The Reliability of User Reviews

User generated reviews are one of the great innovations that came with the growth of the Internet. They are now pervasive, and cover everything from automobiles, to music,  books, movies, and restaurants. I suspect the arrival of user reviews on interactive mobile platforms (such as urbanspoon and amazon) will bring upon us a new wave of interest in such content.

Yet, user-generated reviews suffer from informational problems. Firstly, why would you trust the product recommendations from an online stranger any more than you might somebody else? Secondly, user reviews are often tediously long, contain huge volumes of inconsistent information, and sometimes even degenerate into personal mudslinging matches. This imposes search costs upon the person trying to make sense of reviews. For example I was recently searching for a new lens for my SLR camera (a little hobby on the side), and it took a bit more time than I had expected to visit various photography forums to sort out which products were really good, versus other lenses that suffered quality control problems. Thirdly, there is a growing phenomenon of companies manipulating online information for their own benefit. For instance, last week Seagate was found to be deleting user postings from their website about high defect rates, while a Belkin official was caught out offering cash for good reviews.

The way web operators try to address these problems  is by allowing individual reviewers to build reputations. So, for example, on Amazon, people can decide how many “stars” to award each review, and each reviewer’s profile is publicly viewable. On discussion forums such as slashdot, each participant is rated, and moreover people who post inaccurate comments are often ridiculed or otherwise reprimanded by other users. Visitors to such websites quickly learn to identify reliable contributors and give more attention to what they say relative to others.

Unfortunately, I believe this does not resolve the entire problem because the websites on which people purchase products are distinct from websites where they truly exhibit their skills and knowledge. For example, many people purchase camera equipment from adorama or amazon. But they post photos, critiques and tips for other users on websites like dpreview and photo.net. So, while someone may have build a tremendous reputation on a website like photo.net, this reputation does not “transfer” to their product recommendations on Amazon or adorama.

I believe this presents an interesting entrepreneurial opportunity for someone to offer user-accreditation across different websites. People would have a reason to sign up for such a service, both because it might offer a single login to multiple sites, but also because they don’t have to build their reputations each time they join a different site. Web operators would have a reason to sign up because they would have access to information about each reviewer/user that they would otherwise not be aware of (and this might increase the overall quality of reviews on their site).

What do you think? Is this a good solution, or are there better ones possible?

Author: kwanghui

http://kwanghui.com

11 thoughts on “The Reliability of User Reviews”

  1. I agree that this is a problem and that people are working on this. One intermediary step is the transfer of identity from one site to another. Facebook Connect is one example where people can comment on blogs while using their Facebook profiles. Not exactly what your post was about, but it’s still one step closer to reputation transfer.

    Like

  2. i agree with the writer s comments .but then again information or mis – information is not only restricted to the commercial world only .it exist in all facets of the world .
    . thats why people who control the information world has great control .

    Like

  3. Gwen beat me to the Facebook Connect angle, and there was also the Microsoft Passport program (now the much-changed Live ID) that tried something similar.

    However, they only pertain to identifying information, not meta-data like reputation (or further, credit-worthiness?), so there is plenty of scope for future development along these lines.

    I for one can’t wait until my thumbprint (or whatever) replaces all usernames and passwords, and I can have a single, unified identity across the Web, Grid, Cloud, or whichever name we attach to the electronic pseudo-reality to come.

    All I have left to mention is that the Belkin pay-for-reviews story is not confirmed. It could just be a more subtle version of that infamous “send on this email and Bill Gates will pay you $20” hoax.

    Like

  4. Yes, I think it’s a good solution. Generally we have had the era of firm reputation, and now we’re gradually realising that user reputation is important – and more to the point a project with which we have the tools to deal. Lenders do check out reputations of consumers – as to landlords, but this doesn’t happen much elsewhere that I can think of.

    Like

  5. Trust in reviews is a subset of the wider problem of trust in electronic identities. That is, if you have a trusted electronic identity you can build a reputation system for user reviews (or blog comments). It also means you can build reliable electronic trading systems. If you have wide spread identity fraud then your trading systems or reputation systems break down. A solution to reliable trusted electronic identification turns out to be giving people the ability to have their own secure, private, electronic identity that they themselves trust. If we are confident our own electronic identity is trustworthy then we may be willing to share information about ourselves with others (because we know that others cannot pretend to be us). As the same trustworthiness applies to others we can trust the identity of others who use the system.

