Consider Alice and Bob, who are dating each other
Bob wants to know Alice's password, but Alice doesn't want to give it to Bob because maybe Bob is the NSA
Bob accuses Alice of not trusting him
Here's the thing: Even if Alice trusts Bob, Alice can have a good reason to think Bob may be the NSA
Consider the statement "Alice trusts Bob"
This statement is often used to mean something else
Let us abstract Alice and Bob -- they are each computers with some inputs and outputs
They are also both connected, and connected to each other, to and through the outside environment
This is not any sort of solipsistic argument
It is simply the truth that Alice cannot directly know about Bob
Information about Bob is transferred to Alice through the environment in the form of light, sound, etc
Therefore Alice develops this representation of Bob in her computational substrate
When Alice says she trusts Bob, she is saying she trusts her representation of Bob: Bob'
This is also not some semantic argument on the meaning of trust
To illustrate this, think about someone you know
Clearly, you do not know the person themself, but rather your representation of them
Now the reason Alice does not give her password to Bob is simple
Alice trusts Bob', but she does not trust that Bob is like Bob'
This argument is very appealing because it does not blame anyone
If Alice does not trust Bob, that may be a fault of character on Alice's part
But if Alice trusts Bob', but does not trust that Bob is like Bob', there are two possibilities
Either Alice has a poor ability to process her inputs effectively
In this case, it is a fault not of Alice's character, but computing abilities
Otherwise it is the fault of the environment for not transmitting limited information
Therefore, this view makes trust and caution compatible
Enter your comments and criticism below my dear forumers