>The company has said that most attempts at review fraud are committed by a small minority of hotel and restaurant owners trying to boost their profile unfairly rather than knock another property.
>Anyone can post a review on TripAdvisor, Yelp, and Minube, whereas a guest may only post a review of a hotel on online travel agency platforms if the person actually books at least one night at the hotel through the agency.
>In 2012, TripAdvisor was scolded by a UK watchdog group, the Advertising Standards Authority, for using promotional wording on its site saying that it offered “reviews you can trust… from real travellers… and trusted advice from real travellers.” The group said the company couldn’t back up the claim by proving that all of the reviews were genuine and from real travelers. However, strictly speaking, no system, based on verified bookers or not, could meet that standard.
How valuable is a review? What does it contrast to? Some imaginary background or average experience? It is hard to design a system such that there are no fake reviews. The fake reviews can take different forms: actual customers who were "bribed" (or is that just a discount?); paid reviewers did physically visit but who were bribed, possibly posting under false names, or simply fake reviews written by people who were never accomodated.
Here is my probably brainfart idea which tries to solve both problems at once:
HostCompare may be implemented centralized or decentralized, so I will describe the decentralized version, such that a centralized version can cull away some of the crypto (but then bribing central HostCompare is a vulnerabillity).
HostCompare: instead of reviews, customers booking accomodation in a city through hostcompare, will spend half of their visit at one accomodation, and half at another. The customer can navigate the comparisons between different places, and for each pair already tested view all the facets for which place A is better ">" or worse "<" than place B.
In exchange for the tourist traffic forwarded HostCompare demands a certain percentage of their income. Each host is free to generate its own keypair, and when the tourist arrives shows his passport/ID/... and the host signs the hash of: the name of the visitor and the Comparison's unique ID, and publishes it. This means that you can't create a positive review for your place without also generating customer traffic to your competition and convincing a tourist to co-operate. The comparison is only written once the consecutive stay at both hosts is over.
EDIT: I was going to write my idea more in-depth, including how HostCompare would use part of the profit to encourage (by making cheaper through discount) unexplored comparisons. Also finding cycles in the directed graph. Also how a host bribing a HostCompare tourist would only work if this bribe is offered indefinitely to HostCompare tourists. Furthermore it is more efficient to simply list a lower price at HostCompare since then you are paying less towards HostCommpare. The bribe is effectively a means to be secretly cheaper than listed. Comparisons are also better than reviews since the host can think about how to improve the situation, since it is in fact a comparison instead of a review in vacuum, or a review in comparison with vague memory of travels past, or a review in comparison with some unformalized expectation.