    Once you have a trusted private electronic identity that you trust then you can attach any trusted attribute you like to it. It turns out there is a market for such a system and we have built one and are successfully commercialising it for use by organisations who need to identify people to meet anti money laundering regulations. Our system gives an individual a way to create an electronic identity that they can trust. Individuals trust it because they control access to it and control what they give out from it. This means we should not stop an individual using it for whatever purpose they like – if they are willing to pay for the integration. This means it is available to any entrepreneur who wants to build a reputation system and make it available to individuals. Perhaps an index of the “worthiness” of blog comments on CoreEcon might be a good place for someone to start?

    Like

  6. As the Customer Service Ambassador for Adorama Camera, I was interested to read your posting.

    My role involves actively trawling the Internet each day, seeking out and responding to postings relating to Adorama Camera.
    I have a complex system of alerts and RSS feeds set up, and my responses to customer queries can be found on over 400 websites, blogs, photo sharing forums and shopping sites.

    One recurring problem is the anonymity afforded by the Internet, which enables individuals to post grievances in the ‘heat of the moment’. These lead, frequently, to a flurry of similar complaints.

    Upon further investigation, it is not unusual for a number of these to be found to be complete mis-representation ie, actually relating another company entirely; misunderstanding by the customer as to the nature or function of a product, or a genuine error on the part of our buying department, our shipping department, the website team, our suppliers or distributors, the manufacturer etc – yet the reviewer has chosen to post negative feedback without giving us the opportunity to respond and put matters right.

    Many photographic and shopping sites enable a company the right of response on the site; others enable us to contact the customer directly via a link on the site. However, there are numerous sites where resellers are not enabled to either address negative feedback or to make contact with a disgruntled customer, so cannot correct errors or offer explanations.

    I don’t know what the answer is. There is no denying the power of the Internet; in careless hands, however, it can be extremely destructive. When we can all learn to use the Internet to address our potential ‘audience’ in the manner in which we would speak to them face-to-face, with honesty and integrity, it will become a more valuable tool for all concerned.

    Like

  7. Interesting discussion. Reminded me of this recent article in Slate: http://www.slate.com/id/2208676/

    I certainly think there is an opportunity here to create a tool that will enable users to create secure identities and histories (e.g. reviews of their reviews could be tagged to them etc). Another way to try and build more information into the identity would be to use some indicator of social distance that could be calculated using your online identity (e.g. LinkedIn path distance between i reviewer and j searcher). Along with the ‘histories’ strategy implicit in the secure identity strategy, social distance is a key way humans seek to measure trust offline and so it might be useful here as well.

    Like

  8. I’m wary of consolidated identity. A review site is my approach. Granted not many people come to my out of the way blog but I’m slowly building up a following. I review music and movies and post on the environment. Once people get a flavor to your integrity and taste it’s up to them regarding trust. On line strangers are what they are and you have to take them as such – an axe to grind, fraud, self promotion (that’s my biggest sin). Reviews on general blogs are hit and miss because as you stated they are random strangers. However I believe in sifting I steer away from the overly growing and actually read the 1 star ratings – what is it that people do not like gives me some insight into what the movie, book or music is about. As far as electronics and office equipment I just do my research. There is no easy way around it but I have been thorough and have had success relying on the “kindness of strangers” 🙂

    Like

  9. Thanks, everyone, for your thoughtful comments. A reader also emailed me the following link to a new tool and which might be useful for aggregating the information from different forum websites: http://debategraph.org/

    It might be useful for firms like Adorama trying to make sense of all the user comments being generated out there (thanks for the insights, Helen).

    I thought it’s quite an interesting angle, and worth sharing with you.

    Best regards
    kwang

    Like

  10. great post kwang. i too share your frustrations with review sites.
    just to expand on the points made by gwen and jarrah regarding facebook connect and single sign-in (i.e replacing all of those usernames and passwords with one unified system), i personally beleive that the future will lie in open standards across websites rather than a commercial offering (suggested by kevin). for thoese interested, i suggest checking out the open stack initiative (http://developer.yahoo.net/blog/archives/2008/12/the_open_stack.html) which addresses a lot of the problems raised in the original post.

    Like

Comments are closed.

%d bloggers like this: