2 ms·

This is yet another one of those rambling, wide-dispersal attacks on SV. So here's the thing: we have solved many of these problems. Many of these problems boi

This is yet another one of those rambling, wide-dispersal attacks on SV. So here's the thing: we have solved many of these problems. Many of these problems boil down to trust relationships and increased information in the marketplace. The internet is pretty good at this. Ever order something from a person on E-bay? Why? Heck, with big-ticket items on there, you could easily get ripped off. And back in the early days it happened a lot more. But E-bay implemented a peer-to-peer rating system. Now when you buy anything, you look at the seller's rating. They look at your rating. People with long histories of happy transactions are highly unlikely to ruin it all for just one. It happens, sure, but it's highly unlikely. Nobody is arguing to throw grandmother out into the snow. I think anybody with a brain realizes what various laws were put into place to do. And anybody with a heart would agree with their goals. But it's a fair question to ask if the solutions from 1950 still make sense in the internet age. Many times -- not all, but many times -- they do not. We've got a better way worked out. We figured it out through trial and error and rapid iteration. EDIT: And for those of you who missed the kicker, perhaps I didn't make this clear enough. We've found that peer-to-peer ratings/trust systems statistically outperform enforcement activities, because you are enlisting everybody in the community as an enforcer So you get your great social goals _and_ better adherence. It's a thing of beauty.

rgarcia 14y ago
I agree, but it begs the question: is there a point at which the damage you can inflict with a single bad transaction makes some central authority necessary? e.g. if every 20th car I sell on eBay is a lemon, I still have a 95% rating. If I'm a driver on Uber/Lyft/Sidecar and every 20th person I leave in a ditch I still have 4.75 stars. Obviously regulation isn't meant to stop outright crime like this, but I think it's meant to guarantee a minimum level of service so that the distribution of outcomes has a reasonable lower bound.
- DanielBMarkham 14y ago
  I think it can be. Obviously you wouldn't use a system like this to decide who gets to own nuclear weapons. There are a couple of factors: cost per "incident" and volume of transactions. P2P trust systems work where there is a large volume, the cost-per-incident is relatively low, and, most importantly, the provider looks forward to continued transactions. To me cabs seem to easily fall under this rubric. You know, it's possible to get a medical license and be so inept as to kill people for decades until you get caught. It's possible to get a hoteliers license and spy on your customers and cheat them out of services -- happens all the time. So we have to be clear that we are not comparing absolutes. There will always be problems with either solution. The question, to me, is "which system is more adaptive?" Some systems adapt quickly as conditions change. Most legal systems do not. I'd still want legal recourse for bad actors in many situations. But things like taxicab drivers robbing you? We already have legal recourse, yes? So in reality with the taxi situation we're only talking about the cab fare, not a FUD list of possible really bad things (tm) that could happen in a taxi. Geesh. That'd be a huge list.

rgarcia 14y ago

I agree, but it begs the question: is there a point at which the damage you can inflict with a single bad transaction makes some central authority necessary? e.g. if every 20th car I sell on eBay is a lemon, I still have a 95% rating. If I'm a driver on Uber/Lyft/Sidecar and every 20th person I leave in a ditch I still have 4.75 stars. Obviously regulation isn't meant to stop outright crime like this, but I think it's meant to guarantee a minimum level of service so that the distribution of outcomes has a reasonable lower bound.

DanielBMarkham 14y ago
I think it can be. Obviously you wouldn't use a system like this to decide who gets to own nuclear weapons. There are a couple of factors: cost per "incident" and volume of transactions. P2P trust systems work where there is a large volume, the cost-per-incident is relatively low, and, most importantly, the provider looks forward to continued transactions. To me cabs seem to easily fall under this rubric. You know, it's possible to get a medical license and be so inept as to kill people for decades until you get caught. It's possible to get a hoteliers license and spy on your customers and cheat them out of services -- happens all the time. So we have to be clear that we are not comparing absolutes. There will always be problems with either solution. The question, to me, is "which system is more adaptive?" Some systems adapt quickly as conditions change. Most legal systems do not. I'd still want legal recourse for bad actors in many situations. But things like taxicab drivers robbing you? We already have legal recourse, yes? So in reality with the taxi situation we're only talking about the cab fare, not a FUD list of possible really bad things (tm) that could happen in a taxi. Geesh. That'd be a huge list.