How can you trust recommendations you read on the web? We expose the sock puppets!

 -19-

The web is a great free-for-all that means you have to take care what you believe, but some things are designed to have a value only if they are trustworthy.  One of these things is the humble review - a literary form that goes back probably to before movable type was invented and perhaps even to the first alphabets and the invention of writing. After all what could be more useful to write down than a recommendation of what is the best or a scathing attack on the worst - this is indeed valuable writing. Of course the trouble is that a review only has true value if it can be trusted and this is where the openness of the web brings yet another problem.

What could be easier for an online retailer or bookseller (you can tell I have Amazon in mind but the comments apply widely) than to add the facility to allow customers to comment. Of course the hope is that the customers will write glowing reviews that encourage other users to buy the same thing but, as long as the retailer's inventory is large enough, negative criticisms only serve to move the flock to another pasture. Negative criticisms also mean that users are forewarned about any potential problems and over all this should increase customer satisfaction. If you buy a technical book after a reader has supplied a review which say "it doesn't cover X" then you are far less likely to return the book because you thought it did indeed cover the subject in question.

A small amount of policing to remove obvious crazy or unbalanced submissions can be relied upon from other users - just give then a "report this" button or a "comment" facility and the whole thing is more or less self maintaining.  Overall retailers see customer reviews as generally good even if they are bad!

Now we come to the other side of the transaction - the potential buyer and at this point I'm going to consider the field I know something about, computer books, but the comments remain valid in the broader context. You need to believe the review for it to be of any use to you.

The problem in any small specialized and personalized area such as books is the sock puppet. If you had just had a book published wouldn't you get your friends and colleagues to submit favourable reviews of it? I know I would! It could even be justified on the grounds that one or two glowing reviews will balance the damaging reviews submitted by your personal enemies in order to "bring you down a peg or two". As it's all anonymous, only a valid email address is needed, it is all very easy.

Sock puppets are usually easy to spot they usually write a glowing review - "the best book I've ever read" - and then go on to urge you to buy it. If they are a little more sophisticated they throw in a minor criticism that is designed to throw you off the scent but not the book - "didn't like the cover photo" - or best of all something that is veiled praise such as "the book was far too short". If they are even more sophisticated then they add a review of one or two other often completely unconnected items - often a long review of the book in question and say a two-line review of a particularly fine razor blade refill or a water filter jug.

Sometimes sock puppets are this easy to spot but they can be as subtle as human ingenuity allows and you can never be sure. So what about instilling trust by a coherent, reasoned review? If the reviewer explains why the book is good rather than just saying it's good then this can instill trust. However most sock puppets don't manage to do this and there isn't a tradition of criticism in most areas of book reviewing, apart from literary works where the review is almost an art form in its own right.

To return to the general situation, the community review is mostly a waste of time. The reason is that you can't tell the real reviews from the biased and simply dishonest ones. Is there any way that this situation can be fixed? There have been many attempts to make the community review more useful the latest is Amazon's attempt to qualify a review with an assurance that the reviewer did indeed purchase a copy of the item from Amazon but this just sets the minimum price for an "arranged" review. Of course in the old days when editors edited and magazines published reviews it was easier to believe that the opinions were unbiased and unmotivated by other concerns. Of course you might well not agree with the reviewers but at least it was a good assumption that they weren't the authors' best friends. Can we recover this model in some new implementation? Probably not because readers generally aren't prepared to pay for what they can get for free - only we know that they are just deluding themselves.

Here at iProgrammer we offer a small solution at least as far as computer-related books are concerned. We can't say that it is foolproof as we can't review every single book on the subject - but at least we can follow the creed of complete impartiality and give you honest reviews.