Wednesday, 22 February 2012

The 'Snakeskin problem' and its implications for ridesharing services

The risks of CouchSurfing
I spent a lot of time today browsing through some murky threads on CouchSurfing is a social network where people open up their homes for free for travellers who need a place to stay. There was an unpleasant incident in Japan last month, where Cristiano-de-Angelis, an alleged sexual predator apparently raped a young Japanese woman after having contacted hundreds of other single women on CouchSurfing. He had deleted his profile several times after receiving negative feedback from hosts and was essentially untrackable within the system. A tragic abuse of trust but also evidence that feedback alone is not sufficient to guarantee safety within CouchSurfing.

The snakeskin problem
I'm calling this issue the 'snakeskin problem' - without a formal identity check, users can slough off profiles with negative feedback and start new ones that are not traceable back to the old one.  

Public feedback limits honesty
This isn't the first negative incident on CouchSurfing. Although the website claims that 99.6% of stays are positive (and I've only ever had good experiences myself), this is likely biased by the way that the website collects feedback. Forcing people to write feedback public pretty much guarantees that they're not going to say anything negative for fear of retribution (Teng et al).

Arguably, this public feedback mechanism is the cause of another high profile CouchSurfing rape case, where the culprit behaved inappropriately towards several previous couch surfers, who felt too uncomfortable to leave honest feedback.

Implications for ridesharing
I find these stories frightening and sobering. My ridesharing project, CoCoRide, would open up users to the same potential risks. When passengers get into the car with a stranger, they are opening themselves to all kinds of risks. They surrender almost all control over the situation. In this sense, ridesharing demands at least as much trust, if not more, than CouchSurfing.

The question I've been pondering is what can we do to ensure user's safety beyond just asking them to trust each other. I've come up with three strategies and would appreciate feedback on them.

Strategy one: Identity Verification
The reason why Cristiano-de-Angelis was able to wangle his way into the woman's house in Japan despite having received negative feedback from other couchsurfing hosts was because CouchSurfing has no serious identity verification. They send a postcard with a verification code to your home address, but that's it. No police check, no driver's licence check, no passport photos, nothing. It's weird when you think about it. In Europe, you can't check in to a hotel without handing your passport over. Even in Australia and the US, you have to give your credit card number as a bond, so that if you decide to steal the towels, they can charge you for it. But with CouchSurfing (and with AirBnB for that matter), you don't have to provide any ID. If you choose, you can take on an assumed name and wreak havoc, cutting a swathe through the warm and fuzzy network built on trust.

Even though I'd like to think that most people can be trusted, the evidence from CouchSurfing shows that trust alone is not enough. There needs to be recourse. I believe the act of supplying one's identity will impress upon people the importance of behaving with social decorum. Their information does not need to be shown publicly, but it needs to be on file. Identity verification will prevent people like Mr De Angelis from creating multiple accounts to circumvent negative feedback.

Barriers to identity verification
I reckon the chief reason why CouchSurfing and AirBnB haven't instituted proper identity verification yet is because it's not easy to do. Until recently, there haven't been any good or cheap ways to verify identity online. I'm not absolutely certain, but I think Deloitte's new product, GreenID, could change that. The idea is that people can verify their ID online using their name and address (electoral roll), their phone number (WhitePages), their credit card, their medicare card or their passport (plus a few other options). It all takes less than three minutes. It sounds amazing. If GreenID is available at a decent price, I think this could be the game changer in identity verification. (Disclaimer: I've sent Deloitte an email asking for a free subscription:P)

Strategy two: anonymous packaged feedback
There's been a fair bit of research into the best way to collect honest feedback about other people. Teng et al. demonstrate that CouchSurfing's model of public feedback is almost certainly not the best way to do it. When people know that the other person will see their feedback, they hold back from giving the unvarnished truth. It's the whole concept behind a 360 degree review: workers will only give useful feedback to managers when they have no fear of retribution.

Packaging feedback
So the feedback has to be anonymous. How do you do that though when people are typically going to give feedback right after they share a ride with someone? Can't the other person work out who it was based on the time stamp?

That's why the feedback needs to be packaged. What I mean by that is the feedback isn't released immediately. The feedback goes into a queue until there are two other reviews for the driver/passenger, and then it gets aggregated so that the other person has no way to work out who gave what score.

Strategy three: a proactive complaints team
The reason why AirBnB copped such a beating in the press is because they handled the home wrecking incidents very poorly. In both cases, the company didn't react until the victim had blogged about it and even then the complaints team apparently asked the victim to keep quiet.

CouchSurfing has a bad rap as well for being a bit too neutral after serious complaints. Their official policy is that they will stay out of it unless the police are called. This creates a problem when it might take a few days for the victim to be emotionally stable enough to report the incident to the police, and meanwhile, the perpetrator is potentially attacking other people as happened in the first CouchSurfing rape case.

In light of that, I believe that ridesharing organisations need to have a toll free number that users can ring if anything goes wrong. The complaints team need to be able to handle any emergencies that arise. They need to be able to ring a taxi, ring the police and do absolutely anything else necessary to make the person feel safe.

What are your thoughts?
Will these measures be enough to make people feel safe while ridesharing?

No comments: