Stop Irrational Information Cascades
As the information technology grows rapidly, there are more and more information cascades around us. However, I think it’s not very difficult to stop them, just as we mentioned in class “Cascades are fragile”. I recently read an article “Arresting irrational information cascades”, and there are four strategies presented:
- You can try to apply agreement theorems more faithfully, by ensuring two people discussing their view update both up and down, rather than just copying.
- You could stop the first few people from forming incorrect opinions, or sharing conclusions without being quite confident they are correct.
- You could take theorems with a grain of salt, and make sure at least some people in a group refuse to update their beliefs without looking into the weight of evidence that backs them up, and sounding the alarm for everyone else if it is weak.
- If you are part of a group of people trying to follow AAT, you could all unlearn the natural habit of being more confident about an idea just because many people express it.
I read an old blog post in 2012 about the effect of fake reviews on Yelp. The author pointed out that a good strategy to remain reputation for Yelp is to catch those fake reviews. When I read some reviews on Yelp or Trip advisor, I always doubt the truth of these reviews, because I know that some restaurant will give you a special offer if you write a good review for them.
However, I recently read an experiment about reviews on Yelp and I am surprised about the result. This experiment gathered data from Yelp reviews and used two models to train them, one is an information cascade model and the other one is a random Bernoulli model. The result surprised me that for 75% of the restaurants, the random model is indeed a better model than he cascade model. This suggests that Yelp reviews in many cases may not be influenced by previous reviews, and in fact represent independent observations of the truth of a restaurant. As I relate this result to the article “Arresting irrational information cascades”, I think the result of Yelp experiment makes sense to me. The article “Arresting irrational information cascades” mentioned four strategies to stop information cascades, and I agree the most with the third strategy “at least some people in a group refuse to update their beliefs without looking into the weight of evidence that backs them up, and sounding the alarm for everyone else if it is”. I think in real life many people will update their information completely. They not only consider the information about its goodness or badness, they also consider what is the probability that someone wouldn’t know about the information beforehand. When someone is writing a review for a restaurant on Yelp, he not only receives the information from previous reviews, but also some private signals that come from his own experience with that restaurant. In this case, even though the majority of signals are high, the person might still write a negative review for this restaurant because his private low signals balance the high signals and he can make decision on his own judgment.
I personally think that sometimes if people can update their information rationally and completely, the negative effects that information cascades bring will decrease. The reason why some social media and online review websites still have really high reputation is that there are not too many cascades exist and the reviews are still likely trustworthy.
References:
- https://blogs.cornell.edu/info2040/2012/10/30/the-effects-of-fake-online-reviews-on-todays-information-cascade/
- http://www.overcomingbias.com/2012/10/arresting-irrational-information-cascades.html
- http://snap.stanford.edu/class/cs224w-2011/proj/emmao_Finalwriteup_v1.pdf