Re: New Revisions, New Home

From: Mitch Howe (
Date: Wed Feb 13 2002 - 17:52:46 MST

Eliezer Yudkowsky wrote:
> I do not assert this. I note that there is a still-open question as to
> whether a superintelligence can be hacked into by a better
> or whether hackable flaws preventable by a bounded but transhuman amount
> intelligence.

My apologies. I reread your original statement...

> Any hack requires a vulnerability. Human code often contains
> vulnerabilities because humans have relatively short attention spans and
> can't linger for hours over each line of code. Also, humans often write
> lousy code because we are basically hunter-gatherers and programming is
> not one of the tasks we are evolved to accomplish. It's quite possible
> that code written by a sufficiently smart AI simply would not contain
> code-level flaws, such as buffer overruns and the other flaws that modern
> hackers exploit. To put it another way, the "game" of hacking doesn't
> necessarily continue between superintelligent programmers and
> superintelligent attackers. It could, but it's also possible that given a
> superintelligent programmer there are just no mistakes left to exploit,
> even if the attacker is a hundred times more superintelligent.

.. and I noticed that your wording did not suggest the level of confidence I
mistakenly thought you had in the possibility of 100% secure code. I
actually considered just quoting your entire response, but thought the FAQ
would be better off with a paraphrased version more consistent with the rest
of the document's style. Here is how I have rephrased the incorrect

"Mr. Yudkowsky asserts that it may be possible to create software without
any of the kinds of flaws traditionally exploited by hackers. The truth
today is that imperfect humans find this level of security difficult to
produce, and even more difficult to verify when the only hackers are more or
less evenly matched. But if there is a point at which software can be said
to be 100% secure, then a superintelligence might be able produce a cleanly
secure system that could not be perverted or hacked into by any mind, even
if the attacker were vastly more intelligent than this system."

--Mitch Howe

Do You Yahoo!?
Get your free address at

This archive was generated by hypermail 2.1.5 : Wed Jul 17 2013 - 04:00:37 MDT