Institute For The Future

Reputation systems in the world

A short forever ago (May 2004), I wrote a piece for my Red Herring blog about carpooling, digital identities, and the possibility of turning underused private resources (at least temporarily) into sharable, public ones. (I've put a copy in the extended post.) Recently on Private Revolution, the blog of the founders of the very cool Xtracycle, I came across a post on a similar concept, hitchiker vetting:

 

I’ve been thinking about how cool it would be if there was someway to know that the hitchhiker was cool, and that the driver picking you up wasn’t a serial abductor. What if there was an ebay-like system where you get rated by the people who give you rides or by the riders you pick up? Participants could have an ID card linked to their profile. It could all be checked from your cell phone.

What’s more, the cell phone itself could be used with some of the new “who’s in my neighborhood” tech to find people going your way. This capability seems to be right around the corner. More likely, it’s already here, but no one cares enough about hitchhiking quite yet to make it work. I predict its imminent arrival.

Over at WorldChanging, there’s a piece about a universal reputation rating system. Something like this could incorporate the hitchhiking vetting process. It does seem a bit creepy, having an online reputation for everything you do, like maybe a little too transparent. But there are certainly environmental benefits to the sharing scheme that it supposes: if you can trust more people to borrow your stuff, and vice versa, a whole lot more sharing can potentially enable a lot less “I need my own everything” mentality.

 

How does this connect of the end of cyberspace? It strikes me that if the Internet five or ten years ago seemed attractive as a playground to experiment with identity, or to watch the emergence of new forms of social interaction based on (or enabled and protected by) anonymity, one of the big stories of the last couple years has been the realization that you can do some pretty cool things in the digital world if you reveal the connections between online identities or activities, and real people.

Here's the Red Herring piece I wrote a couple years ago.

Cars as Public Transit

Red Herring (May 2004)

Traffic in the Bay Area is terrible. Not surprising: according to a recent San Jose Mercury News column, there are now 4.5 million cars in the Bay Area, logging 167 million miles every day—over five billion miles a year. Cars pollute the environment, adversely affect our quality of life, and make the Bay Area less attractive and productive. So why don't we have more public transportation? The existing light rail system isn't very heavily-used, and recent projections suggest that ridership on BART will go down in coming decades. The Bay Area is a patchwork of small governments, which adds complexity to any big infrastructure project. A bottom-up approach doesn't work well for mass transit systems, which have to be carefully knit together to attract riders. And for better or worse, today's voters aren't very enthusiastic about large public spending projects, and the state isn't exactly flush with cash.

Further, the changes required to turn the Bay Area into a place where large-scale public transportation would really work shouldn't be underestimated. Proponents of public transportation argue that the region's high cost of living would be lowered if we could build high-density housing served by light-rail trains that would make cars unnecessary. Essentially, we'd have to transform the Valley from a sprawl of suburbs and towns into something more like Manhattan or Tokyo—a sprawl of cities.

And, frankly, most people like cars better than public transportation. They like the convenience of being able to go exactly from Point A to Point B, exactly when they want. There are some other experiments in reducing traffic and the need for car ownership. There are vanpools, shuttles that run between train stations and corporate parks, and car-sharing services. These have loyal users—I rode in a vanpool between Berkeley and Davis for several years—but they're not widely popular. Many of the freeways have carpool lanes, but as anyone who drives during rush hour knows, the existence of these lanes hasn't generated much of a carpooling movement. (When you realize that there are 4.5 million cars and 6.5 million people in the Bay Area, the lack of traffic in the carpool lanes makes more sense.)

It seems an intractable problem. In a region that is not as densely built up as a real city, in which it's hard to live without a car, public transportation is always going to be less attractive than privately-owned transportation. But there's a possible solution, and it comes from an unlikely place: the open source movement. Maybe the answer is to tweak the rules of ownership, to create a system that hooks up drivers and passengers in real time, and creates incentives to share the road. Maybe the answer is to create a public transportation system out of automobiles.

How could cars become public transit? How could you create an ad-hoc public transportation functionality on top of privately-owned cars? Here's a hypothetical model, which is incomplete and doubtless could stand many improvements. (If you see any, feel free to comment.)

The one major piece that would have to be created is an online marketplace that would connect drivers and passengers. It would be part eBay, part old-fashioned ridesharing bulletin board—in fact, let's call it cBay. (Legal correspondence may be directed to "The Future," Red Herring, the South Pole.) But the only other things you'd really need to make it work are cell phones, which lots of people have.

Say I'm a driver. I tell cBay a little about my car—what kind of car I drive, whether it has room for child seats or luggage, that sort of thing. I tell it that I have a regular commute on weekday mornings and afternoons; maybe I also have some other destinations that I go to regularly, and want to have cBay remember them. I can also set certain preferences: I want smoking or nonsmoking passengers; I'm more comfortable with passengers who are the same gender as I am (this isn't a dating service, after all); I speak English and passable Chinese; and I'm willing to make intermediate stops between my home and work, so long as they don't take me more than two blocks away from my normal route.

Passengers would have a similar preference profile: certain regular destinations that cBay can remember, settings regarding linguistic skills, how far away from my destinations I can be dropped off, etc. All this information—as well as information that potential passengers put into the system—can be used to match drivers and passengers, but can be invisible to other participants, for privacy and security.

So let's start it up on a Monday morning. Drivers who are going to work and are willing to take passengers message cBay and tell it where they're going. Riders tell cBay where you are and where you want to go. cBay then generates a list of matches, and messages each party. (If someone doesn't get any matches, they're asked if they want to compromise on some criterion—accept a rider who's a little further out of the way, or a driver who's headed to a place near your destination—and cBay tries another round of matching.)

If both parties accept, the passenger is sent an ID number for that ride. (It seems fairer to make the passenger do a little more work; besides, it's safer for them to be navigating a phone tree than the driver.) They then call the cBay 800 number, enter that ID number, and are connected to the driver. The two work out where to meet and when, and how to identify each other.

When passengers are picked up, they call the 800 number to verify that they're in the car. At the end of the ride, each part calls to say that the ride is over, and are given the chance to rate the other person on punctuality and courtesy; passengers rate how safe they felt with the driver, and how safely they drove; drivers can rate how well the passenger followed instructions, and how safe they felt with the passenger. In this way, drivers and passengers develop reputations that other cBay users can see, and can use in deciding whether to accept rides.

Of course, security will be a concern. So members might have the option of getting regular check-up calls from cBay during a ride. They have to enter a password or ID number, otherwise the police will be notified. Users whose cell phones have GPS, or whose service providers can triangulate their location via signal strength, will have even more security. Finally, passengers with camera cell phones can always take pictures of the car and driver.

So the technology bar isn't very high. Aside from the cBay network, the major thing you'd need is widespread access to cell phones, preferably with messaging capability and Web access; camera phones would provide some extra security. None of these is Buck Rogers technology, though some of it is still a little pricey. But what about incentives?

The obvious immediate incentive would be access to carpool lanes, which are all over the Bay Area, and which can make a drive go quite a bit faster. Passengers might also pay membership or per-use fees, the proceeds of which could be distributed to drivers to cover the costs of gas and labor; the system might also allow drivers and passengers to work out fares.

However, it may be that many drivers who joined up would do so for the indirect benefits—the chance to help the environment, to decrease our fossil fuel consumption, or just to be well thought-of. The open source movement has demonstrated that such intangibles, when made available to a large number of people, can produce surprising results.

Our hypothetical cBay system would provide a way of matching drivers and passengers who share common destinations and similar preferences. You might think of it as a way to increase the load factor on the highways, by increasing the number of passengers. There are a couple other important trends it draws upon.

For one thing, it exemplifies Reed's Law of group-forming networks. Reed's Law holds that the value of such networks increase not as some multiple of the number of members, but as 2^N of members: in other words, it increases exponentially. It should be self-evident that if cBay has a small number of drivers and passengers, the likely number of matches will be very low; with hundreds of thousands of members, that number will rise rapidly. Eric Raymond famously declared, "With enough eyeballs, all bugs are shallow." With enough cars, all transits become shallow—or, to put it another way, there's always someone else going your way. The challenge has been finding them. Microsoft's Marc Smith says that the great thing about the Internet is that if you're one in a million, there are 768 of you online, and you can find each other. cBay would do that for the road.

It can also become a form of social software. Let's say you have several encounters with another cBay member, the two of you get along well, and you'd prefer to have them as a driver or passenger. You could invite them to join the equivalent of a buddy list, which would make them preferred partners on drives.

You can even imagine that a system with a lot of users could support even more fine-grained social links. Imagine being able to organize mobile salons: being able to fill a car with people are all reading the same book, or who want to talk about politics or sports. You might even make drives more productive, for example by pairing up people who'll offer rides in exchange for tutoring in a language.