claimed that the dealer is an established dealer with high
reputation, and has dozens of items for sale regularly each week on
eBay, but I don't see his representability when we are talking about
the whole transaction activities on eBay. I would like to know the
components of the business on eBay. What is the percentage of the
postcard business? Are the transaction activities in this specific category
representable? Especially when the authors are probing the possible
treats to validity in the final part of this paper, they attribute the puzzling
result of the second experiment to the small sampling size. However,
wouldn't that because when buyers are looking for postcards, they don't
care about negative feedbacks that much? After all, it's not like buying a
computer or something expensive or something has high uncertainty of
quality; the loss of buyers is relatively limited. Second, I would like to
know when the authors are conducting the first experiment, if the quantity
of lots that one buyer is purchasing is considered, or, they are only
counting heads of the buyers. If it is the previous case, I am wondering
if the threats to validity from repeat buyers can be omitted to some degree,
since a buyer can order two quantities at one time or buy one quantity each
time separately and lead to the same result.
Travis May
This paper attempted to quantify the value of reputation on eBay by using a controlled, randomized experiment. Two accounts were set up to sell the same goods, one with a positive reputation and the other with a new identity, and buyers were willing to pay a 8% to the high-reputation seller. This goes a long way towards explaining price heterogeneity on the internet in general, as low search costs and product homogeneity leave only differences in trust as a possible explanation for why the market price varies on the internet.
This study leads to a very intriguing potential extension in economics. In order to further quantify the value of reputation, a website could establish a reputation market, in which reputation points could be bought and sold. While this may confuse the typical conception of reputation, it could reinforce the same incentives by providing an explicit price for not being trustworthy and a quantified value of reputation. Furthermore, in typical internet markets, reputation has different values to new sellers and old sellers. Sales on eBay are essentially repeated games, and being an honest dealer is a Nash equilibrium if and only if the punishment to future sales from a negative piece of feedback outweigh the present benefit of cheating on a contract. This means that, as a seller is nearing “retirement,” their incentive to provide quality decreases. A market for reputation could remedy this problem and smooth the incentives over time.
Brian Young
The paper outlines a field experiment performed over eBay to determine to what extent a seller's reputation affects purchases. The results suggest that an established reputation gives sellers a non-negligible advantage.
The discussion of eBay's reliance on trust and the feedback system suggests that as in other systems we've looked at, reputation in eBay doesn't seem to translate well to real reputation. I was reminded of crowdsourcing sites like Taskcn, where there isn't an easy way for talented graphic designers to break out from seeking spec work and into establishing an actual business. In eBay, beyond feedback, there seems to be little to encourage customer loyalty or brand recognition.
I was impressed with the setup of the study, and I don't doubt the results, but I would have liked to see more data points - a user in between strong and no reputation, for instance.
Actually, the part of the study that was most interesting to me was why exactly the established eBay dealer agreed to participate, particularly when the study's primary hypothesis was that he would lose money by selling from the invented accounts instead of his usual account.
Sagar Mehta
Ebay is much like any market suffering from a problem of asymmetric information. Reputation mechanisms attempt to alert users on metrics such as how reliable/timely a seller is. This paper tried to study in a controlled setting the impact reputation plays in determining the price a consumer is willing to pay for some product. While reading this paper, I thought about how the current reputation system utilized by ebay could be improved upon. The system used by ebay seems quite basic, which is both a good and a bad thing. One good aspect of this is that it is very easy to rate someone after receiving an item. One bad aspect I find is that there isn't much optionality in the rating system. There is no inbetween in "positive" vs "negative" ratings. In contrast, amazon marketplace rates sellers on a 5 star system. I think an interesting area for empirical research would be to compare the expressiveness of the rating system with the price a user is willing to pay for an item.
The system performs pretty well for the larger sellers, but once you get down to a few people, the rating obviously becomes less important. For a new user who has no ratings or someone who just sells things at low volume, it might be hard to develop a reputation at all. What does this imply about the nature of ebay? In its most basic sense people think of ebay as a marketplace where anybody can buy or sell used or new goods. But it seems that positive ratings should increase exponentially over time. That is, if you have 1000 positive ratings today, you will increase ratings at a faster clip than someone who starts with 20 positive ratings (by virtue of the fact that more people will buy your product and offer ratings). So, does this mean that ebay is biased towards large, reputable sellers and is really just another avenue for these sellers to push goods?
