SAFE VERSUS “FRIENDLY” USER INTERFACES
I don't understand why these two designs are mutually exclusive. Oftentimes user friendly interfaces limit user abilities to do anything outside the intended purpose. Why would making a system more complicated and unfriendly make it safer?
For some reason, people seem to believe in a calculated number more than actual experience.
Why are the calculated numbers wrong so often? Shouldn't we just be arguing for more accurate testing?
The subtler one is that an oracle’s ability to resolve ambiguity is only as good as its consensus mechanism.
Isn't it just as important to give accurate specifications as it is to use a good consensus mechanism? If both contract writers agree on the intent to purchase candy, but the oracle disagrees with the qualification of candy the outcome would be different than if a different oracle agreed with qualification of candy.
These choices collectively establish the meanings of Python programs — and change those meanings over time. Technical facts depend on socially determined ones.
However if there is the same situation in program with one central authority, such as Microsoft with Windows 10, can they force technical facts through the use of automatic updates and such?