4 ms·

In my mind, focusing on bots is kind of barking up the wrong tree. A lot of fake reviews on Amazon are written by real humans -- my main concern is not figuring

In my mind, focusing on bots is kind of barking up the wrong tree. A lot of fake reviews on Amazon are written by real humans -- my main concern is not figuring out whether a human or a bot wrote a comment, it's figuring out whether or not the comment/review is trustworthy. If we got rid of all of the bots, that wouldn't make Amazon easier for me to use. I feel like scammers have already figured out that humans themselves are relatively cheap to buy.

thejohnconway 7y ago
The article doesn’t focus on bots ( in fact it specifically mentions times when humans pretend to be bots). It’s really about internet fakery in general eroding trust.
- danShumway 7y ago
  Yep, just expanding on the premise of the article and reinforcing that this is a general trust problem, since a large portion of the current conversation around internet trust does revolve around bots.
jmoss20 7y ago
Right -- it's ultimately a trust problem, and we're in an unfortunate negative feedback loop here. Bots aside, what we post on the internet (and believe in general) is a function of the information we consume. That information is increasingly consumed through the internet, and the internet (as typically used) is feeding us increasingly poor quality information. (To be consumed, then shared. Recurse...) We can (and I think do) cope by having different levels of trust for different sorts of internet information. I generally have higher trust in an HN post than an Amazon review, for example. This is useful, but has a dark side: some of the things I (and likely, you) take as a sign to "increase trust" happen to correlate with things that have nothing to do with trust. I trust Amazon reviews more if they have good grammar and spelling, largely because I trust general internet information more when it has good grammar and spelling. But perhaps I shouldn't when e.g. buying a screwdriver (what does writing ability -- and the things it correlates with -- have to do with evaluating a screwdriver?). Coming up with social/political-flavored-information examples is an interesting (and worrying) exercise.
- acid_burn 7y ago
  I find it interesting that when something uses proper grammar and spelling, it may appear more trustworthy. Consider this: a consumer that employs above average grammar and spelling skills to write product reviews may also be more skillful in forming and expressing valuable opinions and assessments of any product. I think that people are naturally inclined to pick up on those signals, whether the correlation is real or not.
  - jmoss20 7y ago
    Even if the correlation is real, what I'm getting at is that it isn't causal. Good grammar probably makes for a good bet on more trustworthiness, but I'm wondering when/how that bet is going to be wrong (and it is, if it's not causal), and how that effects us (on the massive scale that is our info consumption across the entire internet). back to the Amazon example: an expert car mechanic may have worse spelling/grammar than a hobbyist, but I should probably trust their Amazon review of a car part more.
  - 8bitchemistry 7y ago
    Not only that: the exact same written text will appear more trustworthy depending on the typeface [1] (eg Baskerville > Helvetica > Comic Sans) [1] https://opinionator.blogs.nytimes.com/2012/08/08/hear-all-ye-people-hearken-o-earth/ https://opinionator.blogs.nytimes.com/2012/08/08/hear-all-ye...
  - jjwhitaker 7y ago
    There is a component to terribly written phishing emails where the poor spelling and grammar becomes a filter for people who won't fall for it. People who will fall for either won't care or aren't paying attention anyway.
Proven 7y ago
Let me point out that fake review posters that benefit from this (otherwise they wouldn't be doing it) are low income people that other commenters claim are most hurt by low trust internet.
wincy 7y ago
Amazon is weird, I bought a product in the mail the other day and they offered me a $20 coupon (the product only cost $25) if I wrote a five star review. So I wrote a one star review talking about how I don’t want to be bribed and they can’t make me be dishonest and to warn people. But amazon didn’t allow my review. So that sucks, and makes me not trust any reviews on Amazon.
- didibus 7y ago
  I mean, how could they trust you weren't a fraudster from a competing product?
  - pixl97 7y ago
    Because they actually purchased the product?
    - didibus 7y ago
      I don't think that's a great deterrent. Especially for a low cost item, I'm pretty sure it happens a lot that competitors would buy a few products of yours and then leave negative reviews as "Verified Purchase".
      - chongli 7y ago
        Amazon is not a neutral referee though. They have an incentive to block (both fraudulent and legitimate) one star reviews but keep (even fraudulent) five star reviews: better reviews lead to more sales.
- r00fus 7y ago
  Guessing that Amazon has some way for the seller to dispute the review, or they use some heuristics to determine that a comment is “on topic”.