More importantly, how can we improve the usefulness of reputation systems in the case where data (i.e. ratings) are sparse? I think there may be other ways to align the incentives of the seller. For example, say ebay takes a commission for every product a user sells; we could design a system such that the % commission ebay takes decreases with higher ratings on each individual sale. For example, if I get a rating of 5 on a scale of 1-5, ebay may get 0% commissions on my deal, but if I get a rating of 2, ebay will take 3% commissions on my deal. Are there online mechanisms which seek to go beyond just using reputation/ranking as a way to get better quality products?
Xiaolu Yu
The main contribution of this study is providing us with a randomized controlled look at the value of eBay reputations in the natural setting of actual eBay auctions. The findings showed that eBay's feedback system works as suggested by Yhprum's Law, by rewarding sellers who have more positive feedback and hence build up a strong reputation. The paper presents evidence that sellers with a good reputation can expect about 8 percent more revenue than those without a good reputation on eBay.
One surprising result discovered by the researchers is that one or two negative feedbacks did not hurt new sellers, but things remain unclear include why buyers are willing to cut new sellers with negative feedback a little slack, how often cheating occurs among new sellers because of not sufficient attention to negative feedback on them, and whether the price premiums reflect a reputation equilibrium.
As suggested by the authors, it would be interesting to assess whether the market is over-, under-rewarding, or correctly rewarding reputations. In order to do this, it would be necessary to estimate profit by cheating, number of transactions required to build up a strong reputation, and number of cheating to destroy a strong reputation. These factors are also important in justify if the fact "all significant sellers have strong reputations" implies an equilibrium – the Feedback Forum works.
Since the experiment executed in this paper cannot rule out the threats come from repeat customers and multiple purchase, it is reasonable to doubt that a seller are able to manipulate (somehow ask a group of buyers to make multiple purchases) to build up reputation, or a seller can also be targeted by a group of buyers to deliberately lower her reputation. Moreover, eBay model considers the feedback of every user with the same weight, and this could be exploited by the malicious user. For example, users can threaten to leave a negative feedback and therefore potentially destroy the other user's reputation unless they get a discount on their purchase.
Haoqi Zhang
The main contribution of the paper is the detailing of a controlled experiment on ebay to test the affect of reputation on the market's willingness to pay for items. In particular, the author's setup differs from previous experiments in that it is an actual experiment on ebay on which particular items were sold under different seller conditions (e.g. one with lots of positive reputation versus some others with positive reputation (but a few negative ones as well)). This way, the authors are able to better control for factors than other studies because the setup is nearly identical for the two sellers other than reputation.
I found the experimental setup to be sound and the explanations satisfactory. In particular, I liked how the authors critiqued their own experiments and tried to explain the result of the minimal impact of negative impact. On that note, I think the authors should have tested with a longer reputation history, where we have a seller that is not only established but still has quite a negative rating. Based on their argument that most people don't read the details of the ratings, this seems still possible to do in their controlled setup. Also, having more extreme conditions, e.g. a seller with almost all negative feedback, would be interesting to include as well.
Which beings me to this point: while a controlled experiment has many benefits, it is also limited in what it is able to test and the scale of the experiment that can be ran. The postcard setting may not translate to a setting where the object is of higher value (say in the hundreds or thousands of dollars), where a seller may care a lot more about negative feedback and reputation than here.
Alice Gao
The main contribution of this paper is to devise a controlled approach to study how information about reputation affects user behaviours on auction sites such as eBay. This is important because a well devised experiment could help to eliminate the effect of noises as much as possible and therefore reach a well sustained conclusion. The major weakness of this paper is, similarly, that there is still too much noise in the experimental results. Therefore, the conclusion of the paper is not too meaningful in this respect.
In general, I don't find the results of these empirical studies to be useful mainly because there are potentially too many factors affecting the observed results. The authors tried to identify many extra factors whose effects cannot be ruled out for the experiments carried out. Future work can use these analyses and attempt to revise the design of the experiment to eliminate the effects of these factors.
Nikhil Srivastava
Resnick et al provide an interesting empirical study of eBay as a model reputation mechanism. Motivated by a series of studies with conflicting results, this paper seeks to identify the factors that were ignored or conflated in previous studies regarding reputation's role in online marketplaces and to conduct a carefully structured study that eliminates or isolates them. The setup of their experiment is fairly compelling, and after ruling out the relevance of several alternative theories, they show the market rewards established sellers marginally more than new sellers, but does not regard negative feedback as important. Impressively, they take pains to identify several factors that even their disciplined experiment fails to account for, including repeat customers and multiple purchases.
