One thing I feel quite strongly about is scores ‘decaying’ over time. Maybe not completely, but say you take 0.5 as the middle, if someone with a high score doesn’t use the platform for ages, it’s should slowly go down to 0.6 or something. And likewise, someone with a very low score should slowly go up to 0.4 (though perhaps even more slowly).
This would the the calculated score, not the individual ratings.
Aaaand I should also start a new thread on the calculation of the community standing score 
Yep the score is between 0 and 100.
Very much agree. Also because over the course of years, people change.
On the topic of the dehumanising aspect of scoring, I think we should consider not using a star rating system. I think what might be better to see is a clear indication of categories for what the score means. Instead of 1-5 we can have ‘bad experience’ (lowers their score), ‘neutral experience’ (their score is not altered), ‘good experience’, and ‘great experience’. Also using this we can futher prompt ‘bad experience’ ratings to leave reports if necessary.
Agreed, I think we can have a filter (e.g. I only want to see people with scores 70+), but not a sort function. A separate conversation is how we design the host search function to order hosts on the screen. These can be simple things like randomising them or promoting less experienced but good standing hosts, to more complicated things like algorithms to predict good experiences based on profiles and references of the surfer and host.
I’m also going to suggest that somehow we don’t make the community standing score algorithm 100% transparent. The specific strategies of gamification in any environment are notoriously hard to predict, and I really worry about winding up with unintended consequences. I think it’s important to say what is factored into the score (e.g. reference scores, event participation), but not give the exact recipe of how they’re combined.
I also think it’s very important that verification isn’t included in community standing. It’s an important statistic in it’s own right, and shows whether someone is who they say they are, and that their account is duplicated. It’s an objective trust standard, distinct from community standing which is based on the subjective experiences of others.
My first instinct is that bad/neutral/good/great experience is too vague on which aspects of the experience you are giving feedback on. That would entirely invalidate the safety aspect of it.
Perhaps Yes/No options to pointed questions would remove the dehumanising scoring aspect while keeping it factual. I think many people have brought up examples where a host may make you feel uncomfortable, but not unsafe per se. If we’re not allowing for a degree/scoring for feedback on safety, then these questions would have to be separated. E.g.
If no: Would you like to report this user to the safety team? Your report will be completely anonymous and we will take appropriate action if required. Please provide as much detail about your encounter as possible.
If no: Please tell us more about your experience and why you felt uncomfortable. This is completely anonymous (your host/surfer will not see this).
Sorry, just meant for the score part instead of the 1-5 stars, which points to the general enjoyment of the experience. Definitely need these other yes/no questions separately.
Ah hmm. You’re referring to this right?
• I was satisfied with the interaction (including this so that people will be CLEAR on the kind of hosting/surfing experience they’re expecting during the requesting/messaging process)
Good point, I get what you mean. I have another concern regarding using Yes/No instead of a scale though. With displaying the number of Yeses/Nos on a profile, no matter regarding safety or any other aspects, it makes it easier for a creepo to sense who gave a Yes or No, because we’d have to display binary numbers. Or is there another way to show an average with Yes/No answers?
Yeah that’s what I meant.
I think we really want to limit which statistics we show to what’s really important. Where possible, we should include yes/no answers into community standing (e.g. safe/comfortable scores), and for ones we don’t want to include, we should think about how we want to include them. CS’s scores for ‘wants to hang out’ or ‘friendly’ weren’t really useful IMO so I think we should avoid this approach, especially if we think people are tracking who’s leaving responses on their profile. For that ‘wants to hang out’ stuff, is there some other way we can approach it that isn’t a score? For instance could we give prompts or even options for the messages that surfers send to the hosts?
Just off my mind right now, from a conversation I had with @lucas, I was thinking more of allowing the user to indicate if a) they’re accepting guests and b) if they are too busy to spend time with their surfer. Because this varies from period to period (e.g. I may be busy with a project this month but I’m still willing to host, but sometimes I’m not and want to spend time with my surfer), so it helps to set expectations BEFORE the interaction.
This will also (hopefully) form the basis of when they rate the interaction afterwards - on whether it was a pleasant or not. Of course I’m assuming people will use these “want to spend time together” filters during the search process, but this requires testing to prove.
The reason why I’m particularly wanting to address this mismatch of expectations is because in interviews with hosts, a few have mentioned that it’s off-putting when people treat their house like a hotel. Honestly if having a “wanna spend time together” filter solves the problem, then there’s no real need to “rate the experience” per se.
Yes, I agree, I don’t think this should be included at all. So the feedback system now would essentially encompass:
Does this make sense? Do you think there’s a need to rate the interaction in terms of pleasantness if we had a “wanna spend time together” filter?
