Yeah, from the linked paper it looks like 20% is only the percentage of fakes that are statistically obvious.
Other people read the good reviews and determine to make their decision from that.
I read the bad reviews. 1 to 3 star reviews. I'd rather go to a place with a bifrication of bad and good reviews where the bad reviews are stuff I don't care about...
My theory is that if a place sucks it will definitely have bad reviews. A place that is great will still have a few, but they will be over stupid stuff (i.e. I expected something different etc.)
Its the reviewers who need to be reviewed. Or, in a true social sense trusted reviewers should be followed and the rest ignored.
Who can you trust when even trust is gamed?
If they add up, I'll take note. If they don't, I'll presume a fake.
The failure of review services (Amazon, Yelp, and others) to police their reviews adequately really hurts these services. Otherwise they're little more than a yellow pages or directory.