The result is strong, but fairly limited and I wonder the limits on its importance and/or consequences. It seems the problem is eBay's highly limited feedback mechanism that either 1. causes people to be too "afraid" to give negative feedback or 2. serves a marketplace with very few dishonest sellers. Perhaps one should investigate a "dirtier" marketplace to get better data on reputation. I bet drug dealers, fake ID manufacturers, and fleshmongers might have excellent data to provide, but I imagine it's fairly hard to collect properly.
Peter Blair
In this paper the authors invesitage the effect of a seller's online reputation on the perceived risk that a potential buyer observes in doing business with the seller. Instinctually, one expects that a seller with a strong reputation will be seen as a lower risk for a buyer andhence can fetch a premium for her strong reputation. Conversely, one would expect that sellers with a relatively thin reputation would be viewed with more suspicion as a potentially higher risk. Behavior of this type would be consistent with the reality that online cheats can distance themselves from a poor reputation by simply creating a new selling identity -- in equiibrium the market would pay a premium to sellers with long reputations and conversley discount price of unknown sellers as a type of moral hazard. To authors setup a real life experiment selling vintage post cards on ebay to test this hypothesis: 1) sellers with strong reputations get paid a premium 2) sellers will incur decrease in profits that is postiviely correlated with negative feedback. The authors go to great lengths to avoid the perception of an experiment, to portray a similar level of professionalism between the types of sellers in the experiment (sellers with strong reputations and new sellers) and guard against other confounding problems that could lead to omitted variable bias. The result is that that the first hypothesis is confirmed, with buyers willing to pay a premium of 8.1% to sellers with stronger reputations. The second hypothesis, however is not confirmed, perhaps due to the fact that buyers do not bother to check negative feedback, as suggested by the author. I would also add that for low cost items (in this case < $14.99), the risks are so low that it seems reasonable that the penalty for being a mediocre seller is unobservable. Future research directions could be to see how these finding hold when we consider items that have higher sale value and hence greater inherent risk for dealing with an unknown seller. It mightalso be meaningful to distinguish between different type of items -- items such as gold coins and golf clubs are a lot less likely to disapoint since there are clear expectations of what these items are and what they can do, where as with consumer electronics and clothes there seems to be a greater likelihood for mismatches in expectations. One example fo this in in the MA and MA2 experiments that find that negative feedback had no impact on the sale of gold coins. It might also be important to consider whether or ot the items being sold is a bargain relative to its cost when purchased brand new -- a buyer getting a deal may be more likely to disregard or disunt negative feedback and perhaps there is a great sentiment that on ebay one is getting a deal so the risk of buying from a user with negative feedback is supressed by virtue of eBay's struture. At any rate, < 2% of feedback on eBay is negative, so the general perception is that eBay is a wholesome community of buyers and sellers which means that postive feedback will come at a premium and negative feedback is so rare that it's assumed to be close to non-existent. By analogy buying and selling on eBay is like living in a safe environment -- there is little worry or fear about crime and a great emphasis on associating with people in the community who are great people! One finaly comment a better way of observing a price premium for reputation might be to look at the mean of the initial valuation of the object by all bidders, i.e. if one had access to the maximum bid that each new bidder placed this would be a good estimate of how valuable the item is to the potential seller. The final price of an item seems to have both this information in addition to carrying with it the details of the auction, such as the number of bidders and how the auction plays out in real time. It could be that what we want is a measure of how a buyer interprets risk based on reputation in an environment where he/she has an opportunity to reflect on the previous transactions of the seller. Alternatively one can argue that what we really want is a measure of the buyer's trade off between risk and expected return (purchasing a quality item) in a real time setting.
Rory Kulz
So, first, I found it interesting that people on eBay don't bother to
click through to the detailed feedback. tThis surprised me a little
since this is always what I have done every single time I've used eBay
-- look at the negatives in the detailed feedback (the positives are
rarely useful since 99% are just "A+++++") and see how serious they
are.
Second, it's good to see that the reputation system ends up rewarding
those overall with a higher reputation, but eBay's reputation system
has some serious deficits. Some of these are difficult to avoid for
general systems, like collusion and fake bidders, but others are eBay
specific. For one thing, feedback as seller and feedback as buyer are
combined, which allows someone to come on eBay, buy a lot of items,
then seem to be reputable, when in reality, they have joined with the
sole purpose of pedalling poor wares.