thejohnconway 7y ago

The article doesn’t focus on bots ( in fact it specifically mentions times when humans pretend to be bots). It’s really about internet fakery in general eroding trust.

danShumway 7y ago
Yep, just expanding on the premise of the article and reinforcing that this is a general trust problem, since a large portion of the current conversation around internet trust does revolve around bots.

jmoss20 7y ago

Right -- it's ultimately a trust problem, and we're in an unfortunate negative feedback loop here. Bots aside, what we post on the internet (and believe in general) is a function of the information we consume. That information is increasingly consumed through the internet, and the internet (as typically used) is feeding us increasingly poor quality information. (To be consumed, then shared. Recurse...) We can (and I think do) cope by having different levels of trust for different sorts of internet information. I generally have higher trust in an HN post than an Amazon review, for example. This is useful, but has a dark side: some of the things I (and likely, you) take as a sign to "increase trust" happen to correlate with things that have nothing to do with trust. I trust Amazon reviews more if they have good grammar and spelling, largely because I trust general internet information more when it has good grammar and spelling. But perhaps I shouldn't when e.g. buying a screwdriver (what does writing ability -- and the things it correlates with -- have to do with evaluating a screwdriver?). Coming up with social/political-flavored-information examples is an interesting (and worrying) exercise.

acid_burn 7y ago
I find it interesting that when something uses proper grammar and spelling, it may appear more trustworthy. Consider this: a consumer that employs above average grammar and spelling skills to write product reviews may also be more skillful in forming and expressing valuable opinions and assessments of any product. I think that people are naturally inclined to pick up on those signals, whether the correlation is real or not.
- jmoss20 7y ago
  Even if the correlation is real, what I'm getting at is that it isn't causal. Good grammar probably makes for a good bet on more trustworthiness, but I'm wondering when/how that bet is going to be wrong (and it is, if it's not causal), and how that effects us (on the massive scale that is our info consumption across the entire internet). back to the Amazon example: an expert car mechanic may have worse spelling/grammar than a hobbyist, but I should probably trust their Amazon review of a car part more.
- 8bitchemistry 7y ago
  Not only that: the exact same written text will appear more trustworthy depending on the typeface [1] (eg Baskerville > Helvetica > Comic Sans) [1] https://opinionator.blogs.nytimes.com/2012/08/08/hear-all-ye-people-hearken-o-earth/ https://opinionator.blogs.nytimes.com/2012/08/08/hear-all-ye...
- jjwhitaker 7y ago
  There is a component to terribly written phishing emails where the poor spelling and grammar becomes a filter for people who won't fall for it. People who will fall for either won't care or aren't paying attention anyway.

Proven 7y ago

Let me point out that fake review posters that benefit from this (otherwise they wouldn't be doing it) are low income people that other commenters claim are most hurt by low trust internet.

wincy 7y ago

Amazon is weird, I bought a product in the mail the other day and they offered me a $20 coupon (the product only cost $25) if I wrote a five star review. So I wrote a one star review talking about how I don’t want to be bribed and they can’t make me be dishonest and to warn people. But amazon didn’t allow my review. So that sucks, and makes me not trust any reviews on Amazon.

didibus 7y ago
I mean, how could they trust you weren't a fraudster from a competing product?
- pixl97 7y ago
  Because they actually purchased the product?
  - didibus 7y ago
    I don't think that's a great deterrent. Especially for a low cost item, I'm pretty sure it happens a lot that competitors would buy a few products of yours and then leave negative reviews as "Verified Purchase".
    - chongli 7y ago
      Amazon is not a neutral referee though. They have an incentive to block (both fraudulent and legitimate) one star reviews but keep (even fraudulent) five star reviews: better reviews lead to more sales.
r00fus 7y ago
Guessing that Amazon has some way for the seller to dispute the review, or they use some heuristics to determine that a comment is “on topic”.