Yeah I think this is where the ‘experience score’ part of the rating comes into it, it speaks to if your general enjoyment but also if it matched your expectations. And that will be more general than the few criteria we can come up with like ‘wants to hang out’ (not that it’s unimportant, just that there’s all kinds of ways the expectations can mismatch the experience)
That’s part of why I think we definitely need this general ‘experience score’, it will reflect whether people’s expectations match their experiences with a person.
I imagined these would be binary options. Do you think there’s scope here for a range this large to be meaningful? Is a 4 for safety or discomfort a bad score because it’s not 5?
How would you display the safety scores on a user’s profile then? 3 Yeses and 2 Nos? Or are you saying not to explicitly state it at all?
Yeah was imagining it would just be included into the score, and quite heavily too for Nos. Otherwise we’re going to have like 5 statistics that people have to look at.
I think this is really finetuning stuff though, and we won’t be able to really see what works until it’s tested in practice, at which point we can iterate. But generally I think start simple and expand from there.
I disagree with making a safety score implicit. It wouldn’t be useful for me as a vulnerable surfer when I look at profiles with similar community scores. It also won’t serve as proper, explicit feedback to users on what is affecting their score negatively, so that they can take the required action to improve it.
I’d like to hear more views from others for this!
Just wanted to chime in here because when CS did this, it made “super hosts” very upset. I get that we want to make it more fair, but it doesn’t seem fair when people who have lots of experience and positive feedback are randomly pushed to the bottom.
(I am going to start a thread on what I think would do that better - displaying an acceptance rate so that busier/more popular “super hosts“ may seem less appealing (or more appealing, depending on how up you are for a challenge). Hosts that get less requests would have a higher acceptance rate naturally 
How about making the “can host” and the “wants to meet” two different things?
In CS you could set your “status” as can host/ cannot host/ wants to meet, and in the search result, if you wanted to find locals to hang out but you didn´t need a host, all the people that had “hosting” on were filtered out. Wich is not accurate. I´m open to host, but I´m also open to go to a museum, so I will set up “wants to host”+“wants to meet (or hang out or whatever)”, and maybe some people are open to host, but don´t really want to hang out with their guest, or don´t have time, so “want to host”+“doesn´t want to hang out” would be clear and solve this two issues.
I would chang this to “please leave a note on the guestbook” or “please describe your experience”. For me references are way more important than any point system, or anything wirtten in the profile by the person. What others say about the person is what gives you the real idea of how the interaction is going to be. And also is what keeps the trust system running. Let´s please reinforce the idea of leaving a reference, not make it an additional nuance “just if you want to”. Please?
Very good point, it seems not just the anonymous component is important, the outward feedback component deserves just as much emphasis. Will definitely keep this in mind. Thank you for sharing your thoughts on this! 
I feel that there are many factors to why people may not want to come forward and call their harassers out publicly. A quick google search will help you understand why.
That’s what you’re talking about right? Because the only anonymous portion we’re proposing is the part where we ask members if they felt SAFE and COMFORTABLE (i.e. to check that life was not in danger, there was no unwanted sexual advances) throughout the encounter. References are still public. If you’re going to call victims cowards and “stabbing” their harassers in the back, then it’s literally victim shaming.
Hope you can clarify your stand on this, because I am not comfortable with having this kind of mentality on the platform.
I stand with Kelly on this, too. Platforms that have been developed later than Couchsurfing have benefited greatly from having an anonymous element to references. In a world where people are increasingly sensitive and egos are fragile, it’s not that easy to leave a candid negative reference without fear of being attacked or shamed.
I don’t think anonymous negative references are a wise idea. In the world we live and, particularly in Brazil nowadays, politics go hand in hand with fake news and “trollers” staining reputations through gossips and rumors is easy and it’s being widely used. If a guest didn’t like something about the host it could just use this unproven “reference” as a vendetta. It’s better to say it out in the open, so that the host has at least a way to reply.
I think there is a misunderstanding here. There is ZERO intention for us to have a public, anonymous reference function. When we say anonymous aspect to references, it is solely on letting users anonymously score the SAFETY of each interaction. Each interaction will still have a public reference.
So once again, I have no idea where this thread is going because I don’t think allowing victims a safe way to raise the flag on predators is “cowardly” and “backstabbing”.
Thanks for sharing the article, @kellyt!
Indeed, an anonymous option for feedback doesn’t give power to cowards. It gives power to voices that don’t want to end up in a shouting match whenever they rise an issue. A lot of people are just too accustomed to being able to shout down accusations without constraints.
Also can’t see the danger here being anonymous vendettas. We’re talking about a medium value from unrelated encounters. No need to mull over the effects of one wrong feedback. And the people that will be worried about the overall picture are exactly the people that should be worried.