As an example, a classic eBay scam is the listing that purports to
offer a free X, with small-print terms that say they are offering
"documents about where to find a free X." For example, X could be
"videos online" and the seller is really just providing a textfile
that says "YouTube.com, Hulu.com, ..." If you go on eBay and look at a
few of these -- among them I found free credit card processing, free
online pornography, free mp3 players -- they all involve sellers with
over
But yet it's obvious from their listings that they are scammers, so a
better designed reputation (it is not clear that this is possible, but
it certainly seems so) should be expected to be more robust against
these sorts of attacks. I would also expect that a better reputation
system would influence market dynamics more substantially than eBay's
did in this paper.
Michael Aubourg
First, it is true that we should expect that buyers will not provide information to help determine seller reputations, since to do so incurs a cost, and free riding is hard to
punish. However, Ebay should reward people who give information about seller reputations. The more information they give, the higher the reward should be, providing it is capped.
The fact that buyers ans sellers can communicate thanks to a parallel way, which are the email, turn the estimation problem much harder.
The researcher in this paper implemented the correct experiment since they varied the seller’s reputation without varying his responsiveness to emails from potential
bidders, his skill at listing items or the quality of his goods, which was necessary so as to obtain reliable results.
Then, we see that the negative comment play an important role in Ebay. For that reason I think it should be treated in a more precise way :
-1/0/1 is a very short choice array. We should either enlarge this scale (to distinguish late shipping from no shipping at all) or implement a small satisfaction poll system. Instead of allowing the buyer to write a sentence as a comment, they should answer a small poll about their satisfaction. Hence, buyers could understand what is wrong in the seller's reputation : It is easier to see a poll's result instead of looking for all previous comments, posted in historic.
Then, if a seller's problem is only slow shipping, a buyer who doesn't care at all about shipping date shouldn't change his trust in that person. However now, it is not the case.
Furthermore, what if I am a new buyer, in Ebay and I don't really know what I should reply as a satisfaction grade. Let's say that I grade a seller with -1.
This has a strong consequence if the seller is new. Here is my point : Ebay should weight differently, people with a different seniority.
In addition, another feature I would have add to the reputation system is a "truth vote". This feature should allow anybody who see a selling article to rate/grade it.
This sounds maybe weird at first sight, but it does make sense for several reason :
- Let's imagine I am a professional in a special field. I am looking for a special kind of object. Then I find a selling object which is obviously a fake one, or even worse, a fake selling offer,
then, I won't even bid on it. But no one would know what I think about it. So I think it is important to take it into account.
-Now, imagine people are suspicious about a selling offer. I am a professional, and I know that the reported quality is accurate because I sold it before for instance... I should be able to report my truth, even if I am not interested in buying it.
Hence, since people are known under their identity in Ebay, it is quiet easy to avoid multi vote. Furthermore, we could think about a reward system which would reward people who give their advice about selling offer. (but the reward should be independent from their reputation. For instance, they could have small fee discount.)
Finally, it is surprising that the feedback is not weighted by price of the transaction.
What if a problem occurred about a postcard ? It is not a big deal. The buyer will lose a couple of dollars in the worst case.
What if it is a car, or an excellent electric guitar which sells ? Now it is a big deal, and the problem concerns thousands of dollar.
Hence, it is really crazy to consider a
If I want to improve my reputation : This is the trick : I will sell many very low-price objects, and will deliver/ship/provide informations properly, in order to receive good feedback.
As soon as I receive enough good feedback, I will then be
able to sell more expensive objects. This is what I personally did. And this is
supported by a previous research led by people from the
For that reason, even if common people, and researchers trust Ebay reputation system, I am rather suspicious. I always try to figure out the amount of the transactions which led to a positive feedback.
Ziyad Aljarboua
This paper shows how sellers' reputation matters for buyers. It shows the
importance of reputation and buyer feedback for online markets as this is the
only way for buyers to verify the product they are about to buy. The reputation
system in websites such as eBay provides an alternative for the physical
inspection of products by presenting the behavior of sellers over the past
sales and buyer's satisfaction.
In this 12-week experiment, several seller accounts were used where one had a
high reputation and large number of past sales and the rest were relatively new
account with no feedback. Matched pairs of auction lots were sold through the
accounts for same price and shipping cost.
