For the large part, such a trust network already exists with Facebook. Yes, it's not perfect, and yes, it probably can be better, but when you want some form of trust it's usually very powerful to use Facebook Connect to allow a user to see another user's FB profile.

I've been recently involved in creating a site for sports meetups (like, say, for a casual game of tennis), and FB Connect came up pretty early on in our discussions as a way of getting people to trust each other.

Trust networks are too important to leave in the care of one company.

But because of the network effect, they'll always be in the care of one something.

It's an interesting idea and I was with him until he said the government should be involved to enforce this. The government doesn't regulate any meaningful relationships in my life so why should it regulate who I decide to trust? Seems like a fantasy to even suggest that it could do so.

Really? so you don't have any "meaningful relationships" that involve contracts, laws, things like that? You don't drive a car? Use a phone? Benefit from Police and Fire protection? Those things are all government regulated.

Really when you think about it, government has a lot to do right now with regulating trust. Think of how much more difficult trust between businesses would be if contracts didn't have the force of law? If disputes couldn't be settled in the courts?

Meaningful might be something else to you than it does to me. The meaningful relationships in my life include my wife, my child, my family, my friends, and of course spiritual relationships. The government does nothing in these.

I don't want a relationship with the police department, the fire department and I don't see a purpose for the government to regulate phones.

Then there is also the issue that contracts can be voided if the government decides to.

The governmet issues ID cards, which are used in diverse trust situations.

I agree that government shouldn't take an active role in running the system, but protecting the system by preventing abuse seems like a good fit. Maybe something very similar to the way government enforces contracts.

I see two other roles the government can play. The first is funding research into the best way to build a distributed trust network.

The other, perhaps more important, role is in creating and enforcing standards. (This role might emerge naturally from the results of research.) There must be a defined set of features that anyone providing a "trust network client" has to implement. Preferably such standards would include a provision that requires providers to let users package up their "trust data" and take it to another provider, with a guarantee that the original provider would not maintain it. (After all, you might decide that you don't trust your provider!) Otherwise, it will be the same old story: every private entity in the game will want to control their users' data, and prevent them from going elsewhere. Any moderately successful trust network will be incompatible with the others, and even those that become popular will live and die with the entities behind them.

Imagine if the Internet had been built like that.

There might be something to it. I doubt he's thinking of a big brother scenario; he was referring to checks and balances. The government has a lot of information that would be useful about whom to trust and perhaps Craig was mostly thinking about government in a collaborative, information-sharing role.

But at first glance the thought of the government telling me whom to trust gives me the creeps, too.

At http://ycombinator.com/ideas.html, pg speculates that ideas 4 (outsourced IT), 18 (the WebOS), and 19 (application and/or data hosting) can be bundled.

I think Newmark's vision may belong in this bundle as well. It somehow feels like part of the same domain. Though as pg says, "the way to find such a grand, overarching solution is probably not to approach it directly, but to start by solving smaller, specific problems, then gradually expand your scope."

So the real problem is how to crack it open? What are the first steps that lead fruitfully into the bigger problem domain?

It's a great idea ... I bet it ends up just like openid though .... every big company competing to be the provider and hesitant to accept anyone else on their own platforms.

Check out CouchSurfing for an example of how to solve this problem. Craig is just talking about how to make it universal rather than site-specific.

CouchSurfing provides enough trust data that you'll have someone into your home based on a web profile. Some of the mechanisms include personal references, friend lists, community awards (that are meaningful and hard to get), and unrevokable vouches.

Cafe Reverie FTW.

They have pretty good lunches - he's in the patio in back.