Host Finding and the Super-Host Effect


There is a prevalence of super-hosts on the Couchsurfing™ platform. These are hosts which have far more references than other hosts in their area. While very active hosts can be extremely beneficial for the community, unfortunately when combined with the message restrictions (10 per week for 95% of users) and the host matching system, the effect is detrimental and can cause "couch surfing fatigue" for the super-host, while other hosts do not receive any requests. Here's how it works.

A surfer is looking for a host in a new city for two nights, but she is restricted to only sending 10 requests per week so might only send three or so messages to potential hosts. To maximize her chance of getting an offer to stay, she only messages hosts who are active and have a lot of references. Someone agrees to host her, great!

What happens here though is that it incentivizes surfers into only contacting well-established hosts, which in turn makes them even more established. New hosts find it harder to start hosting and hosts with fewer references drop out of the community as they get few requests. The host pool shrinks to just a small handful of super-hosts. Fewer hosts results in fewer surfers being able to use the platform. With fewer people in the community, there are also fewer organized events. In the worst cases, predatory hosts gain a foothold, reducing communities to only a small group of male hosts who almost exclusively host attractive women.

What's important to look at here is the host distribution in each city. In the super-host case, there are a small group of people who do most of the hosting. However, in a healthy community, the surfers should be distributed much more evenly between hosts, promoting more hosts and a larger active community. Couchsurfing™ has not done much to address this or incentivize healthy communities. One reason for this might be that addressing it would involve removing the message restrictions which is a key part of their revenue stream.

The host matching system also encourages the super-host effect. The current system is largely built on the number of reviews a person has, and despite this not being considered a good measure of trust or community standing, it is the only information that users have access to. Mix in a poorly built filter system, and that explains most of how surfers find hosts. If there were metrics other than reviews that users could use, it would mean that the number of reviews were less important and there would be a more even spread of hosting requests. Other metrics could be used such as shared values and community standing from a broader range of activities (hosting/surfing/organizing/hangouts etc.).

Our solution is a Community Standing score that counteracts the super-host effect and removing the message cap, reducing fatigue from current super-hosts and allowing more members to have healthy and safe couch surfing experiences.

Our plan to fix it: Better host finding