XPost: alt.folklore.computers
From: invalid@invalid.invalid
Bobbie Sellers writes:
> On 12/28/25 22:40, rbowman wrote:
>> On Mon, 29 Dec 2025 01:17:50 -0500, c186282 wrote:
>>> For most 'office' uses you do NOT need AES-256 encryption for yer
>>> damned payroll or budget files. Nobody, not even Vlad or Xi, CARES.
>> We were dealing with NPS sites like Yellowstone and Rocky
>> Mountain. The Dept. of Interior certainly cares.
>
> Any employer who has a payroll has the Social Security numbers
> of the employees. This certainly deserves the best encryption that
> can be set up.
>
> The encryption schemes are all breakable with enough power
> brought to bear.
At this point you should take a moment to work out how much ‘enough’ is.
Let’s assume that:
- you have a circuit design that can do a single AES-128 key
schedule and decrypt operation in one cycle
- you can fit a million copies of this design onto one chip
- you can run the chip at 10GHz
- you can manufacture a trillion instances of the chip
(and put enough of a computer around them to do something useful)
- you can somehow power and cool this unrealistically large
supercomputer.
This lets you test 10^6 * 10^10 * 10^12 = 10^28 keys per second. There
are 2^128 possible AES-128 keys so it will take you a little over one
thousand years to break AES-128. Your SSN will no longer be relevant by
this point.
All the performance numbers are deliberately overestimated, so the real
timespan is probably much larger.
For AES-256 the result is 10^41 years. The universe will end first.
> Another reason to worry about AI and vast amounts of computing
> power brought to bear on what exactly.
‘AI’ does not make computers any more powerful. For anything numerate it
has a good chance of confidently telling you the wrong answer.
--
https://www.greenend.org.uk/rjk/
--- SoupGate-Win32 v1.05
* Origin: you cannot sedate... all the things you hate (1:229/2)
|