As hypothesized, it was observed that buyers are more likely to buy form sellers
with strong reputation with higher price for item than sellers with no
reputation who offer the same item for less. Also, negative feedback was found
to impact new sellers with no or weak reputation. Sellers with negative
feedback are more likely to make lower profits that similar new sellers with no
negative feedback.
While these results seem reasonable, findings of such study are likely to vary
depending on how the experiment is carried out. Many factors could affect such
an experiment. First are the items being auctioned out. The fact that this
experiment only sold vintage postcards restricts the audience (possible buyers)
to those who are looking for such items and yields a less comprehensive study.
Other factors such as the timing of the experiment might significantly affect
the study. Also, other sellers who are not part of the experiment and selling
same items would directly affect the study as they are competing for same
buyers.
Nicholas Wells
This study finds that eBays reputation scores are important in getting a higher price on goods. This makes sense in light of how a market works. Customers should be willing to pay a premium on an increased confidence of delivery on quality goods. Reputation is one way of indicating ones quality as a seller. The dynamic is different on the internet because, as the paper points out, reputation is now common knowledge. One interesting thing about this is that now reputation may literally be the only thing differentiating different sellers and their product. That being the case, of course it makes sense that reputation will be a factor.
I looked up Yhprums Law and Wikipedia notes that sometimes systems that should not work, work nevertheless (Resnick et al. 2004 regarding eBay). This is the opposite of Murphys law. The idea here is that there is no incentive for anyone to participate in the Feedback Forum in an altruistic way.
Subhash Arja
This paper conducts a controlled experiment on the reputation of the on-line auction site eBay. The authors sought to check how much more a customer is willing to pay to buy from a seller with a STRONG reputation versus a seller that was relatively new to the system. The result of this experiment would be very useful to the eBay community as well as eBay itself, so that it may consider alternative incentive models to remove any inefficiencies in the system. The second hypothesis the authors want to test is the theory that new sellers with negative feedback will obtain lower profits than those without negative feedback. The authors make it a point to mention that, since this is a controlled experiment, they eliminate the effects of email communication as well as investigate the impact of reputation for non-standardized goods. Overall, I thought that the experiment model was sound, but, while the authors were surprised by the results, I was less shocked.
The authors do find that buyers are willing to spend on average of 8.1% more to buy from someone with an established STRONG reputation as opposed to a seller that is relatively new. I found this a natural result since I have employed the same thought process when purchasing from sites such as Amazon and Half.com. I am willing to pay a little more to a seller that has a higher rating as a result of many reviews versus one that has a high rating with much fewer reviews. The authors were surprised with the results of the second experiment after finding that negative feedback had little to no impact on the buyers. I thought this occurred because if a person is new and has negative feedback, a buyer will give the seller the benefit of the doubt in many cases. Also, the type of negative feedback varies, and buyers can discern trivial complaints such as "shipping was a day late" versus "item never arrived". Therefore, the authors would have obtained more detailed results if they had done a more detailed study on the length and degree of negative feedback. Another project idea that is an extension of this experiment is studying the effect of seller presentation of the item. This study would be conducted on sellers' webpages for the product and how this impacts the chances of purchase.
Malvika Rao
I found the paper "The social cost of cheap pseudonyms"
to be very
interesting. It is a brave attempt
to find a solution to the problem of
multiple and new identities on the
internet without compromising the
flexibility and accessibility of
online systems.
Yet the solutions proposed appear to be "patch" solutions
rather than
solutions that are more fundamental
to the nature of the problem.
Admittedly this is a very difficult problem.
For example, in the case of once-in-a-lifetime identifiers,
how is an
intermediary selected to guarantee
the integrity of the intermediary? What
are the incentives for this
intermediary to perform their job correctly?
This also discourages people from having multiple identities
where each
identity plays cooperatively and
always behaves towards the social good.
Why should that be punished?
The "pay your dues" (PYD) model seems a better bet.
While it does lead to
some inefficiencies it seems to be
a more natural mechanism that meshes
well with the philosophy of the
internet.
It is interesting to try to think of a mechanism where
subsequent
identifier registrations are costly
but not punished unless a deviation
occurs. unfortunately
there appears to be no natural and "automatic" way
of differentiating between the
first registration and subsequent
registrations.
The paper "The value of reputation on eBay" reveals
that established
identities fared better than new
seller identities. This is unsurprising
and presents a natural incentive
for sellers to keep their identities in
the long-run.