Scientifically rigorous computation: was "You don't need delegation, but it's very useful"
Carl Hewitt
Thanks Christine!
We need a rigorous scientific understanding of computation that is free of dogma.
As in other branches of science, that means mathematical foundations.
Such foundations can help address critical practical challenges such as the following:
Currently, a Great Power cyberattack campaign can take down a country’s economy and
keep it down for a long time [“Sandworm” Greenberg 2019] . Unless there is drastic change, it is expected that
there is a particular country that will be so attacked within the next 7 years in the context of a larger crisis.
To prevent successful cyberattacks, an Apollo-scale project is required to develop and deploy
current scientific understanding.
See the following video for more information:
“Preventing Successful Cyberattacks Using Strongly-typed Actors”
https://www.youtube.com/watch?v=AJP1VL7shiI
Regards,
Carl
https://professorhewitt.blogspot.com/
-----Original Message-----
From: fr...@googlegroups.com <fr...@googlegroups.com> On Behalf Of Christine Lemmer-Webber
Sent: Wednesday, July 28, 2021 19:22
To: fr...@googlegroups.com
Cc: Alan Karp <alan...@gmail.com>
Subject: Re: [friam] You don't need delegation, but it's very useful
Alan Karp writes:
> (Previously sent to cap-talk, but I haven't gotten any responses.)
>
> from a comment made on a standards email list.
>
>
> That's not what I think. I suppose you could avoid needing
> attenuation by having a separate capability for each possible
> permission (or worse, combination of permissions), but I don't know
> how to do responsibility tracking without it.
In a sense, since you cannot prevent delegation, maybe you don't need it, in that you'll always have it. ;)
One doesn't need memory reclamation either, but it is very useful :)
- Christine
PS: I think the earliest implementations of lisp machines, maybe pre-CADR even, actually didn't have first-class memory reclamation support either (hardware based GC came later), but also had it anyway... what people would do is write out the entire running "world" to disk and come back later and boot it again, as a kind of hardcore stop-and-copy. I think I read this in the lisp history writeup or something... I like the story, so I hope I didn't get it wrong.