Politeness, trust and ice cream.

[s1113645@tesla.cc.uottawa.ca]

On Sun, 9 Jun 1996 ichudov@algebra.com wrote:

Fascinating stuff about polite gun-owners deleted.

> If we think about anonymous computer contractors and anonymous
> employers, the interesting question is how to maintain reputations and
> how to check references.

I think it comes down to "who watches the watchers". Do you trust 
Business Week's ratings of so and so management consultant? Do you trust 
The Cypherpunk Guide To Business Magazines's ratings of Business Week's 
ratings of consultants? How about Joe Usenet's assessment of the above on
misc.stocks.slander? This begins to resemble the problem of the pgp Web 
of Trust. This problem already exists in the non-anonymous 
flesh-and-paper world. Who do you trust to tell you who to trust (and so on)?
Do you trust journalists who take ad money?

Presumably, once a decent profit model evolves for net publishing, there 
will be some incentive for customers to give you their opinions, and for 
others to gather them. Do you have a clear path of trust (or faith 
or some other quality) proceeding either to them or to their stated 
customers? 

I forsee many variations of trust webs to determine the quality of
ratings. Eg. I am 50% in agreement with Hal's taste in ice cream, 10% in
agreement with Declan's and 75% with the Economist's. I have signed this
with my key. Do the math to see how much you trust my assessment of Tim
Horton's chocolate pecan fudge. You decide how to do the math. Tim and Hal
had some really nice articles on this last month 

---Begin PGC Comment---
KeyID 0xF00

C1: Payment-Statement: I have not been payed to make the above 
endorsement.

C2: Coercion-Statement: I have not been coerced into making the above 
statement  

---Begin PGC Signature---
13235097u13251-9233u5v123rsdf;lkhjs -882351932u4v
---End   PGc Signature---
Pretty Good Commentary is a copyrighted trademark of KeyID 0xF00
---End   PGC Comment---

Reminds me that I should grab an AI book real soon now. 

Another thing one could use is a pseudonymous open-booking protocol (I 
didn't read Eric's post, so I don't know if it's any good) to determine 
if alleged customers are the real article. Offhand, I'd venture a guess 
that we'd see the above problem again, which hints at the importance of a 
good generalized trust or agreement calculus (and calculator) for formalized 
comments.

I don't want to imagine how bad the traffic will get on IETF mailing 
lists to standardize trust comments.