4 ms·

...at least until review factories start producing two-star reviews describing tradeoffs that would seem reasonable to most people.

by cs702 7y ago

jperras 7y ago
Considering that most ranking algorithms don't a priori give more weight to lower ratings, this is probably fine.
CamelCaseName 7y ago
When it comes to online marketplaces, this is an often repeated idea, but absolutely nonsensical for two reasons: 1. The number of people who read reviews are a small, small, fraction of total buyers. Most people look at review count and the star rating (not the number, the image). But more importantly, 2. Ranking in search results is the #1 factor in driving sales. Everything else is a distant #3. Search engines will reduce rankings of products with negative reviews. (This my anecdotal experience after 2+ years of being an Amazon 3P seller)
- cs702 7y ago
  It's not so simple, because the machines that rank products and serve search results are evaluating the reviews too. As those machines look for signals of "authenticity," review factories learn to manufacture reviews that fake those signals. If adding a small mix of two-star reviews helps with search rankings, the review factories will start doing it in a heartbeat.
thebigshane 7y ago
describing tradeoffs that would seem reasonable to most people Sounds like "Mission Accomplished" to me! https://xkcd.com/810/ https://xkcd.com/810/ (Transcript...) Cueball: Spammers are breaking traditional Captchas with AI, so I've built a new system. It asks users to rate a slate of comments as "Constructive" or "Not constructive." Cueball: Then it has them reply with comments of their own, which are later rated by other users. Megan: But what will you do when spammers train their bots to make automated constructive and helpful comments? Cueball: Mission. Fucking. Accomplished. (from https://www.explainxkcd.com/wiki/index.php/810:_Constructive https://www.explainxkcd.com/wiki/index.php/810:_Constructive)
- dredmorbius 7y ago
  I increasingly see that as one of the most misguided xkcd strips ever. Unfortunately, the endgame for bot spam indistinguishable from humans isn't some benign silicon intelligence, but an epistemic war for attention, spending, manipulation, and/or fraud, benefitting deep pockets and highly motivated parties. It ends poorly.

...at least until review factories start producing two-star reviews describing tradeoffs that would seem reasonable to most people.

by cs702 7y ago

...at least until review factories start producing two-star reviews describing tradeoffs that would seem reasonable to most people.

jperras 7y ago

Considering that most ranking algorithms don't a priori give more weight to lower ratings, this is probably fine.

CamelCaseName 7y ago

When it comes to online marketplaces, this is an often repeated idea, but absolutely nonsensical for two reasons: 1. The number of people who read reviews are a small, small, fraction of total buyers. Most people look at review count and the star rating (not the number, the image). But more importantly, 2. Ranking in search results is the #1 factor in driving sales. Everything else is a distant #3. Search engines will reduce rankings of products with negative reviews. (This my anecdotal experience after 2+ years of being an Amazon 3P seller)

cs702 7y ago
It's not so simple, because the machines that rank products and serve search results are evaluating the reviews too. As those machines look for signals of "authenticity," review factories learn to manufacture reviews that fake those signals. If adding a small mix of two-star reviews helps with search rankings, the review factories will start doing it in a heartbeat.

thebigshane 7y ago

describing tradeoffs that would seem reasonable to most people Sounds like "Mission Accomplished" to me! https://xkcd.com/810/ https://xkcd.com/810/ (Transcript...) Cueball: Spammers are breaking traditional Captchas with AI, so I've built a new system. It asks users to rate a slate of comments as "Constructive" or "Not constructive." Cueball: Then it has them reply with comments of their own, which are later rated by other users. Megan: But what will you do when spammers train their bots to make automated constructive and helpful comments? Cueball: Mission. Fucking. Accomplished. (from https://www.explainxkcd.com/wiki/index.php/810:_Constructive https://www.explainxkcd.com/wiki/index.php/810:_Constructive)

dredmorbius 7y ago
I increasingly see that as one of the most misguided xkcd strips ever. Unfortunately, the endgame for bot spam indistinguishable from humans isn't some benign silicon intelligence, but an epistemic war for attention, spending, manipulation, and/or fraud, benefitting deep pockets and highly motivated parties. It ends poorly.