[cap-talk] Montesquieu, the *-property and implicit grantability
jed at nersc.gov
Wed Jan 4 20:07:09 EST 2006
(sorry I didn't respond to this earlier Rob, I missed it going by)
At 11:32 PM 12/2/2005, Rob J Meijer wrote:
>Jed has been making the argument that non-grantable capabilities and
>any other form of blocking capability transfer is bad.
>I think I've got a chalanging scenario for Jed to proof that this
>argument is valid and alternatives are available.
>Montesquieu devides the political authority into the legislative,
>executive and judicial powers. Lets say that these 3 are to be described
>as seperate sets of capabilities, lets also say that all 3 are than
>to create 3 seperate POLA based subsystems. Lets also say that all sub
>authorities are held by a large amounth of low level officials.
>Now lets say that a subset of the low-level officials are fooled by a
>potential despote to give him their authority.
>With or without non-grantable capabilities, but with available
>side-chanels, could one prevent the the 3 seperate political powers to
>acumulate with such a potential despote?
I argue no. The use of the term "side channels" in the above seems rather
odd to me. Still, if such a despot arises and can convince the low level
authorities to act on his behalf (whether directly or by granting him
their capabilities - that aspect seems an implementation detail to me),
then he can assume their authority.
>I used to think that seperated 'do' and 'grant' model surficed, but
>after Jeds argumentation I am no longer sure that modern civilications
>that implement Montesquieus principles like that are not vulnerable
>to such vulnerabilities as described by Jed. If it does not, and
>seperation is futile, than what organisational measures would be needed
>to circumvent this vulnerability?
Dare I say it? POLA. To use your metaphor, if a despot arises, one
hopes she will only be able to convince a small number of low level
officials to give her their authorities. One hopes that such delegations
become visible when used (e.g. logging use of the registered low level
capabilities), the low level capabilities will be revoked, the low level
officials fired, and the system will right itself.
>It apears to me that history seems to
>show that political systems somehow have this problem tackled, but
>if not by virtue of seperation of grant and do capabilities, it is
>unclear to me what other saveguards are in place.
I believe it amounts to the above.
More information about the cap-talk